Research Solutions Inc Aktie

Good afternoon, everyone, and thank you for participating in today's conference call to discuss Research Solutions' Financial and Operating Results for its Fiscal 2026 Third Quarter Ended March 31, 2026. As a reminder, this conference is being recorded.

I'd like to now turn the conference over to your host, John Beisler, Investor Relations.

Thank you, operator. Good afternoon, everyone. Thank you for joining us today for Research Solutions' third quarter fiscal year 2026 earnings call. On the call today are Roy W. Olivier, President, Chairman, and Chief Executive Officer; and Dave Kutil, Chief Financial Officer. After the market closed this afternoon, the company issued a press release announcing its results for the third quarter of fiscal 2026. The release is available on the company's website, researchsolutions.com.

Before Roy and Dave begin their prepared remarks, I would like to remind you that some of the statements made today will be forward-looking and made under the Private Securities Litigation Reform Act of 1995. Actual results may differ materially from those expressed or implied due to a variety of factors. We refer you to Research Solutions' recent filings with the SEC for a more detailed discussion of the risks and uncertainties that could impact the company's future operating results and financial conditions.

Also on today's call, management will reference certain non-GAAP financial measures, which we believe provide useful information for investors. A reconciliation of those measures to GAAP measures is included in today's earnings press release as well. Finally, I would like to again remind everyone this call is being recorded and made available for replay via a link on the company's website.

I would now like to turn the call over to Roy W. Olivier. Roy?

Thanks, John. The third quarter represented improving EBITDA and net income but was certainly lower than our expectations in terms of top line growth. Churn continues to be an area that needs action, which I will talk about in more detail later in the call. New bookings were good at 61 new or upsell logos, representing $961,000 in ARR. Unfortunately, churn was 46 logos, representing $398,000 in ARR. Most of those were smaller accounts. However, there was 1 large account in that total that represented about $130,000.

As a reminder, about 1/3 of churn is uncontrollable. That category is primarily driven by one of our customers being acquired, going out of business, or experiencing a reorg that includes eliminating the research function. 2/3 are controllable, and that's where our focus is, and none of it is related to "AI usage". There was a lot of good news in the quarter. Academic and corporate sales were both strong. We signed sizable academic deals in Johannesburg, Singapore, and several with top U.S. universities, including my alma mater, Texas A&M. These deals are primarily Scite deals. We also signed several large corporate deals at levels well above or at historic average sales price or ASP. This is a good mix of both Scite and Article Galaxy deals. About 70% of those deals are AG deals. The point is that new bookings are executing well, and I feel good about our new bookings growth.

We have been using AI internally in virtually every area of the business. The product teams have seen a nice uptick in productivity as a result. Due to this, we can now assign software development issues to AI and then have a developer check that work. This has accelerated the number of new items in each release. In addition, during the quarter, we released 2 new AI-based products. These are called MCPs, which stands for Model Context Protocol. You can think of an MCP as similar to a software API, but it is a connector to integrate an AI-based LLM like ChatGPT or Claude or an internally developed LLM into Article Galaxy or Scite. This is part of our headless strategy, which means we want to be where our customers are working. I'll talk about that also more later in the call.

I'll discuss all this with you in a little bit more detail, but for now, I'll have Dave walk you through the results in more detail. Dave?

Thank you, Roy, and good afternoon, everyone. Total revenue for the third quarter of fiscal 2026 was $12.1 million compared to $12.7 million in the third quarter of fiscal 2025. Growth in platform subscription revenue was more than offset by a decline in our lower-margin transactions business.

Our platform subscription revenue increased approximately 7% to $5.2 million. The growth was primarily driven by a net increase of 15 platform deployments over the prior year, as well as expansion within our existing customer base through upsells and cross-sells. Platform revenue accounted for about 43% of our total revenue for the quarter, compared to approximately 38% in the prior year quarter, as this mix continues to move towards the higher-margin platform business.

We ended the quarter with $22.1 million in annual recurring revenue, or ARR, up 8.5% year-over-year, consisting of approximately $15.7 million in B2B ARR and approximately $6.4 million in normalized ARR associated with Scite's B2C subscribers. Although it was down 7.5% year-over-year, B2C ARR has shown signs of improvement in recent months as the MCP launch that Roy mentioned earlier has increased both conversion and retention. Net incremental platform ARR for the quarter was approximately $317,000. Please see today's press release for how we define and use annual recurring revenue and other non-GAAP items.

Transaction revenue for the third quarter was $7.0 million compared to $7.8 million in the prior year quarter. The softness was driven by a previously discussed churned account and volume reductions from a small number of larger customers. It is notable that the monthly trend inside the quarter showed meaningful directional improvement, and while not a full recovery, it is an early sign of stabilization. Our total active transaction customer count for the quarter was 1,346 compared to 1,380 in the same period a year ago.

Gross profit for the third quarter was $6.3 million, essentially flat with the prior year quarter from a dollar basis standpoint on lower revenue. Gross margin was 51.7%, a 220 basis point improvement over the third quarter of fiscal 2025. The increase was driven primarily by the ongoing revenue mix shift towards our higher-margin platforms business.

On a trailing 12-month basis, the company's blended gross margin now stands at 51.4%. The platform business recorded a gross margin of 86.4% compared to 87.4% in the prior year quarter, reflecting modest hosting and infrastructure investments to support our AI and integration road map, and is still well within our high-to-mid 80% target range. Gross margin in our transaction business was 26%, essentially unchanged from the third quarter of fiscal 2025.

Total operating expenses in the quarter were $5.2 million compared to $5.7 million in the prior year quarter. The improvement was driven primarily by lower general and administrative expenses and lower stock-based compensation, partially offset by continued investment in sales, marketing, and product. This is further evidence that the company's profitability improvement is not just emanating from margin mix. It is also a function of our operating discipline. We are being deliberate about G&A spend, keeping it contained while we put resources to work against growth initiatives. We expect the discipline to translate into continued operating leverage.

Net income for the quarter was $860,000, or $0.03 per diluted share, compared to net income of $216,000 or $0.01 per diluted share in the prior year quarter, an increase of approximately 297%. Adjusted EBITDA for the quarter was $1.6 million compared to $1.4 million in the year ago quarter, a 14% increase. On a trailing 12-month basis, our adjusted EBITDA margin was 12.3%, a 220 basis point increase from the prior year quarter. And trailing 12-month adjusted EBITDA now stands at $6 million.

Turning to our balance sheet. Cash and cash equivalents as of quarter end were $12.1 million, essentially unchanged from our 2025 fiscal year-end, even after funding the Scite earnout payments made during the first 9 months of fiscal 2026. That consisted of approximately $3.7 million in cash and the issuance of approximately 739,000 shares of stock. We ended the quarter with no outstanding borrowings on our revolving line of credit, providing additional flexibility on the balance sheet.

Cash flow from operations for the quarter were $1.0 million compared to $2.9 million in the prior year quarter. The decline reflects the timing of customer billings and strategic prepays rather than a change in underlying earnings power or a change in the collectability of receivables. Trailing 12 months cash flow from operations was $5.7 million. We are entering the final quarter of fiscal 2026 with the aim of delivering adjusted EBITDA growth over the prior year, driven by continued platform subscription growth, improved retention, growing stabilization in transactions, and disciplined expense management.

We expect to exit fiscal 2026 with stronger earning power and cash generation, reinforcing the durability of our Software-as-a-Service platform model and our ability to invest behind the next phase of growth.

I'll now turn the call back to Roy. Roy?

Thanks, Dave. By the way, Josh, who normally attends these calls, is out today. He and his wife had a baby last week, and he's home taking care of the baby. So turning back to B2B bookings. The academic and corporate teams are doing well. As a reminder, we built out the academic focus team back in 2025 or fiscal year '25. The corporate team has been around for many, many years. The corporate team did well during the quarter, booking around $400,000. The academic team generated about $265,000, even though it's a seasonally slow time for them. The remainder or the balance of the new bookings was generated by the CSM upsell-renewal team.

While many reps on those teams are still new or in their first year, the performance on those teams is on track and improving. As I mentioned earlier, B2B churn is the primary drag on top line growth. We are doing several things to improve churn, including reorganizing the team, adding additional resources to that team, installing new technology to move us from reactive to proactive, as well as identifying low-usage users and kicking off workflows to reengage with those users. Finally, we are improving our onboarding and training to ensure that we improve customer adoption. While this is not something that gets fixed in a quarter, I feel very good about our plan to get this back to historic levels. We'll report more on this during our next call.

Turning to B2C. We have started to see some improvement from previous quarters. Product enhancements and the new AI-based solutions I previously mentioned helped reverse what was a steady decline in that business. We saw less new trials on far less digital spend, but have kept MRR about flat quarter-over-quarter. We've reduced our CAC by about 24%, and we've increased our lifetime value, or LTV. We have had a good start to Q4, but keep in mind that we'll be entering the low season as universities let out for summer. That said, we have some exciting new versions of the B2C product and those are growing nicely.

Product strategy and innovation are the drivers of all things for us. We've continued to improve the Scite and AG products. As a reminder, we deliver Scite and AG's value 3 ways. First is to customers that literally license the software and use it daily. Second, we provide that same functionality and value via software APIs for customers that have internally developed their own solutions but need our unique capability at points in that workflow. As mentioned previously, the third option is the new 2 AI-based products that connect AI LLMs to the product. These, again, are called MCPs, and they are 1-click connectors that allow the value of Scite and AG to work within an LLM. They are working now with ChatGPT and Claude.

We have this running for all B2C and some B2B users today. Back to our previous quarters where we talked about our headless strategy to be where the customer is, many of our users are starting to work in an LLM. They can ask a question of that LLM, and interact with Scite content in a copyright-compliant way. Once they get a fully cited answer, they can ask that LLM for the articles, and it will talk to Article Galaxy and go get that content for the user in a copyright-compliant way. This basically allows an LLM to work with Article Galaxy and/or Scite, again, in a copyright compliant way. This will be a big part of our future growth.

The Scite MCP was launched about 2 months ago, the AG MCP a month ago. We already have a sales pipeline of more than $1 million in opportunities for these new products. These MCPs will work with academic or corporate customers. In addition, we have integrated the Resolute databases into these MCPs.

As a reminder, we purchased Resolute about 2 years ago, and 1 of the assets in that acquisition was access to over a dozen curated databases relevant to research. We have integrated those into our MCP. So when a user asks a question in the AI tool, the tool's answer today will include patents, clinical trial, and major grant data. This means users can -- nope, sorry, wrong page-- can start and end their research journey in the AI LLM of their choice. There's a tremendous amount of excitement internally and across our customer base about these databases being integrated. We'll announce additional data sets as we add them.

Integrating the unique value we can deliver with where our customers are doing research will remain the cornerstone of our strategy. In short, AI will not eliminate research, but it will change how it is accessed.

Regarding document delivery or DocDel, we expect to see improvement in Q4 in terms of FY -- I'm sorry, in terms of year-over-year performance. As noted on earlier earnings calls, most of this decline was driven by a handful of customers. We have a large customer churn impacting the year-over-year decline, and a few other customers are doing less research than they did a year ago. Other than those customers, we are seeing increases in DocDel, particularly in OA or other free DocDel, but paid is slightly down.

Regarding M&A, we continue to look for interesting opportunities and are progressing with some of those opportunities. That said, we are currently trading at 2.7x to 3.4x based on TTM enterprise value to revenue, depending on how you value the DocDels. This makes it challenging to pay the multiple sellers expect, which are still based on pre-SaaS software values declining due to AI concerns.

To summarize, I'm happy with the cash flow, net income, EBITDA, sales execution, and product work we did in Q3. I don't think our numbers reflect the great work we are doing in those areas. I do believe that progress will show up in future quarters and in FY '27.

Now I'll pass it over to the operator for questions. Operator?

[Operator Instructions] And we'll take our first question from Jacob Stephan with Lake Street Capital Markets.

Roy, I guess, first, I just want to touch on the B2B churn a little bit. Obviously, it sounds like a great new logo quarter, hindered by some of that churn. But I guess what are the pain points of customers? What are you hearing from them as reasons for the churn?

Basically, it comes down to customers that have limited engagement. So in other words, we don't see a lot of usage of the platform or they don't purchase enough articles, in Article Galaxy's case, to make the ROI work for the platform. Some cases in this economy, customers are simply looking for ways to save money. So they may not be a big research organization, but they want to cut costs associated with doing that. But for us, the priorities moving forward are improving onboarding and training to ensure that at the end of a 90-day window, we have a majority of the researchers logged in, having used the product, and been trained on the product.

Number two, we want to monitor usage of the product so that when we see a cohort of customers that are not using the product, we can automatically kick off a 3-or 4-point communication to that customer via in-product messaging and email to reengage them and get them to use the product.

And number three, we know there are certain features in our product that are very high renewal features. In other words, people that use certain features in the product use it heavily and they very rarely churn out. So just like usage, we will monitor cohorts of customers that are not using those features and kick off workflows to help them know those features are there, help them know how to use those features. But for us, a majority of that churn is related either to lack of ROI, which can come from lack of usage, or can come from lower DocDel purchases, and lack of engagement on user parts. In other words, they're not using the product as much as they had hoped.

Maybe just touching on the B2C side then. I know we're entering a seasonal slowdown just as students are leaving class and everything. But maybe help me think about the improving CAC metrics. Are you spending less on marketing and still seeing better conversion? I guess, numerator versus denominator type question.

Yes, we are spending significantly less on digital ad spend, even though we are seeing our conversion rate out of the funnel. In other words, that creates a trial user. The conversion rates actually improved, which leads to flattish MRR on much less digital spend. And then some of the things we've released like the MCPs we've talked about have really had a big impact on retention. So we're seeing those customers stay longer, hence the improvement in lifetime value there. So I think if we continue to execute there, we may be able to get that thing growing again.

And maybe just one -- go ahead.

I was just going to say on the advertising spend side, we are able to -- I think I mentioned in the last call as well -- we are able to really manage that on a week-to-week basis. So we can turn it on as we get into the fall season, the return of academic cohort. And then as we approach summer, we can be a lot more deliberate on that advertising spend. So we'll continue to do that. And really having visibility into that and how that's converting has helped us manage costs and also pull that line where we're still keeping the MRR growing.

And maybe just last question for me. I know you guys are innovating on the AI front, making Scite and Article Galaxy integrated through MCPs, and also the integration of Resolute. But maybe if you could look 12 to 18 months out, I guess, what does your AI road map look like for new product launches? Where are you seeing maybe pockets of opportunity for new product development?

Yes. I think, obviously, Scite as a platform was developed as 100% AI. I think Article Galaxy has some opportunities within the Article Galaxy platform to implement AI in a copyright-compliant way to do several things to take friction out of the research process. So for those customers that use AG on a daily basis, they'll be able to get summaries of stuff that are in a folder, assuming they have the rights to do that; they'll be able to get a summary of an article, assuming they have the rights to do that; they'll be able to extract tables and other information, again, assuming they have the rights to do that.

So we'll continue to add. I think there are several points in Article Galaxy we can continue to enhance by layering AI in on top of that workflow. Both the MCPs are 100% AI. And I think the real value add there is we have another roughly 9 or 10 curated databases out of Resolute that we can integrate initially with those MCPs. So an MCP user will see, here's patent results to your question, here's clinical trial results to your question, and then ultimately, we'll add those other databases, which include drug databases and other research associated databases that'll be helpful to those researchers that they can add, turn on, turn off. Those are also revenue opportunities for us.

And then once they're in the MCP, which will happen actually very quickly, then we'll circle back and be able to integrate those into Scite, AG, or both so that you can have that access as part of your search results in Scite or in AG. So it's continuing to add basically curated unique data that gets added into the answers of the questions that you're asking, if you want them added, and giving you a broader view of all information related to your query instead of just scientific research.

[Operator Instructions] We'll take our next question from Derek Greenberg with Maxim Group.

My first is just on the Scite B2C2B pipeline. You guys had highlighted that in past quarters. I was just wondering if there's any progress on that front, just in terms of what you're seeing in the pipeline buildout from those opportunities.

I didn't pull that specific slice of our pipeline. I was looking at the MCP part that I mentioned in the call. But that continues to be a driver of a lot of our B2B sales, but I cannot tell you here whether it went up, went down, or was flat during the quarter.

Yes. I think I can give you a little color on that. It increased a little bit. We were talking about sizing it up about $100,000, and it went up probably about 50%. And we are, as Roy alluded to, starting to look at some of the newer products in the B2C side that will give users a little more MCP usage over the basic plan. So we think that some of the power users as well, and in the future teams, 2 to 50 seats, we think we can capture those cohorts in the B2C platform and then let them see the value of the MCP and other features on the B2C side and then convert them into B2B as well.

So we have a lot more of a road map on the B2C side. And like I said in my remarks, the retention and the engagement is much improved once we release the MCP. So that is a pathway from B2C to B2B as well.

And then my next question is just how to think about usage right now on the headless strategy? Like what you're seeing in terms of percent of pulls and usage that are coming outside of your core platforms and within their own workflows versus on the platform? And what you expect that mix to be over time as these new technologies gain more usage?

Yes. So usage of the MCP products, we see a multiple higher than our own products. Once they're integrated, they're integrated into all users that are pounding on an LLM, whether they're Claude or ChatGPT all day, every day. So we definitely are seeing usage that are literally, in some cases, a pretty big multiple over what we would see in a typical enterprise customer of the same size or same number of user seats. I will say a byproduct or related to what Dave commented on is that we are positioning the product at usage-based pricing.

So we are not doing MCPs that are basically unlimited. There is a limit. And when you hit the limit, you have to move to the bigger limit from a pricing point of view because we do believe that while we're going through this transition from our historic SaaS software and API business to more and more MCP-associated revenue, that revenue is going to be derived based on usage, which is a multiple of the SaaS software platform, as I mentioned. And we want to make sure we're not trying to sell a seat product into an environment that's 10x the usage without capturing the value in revenue to us of that usage, if that helps.

And then just my last question. You had previously said that you were seeing -- you were considering potentially selling into new segments such as like financial institutions, hedge funds, investment banks for their research. I was wondering if you've seen any traction in those segments or in any other new segments or if you had any comments on that.

We have a few cats and dogs in those segments. But right now, we are very focused in the new sales groups on corporate and academic, and I hesitate to really pull people off of what appears to be a pretty good pipeline and a pretty good TAM to go after these other verticals other than react to inbounds from those other verticals. So if we see one that we really think will yield results, then we may hire additional salespeople to focus just on those markets, but we have not done that at this point.

Thank you. At this time, there are no further questions in queue. I will now turn the meeting back to Roy Olivier for any additional or closing remarks.

All right. Well, thanks, everybody, for joining us. As a reminder, we will be attending the Three Part Advisors IDEAS Conference in New York on June 10. Qualified investors that are interested in attending, please reach out to Three Part Advisors to get scheduled. We look forward to speaking to you in September and to discuss the fourth quarter and full fiscal year results. Have a great day.

Thank you. This brings us to the end of today's meeting. We appreciate your time and participation. You may now disconnect.

[Operator Instructions] Good afternoon, everyone, and thank you for participating in today's conference call to discuss Research Solutions' financial and operating results for its fiscal 2026 second quarter ended December 31, 2025. As a reminder, this conference is being recorded. I'd like to now turn the conference over to your host, John Beisler, Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, everyone. Thank you for joining us today for the Research Solutions Second Quarter Fiscal Year 2026 Earnings Call. On the call with me today are Roy W. Olivier, President and Chief Executive Officer; Dave Kutil, Chief Financial Officer; and Josh Nicholson, Chief Strategy Officer.

After the market closed this afternoon, the company issued a press release announcing its results for the second quarter of fiscal 2026. The release is available on the company's website, researchsolutions.com. Before Roy, Dave and Josh begin their prepared remarks, I would like to remind you that some of the statements made today will be forward-looking and are made under the Private Securities Litigation Reform Act of 1995. Actual results may differ materially from those expressed or implied due to a variety of factors. We refer you to Research Solutions' recent filings with the SEC for a more detailed discussion of the risks and uncertainties that could impact the company's future operating results and financial condition.

Also on today's call, management will reference certain non-GAAP financial measures, which we believe provide useful information for investors. A reconciliation of those measures to GAAP measures is included in today's earnings press release as well. Finally, I would like to remind everyone this call is being recorded and made available for replay via a link on the company's website.

I will now turn the call over to President and CEO, Roy W. Olivier. Roy?

Thank you, John. Our fiscal Q2 was a bit of a mixed bag in terms of results. We experienced a decline in our year-over-year transactions resulting results primarily because of one known churned account and a few larger customers that saw significant declines in volume. Excluding those accounts, we were down about 2%, which would be in the range of what we consider normal in the business driven by our corporate research intensity and academic budgets. We expect that decline to continue in the second half of the year. We do have some levers to pull in this segment that we think will improve the results, and we're working hard on those initiatives.

We also continue to see increased competition in the B2C segment as reflected in our Q2 results and expect to see that headwind for the second half of the year. I'll also remind you that we focus on annual or multiyear agreements in the B2B versus the B2C business being primarily a month-to-month subscriber-based business. We will continue to focus on B2B annual and multiyear agreements as our primary growth driver going forward.

All that said, we are making product and sales process improvements to increase our conversion rate in this business neatly. We are shipping improvements, both software and new data like adding patent data faster than we ever have. We'll continue to innovate and improve our sales process and marketing investment to improve results in this area. Josh will speak to some of those product improvements later in the call.

We did see strong results in our B2B segment in both ARR bookings and net ARR bookings. The 47 net new deployments in B2B underscores that customers are excited about our current and future product development. The results reflect a better product sold through a disciplined and focused sales team. We saw good results across our corporate, academic, and upsell teams. We continue to see very strong year-over-year ASP growth at the pipeline stage. And as noted in our earnings release, we saw a 6% ASP increase during the quarter.

To be clear, I'm very happy with our results in terms of operating and net income, EBITDA and cash flow. We do have work to do in terms of the overall SaaS growth rate, and we'll work hard to return that rate to 20-plus percent.

I'd like to pass the call over to Dave to walk you through the fiscal quarter and full year 2025 -- I'm sorry, the fiscal quarter and FY '26 year-to-date financial results in detail, and then I'll wrap up with some comments and outlook for 2026. Dave?

Thank you, Roy. Good afternoon, everyone. Total revenue for the second quarter of fiscal 2026 was $11.8 million compared to $11.9 million in the second quarter of fiscal 2025. Our platform subscription revenue increased roughly 14% to $5.2 million. The growth was primarily driven by a net increase of 47 platform deployments from the prior year as well as expansion within our existing customer base through upsells and cross-sells. We ended the quarter with $21.8 million in annual recurring revenue, or ARR, up 14% year-over-year, which consisted of roughly $15.3 million in B2B ARR and approximately $6.4 million in normalized ARR associated with sites B2C subscribers.

Total incremental ARR for the quarter was $560,000, the highest organic second quarter in company history. While we did experience softness in B2C ARR, this was primarily driven by a concerted pullback in certain marketing channels where trial to subscriber conversion was lagging. We are prioritizing cohort quality and retention over short-term B2C growth, which we believe positions this part of the business for more durable and profitable contribution over time. Please see today's press release for how we define and use annual recurring revenue and other non-GAAP items.

Transaction revenue for the second quarter was $6.6 million compared to $7.3 million in the prior year quarter. The softness in transaction revenue was largely driven by previously discussed churned account and volume reductions from a small number of larger customers. Our total active customer count for the quarter was 1,321 compared to 1,384 in the same period a year ago. Gross profit for the first quarter was $6.2 million, up 6% from the prior year quarter. Gross margin was 52.4%, which constitutes a 350 basis point improvement over the second quarter of 2025. The increase is due to the ongoing revenue mix shift towards our higher-margin platforms business, which was also enhanced by expanding gross margins in the Platforms business.

Platform revenue accounted for 44% of the revenue in the quarter compared to 39% in the prior year quarter. On a trailing 12-month basis, the company's blended gross margin now stands at 50.8%. The Platform business recorded a gross margin of 88.1%, a 160 basis point increase compared to the prior year. We have been able to continue to expand gross margins in the platform business as our labor and hosting costs continue to increase at a proportionately slower pace than our revenues.

Gross margin in our transaction business was 24% compared to 25.2% in the prior year quarter. The decrease was primarily attributable to lower fixed cost leverage due to the reduced revenue base and some pressure on copyright-related margins. Total operating expenses in the quarter were $5.4 million compared to $5.7 million in the prior year quarter. Lower general and administrative expenses and lower stock-based compensation more than offset an increase in sales and marketing investments. We have been very intentional about keeping general and administrative costs contained as we allocate more resources towards growth initiatives, and we expect to continue driving operating leverage as we scale the business.

Net income for the quarter was $547,000 or $0.02 per diluted share compared to a net loss of $2 million or $0.07 per share in the prior year quarter. The prior quarter's results included a charge of approximately $2.4 million that was related to the projected contingent earn-out liability for site. Adjusted EBITDA for the quarter was $1.3 million compared to $963,000 in the year ago quarter or a 36% increase. On a trailing 12-month basis, our adjusted EBITDA margin was 11.8%, a 230 basis point increase from the end of calendar '24. I would also like to point out that historically, Q3 and Q4 are typically our strongest quarters for adjusted EBITDA. Turning to our balance sheet.

Cash and cash equivalents as of December 31, 2025, were $12.3 million versus $12.2 million on June 30, 2025. This includes 2 payments related to the site earnout, which consisted of a total of $2.6 million in cash and the issuance of approximately 487,000 shares of stock. Cash flow from operations was $1.4 million, a 35% increase from $1 million in the second quarter of fiscal '25, reflecting both higher profitability and disciplined working capital management. We continue to believe we can grow our cash balance in fiscal '26 while funding the remaining site earn-out obligations entirely from operating cash flow. We also ended the quarter with no outstanding borrowings on our revolving line of credit, providing additional flexibility in our balance sheet. As we move into the traditionally stronger back half of our fiscal year, we believe our balance sheet is well positioned to capitalize on high-return growth opportunities through a strategic approach.

Looking back at the first half of the fiscal year, we delivered a meaningful year-over-year increase in net deployments, cash from operations and EBITDA. Even with top line pressure in B2C and transactions, we grew gross profit and improved our overall profitability profile, demonstrating the operating leverage inherent in our business model as we scale platform ARR. For the back half of the year, our objective remains to exceed fiscal 2025 EBITDA levels in each of the remaining quarters and to drive further growth in the cash flows generated by the business. We expect this to be supported by ongoing platform subscription growth, a more stable trajectory in transactions and continued rigor around operating expenses, even as we make targeted investments in sales, marketing and product.

Taken together, we believe we are positioned to exit the fiscal year with a stronger earnings and cash generation run rate while also maintaining balance sheet flexibility to fund disciplined high-return growth initiatives. I'll now turn the call back to Roy. Roy?

Thanks, Dave. I'd like to remind our investors that we typically outperformed the first half of the year in the second half. We expect that seasonality to continue with B2B and transactions. One of the reasons we have seen strong B2B growth is due to where we are going from a product perspective. We believe we're on the right track and customers are voting with their wallet, confirming our belief. I think we've all seen the recent market response to AI over the last 10 days, which impacted our stock price like many others. We are clearly in a sell-first reaction to AI news.

In our view, enterprise software that accelerates work does face some level of disruption. We continue to believe that our unique capabilities and data are not exposed to those tools or LLMs in general. We offer less hallucinations and better search results because we can search behind paywalls. We also offer rights management, which includes what rights you have with articles you have purchased, things like AI rights, reprint rights, et cetera. We manage the company or library's entitlements, which is what articles they buy one by one, what they subscribe to, what they have discount packages for and what their library already has in it. We offer unique tools to help researchers understand the quality and efficiency of articles they're looking at.

While we continue to develop a full workflow tool like the Bloomberg terminal, but for researchers, we're also focused on delivering that unique capability within whatever workflow the customer has. That sometimes is their own tools. And moving forward, we will support companies that standardize on an LLM. We have that working now. I can use Claude or ChatGPT to ask a research-focused question, and Claude will use our site connector to generate a fully cited answer using only peer-reviewed articles, leveraging our unique data, our citation graph and our rights management data. The user can then ask Claude to talk to our products directly in the tool to get those cited articles and we will integrate within the rights management, the company library and their entitlements to deliver those articles. We can even report what rights they have for those that, in many cases, help them acquire the rights to do what they need to do.

For example, they may want to summarize several articles in AI to prep for a meeting. We can help them do that in a copyright compliant way. Josh will speak to our new products shortly. In fact, I'll turn the call over to Josh to talk a little bit about our thinking on product strategy and usage. Josh?

Yes. Thanks, Roy. Hello, everyone. So today, I want to talk about 2 fundamental shifts happening in our business that represent not just operational improvements, but a transformation in how research solutions creates value. These shifts center on our move to API and AI integration and our evolution from a document delivery company to what I call an Answers and access platform. These shifts have been ongoing, and I've discussed them in previous calls, and I think position us well for all the macro events happening with regards to AI. I'm going to talk about 2 things here, API and AI integration sales becoming infrastructure and then another theme later.

The biggest change in our business is how customers are buying from us, both new customers as well as many of our original customers instead of just purchasing seats for researchers to log into our platforms, customers are now integrating our services directly into their own systems and AI tools. We're no longer only selling seats to a platform. We're selling infrastructure. top pharmaceutical company doesn't want their researchers to switch between 10 different tools. They want citation verification and Article Galaxy rights cleared content accessible from wherever their researchers already work, whether that's an internal AI system or a proprietary drug discovery platform. This is what we call API sales and is driving some of our largest contracts to date. These deals are typically 2x to 5x larger than traditional seat-based sales and they're stickier.

When we are embedded in a customer's core infrastructure, we become mission-critical, not just another subscription they might cancel. Here's why these matter. Every company building AI tools for research faces the same problems. They need accurate citations, copyright compliance and access to scientific articles. Instead of each company solving these problems separately with thousands of publishers, we're building the solution once and making it available to all of them through simple integration. This leverages and repositions our partnerships with publishers, which have taken over a decade to form in some cases. We now provide access to more than a dozen specialized research data sets spanning different disciplines and regions from life sciences to physical sciences, from research articles to patents and to clinical trials and more. When an AI platform integrates with us, they're not getting access to one database, they're getting comprehensive coverage of global research. That breadth is what makes us valuable infrastructure.

We're already seeing strong traction with this approach. Multiple enterprises are beginning to use our technology to power their internal AI research assistance. While still early, publishers are also in discussion with us about using our infrastructure to offer their subscriptions with AI access to their customers without having to build the copyright and citation systems themselves. The bottom line, we're moving from selling seat-based software subscriptions to selling infrastructure. Infrastructure businesses have better economics, higher contract values, lower churn and more expansion opportunities. One large pharma company is already making over 1 million calls to our API per year.

Second stage I want to talk about is from documents to what I call Answers and Access. This brings me to this bigger picture. Research Solutions is evolving from a document delivery business to an Answers and Access platform. The old model is straightforward. Researchers need articles, we deliver articles transaction complete. The new model is different. Researchers don't start by thinking, I need this specific article. They start with questions. I need to understand this mechanism. I need to validate this hypothesis. I need evidence for this claim. They're looking for answers first, not documents. Our job is to provide both. We give them answers through site, helping them find relevant citations, verify claims and understand what the research says. When they need to go deeper, we give them access through Article Galaxy, delivering the full articles with proper copyright clearance.

That's what I mean by Answers and Access. We're not abandoning document delivery, we're making it part of a complete solution that starts with answering research questions. The market is validating this approach. Customers who only used us for article fulfillment are now deploying site across their entire organization for literature view and research validation. Academic institutions are expanding from researcher access to broader applications and publishers who just saw us as a fulfillment vendor are now asking how we can help them create new revenue from AI usage of their content. Here's why this matters.

These 2 ships, API integration sales and Answers and Access work together and reinforce each other. Our API approach makes it easy to embed our services wherever researchers need answers and our Answers and Access model means we're not just a commodity service, but mission-critical infrastructure for research decisions. This is why we're building integrations and developer tools, including a new API for Article Galaxy, which we have recently deployed. This is why we believe the long-term value of research solutions come from being the infrastructure that makes AI trustworthy for scientific research. No longer just in the document delivery business. We are in the research intelligence business in a world where AI is becoming the primary way people access information, the winners will be the companies that provide verified compliant citation-backed infrastructure that researchers and companies can trust. That's the future we're building for research solutions is going, and we're excited by that.

Thank you, and I'll turn it back to Roy now.

Thanks, Josh. There was a recent post on X talking about there being 2 kinds of software companies. Type 1 is software humans click on, things like dashboards, CRMs, project management, other tools, basically products that provide a human interface to do a task. If all you do is automate a task, then AI is a threat. Type 2 is software bots. These include APIs, databases, authentication, infrastructure and more. These are functions that are unique like curated data, data behind paywalls, customer entitlement information, customer library information, rights information, et cetera. These have unique capability and typically have tolls or charges associated with usage.

Our company started by having unique access to the world's peer-reviewed scientific content. We still have unique capability in this area today in many ways, including capability to help customers find obscure or hard to obtain content. We then added entitlement capability, rights information, the ability to store a corporate library, et cetera. From that, we created Article Galaxy to use these unique databases, authentication capability and rights information to service our customer. What's happening now is we're going full circle by applying our unique core capabilities to AI and LLMs. We continue to believe that we're well-positioned to take advantage of this transition to an AI world. Our strong B2B results give us confidence that we're on the right track.

With that, I'd like to turn the call back over to our operator for Q&A. Operator?

[Operator Instructions] And we'll take our first question from Jacob Stephan with Lake Street Capital Markets.

Maybe just first to start off, really nice B2B results. I'm wondering if you could touch a little bit on the pipeline there. Maybe kind of compare and contrast are the larger deals specifically related to the kind of headless API, AI model that you guys have been working on the last few quarters here?

Yes. The pipeline has grown consistently quarter-over-quarter and year-over-year. And a lot of the pipeline now is these "API deals" where we're integrating with larger customers in the workflow. Therefore, they're much larger deals.

And maybe from a B2C standpoint, obviously, you have kind of the favorable student enrollment trends working back in your favor. Any kind of early anecdotal comments you could give us on how that business is maybe returning to growth in the second half?

Yes. I don't think we knew if it will. That's why I specifically excluded it when I said second half of the year, we'll see transactions in B2B business be stronger than the first half. What we're seeing today is 2 things. One is much more competitive digital marketing spend to attract trial customers. So somebody is out there doing Google search, they find us. Now they also find everybody else that's entered this space in the last year or 2, and we're all competing for the same keywords. So we're seeing a lot more competitive to get users to the platform. More importantly, we're seeing the conversion rate of trials to users lower than it was a year ago.

In other words, we actually still attract as many trials as we did a year ago, we're just converting at a lower rate. And that's where Josh and his team are doing a lot of great work to add things like patents, patent data on top of the peer-reviewed scientific research data, new features and functionalities in order to try to move that conversion rate up. Anything you want to add there, Josh?

Yes, I think that's right. I think we've always focused on having more stable, higher per user basis in B2B licenses. And so a lot of product decisions have been to -- around that, right? And so we're seeing this B2C2B sales motion, which oftentimes someone will come in, test it for a month and then they bring it to their whole team at the large pharma. So I think a lot of this is also us pushing more and with more focus on B2B, where the money is a higher average cost per user and also more stable contracts, yearly by default versus monthly, which is where most B2C is.

Yes. I will say also, in Q2, we did see a reduction in churn year-over-year. and that has continued into the third quarter here. So that trend is going in the right direction. It's exactly what Roy talked about with the conversion, trials to new subscribers that's really impacting us.

Maybe just one last one. Dave, you've been there for a couple of months, it seems like now. Maybe top 3 things that you're kind of looking at and to make improvements at.

Yes. So we are really looking at what we can control. So for me, it's the operating expense rigor I talked about. We're going through. We have a cost savings program that we're looking at all of our vendors, looking for redundancies and really trying to improve that, and we're excited about the results there. So that's the first thing. Second thing is really looking into and diving in with the sales team and really trying to understand the root cause of some of the churn in the business and make sure we're proactively addressing that. And then working capital management is really important to us. Again, one of the things we can definitely control. We're proud of the results that we've had with cash flow from ops, again, funding the site earn-out payments out of operating cash flow, and we continue to manage that and look for cash to increase. So we're meeting on a weekly basis and making sure we're achieving our projections and having a forecast that we can deliver to shareholders.

We'll take our next question from Richard Baldry with ROTH Capital.

You've done a good job talking about using AI sort of at a productized level externally. Talk also about -- it seems to me there's a lot of room for adoption of AI technologies internally to lower costs or improve output speed, efficiencies, whatever in areas like development or back office. So how far into that sort of process do you feel like you are? Do you think you can meaningfully control as you grow or lower costs on the fly?

Josh, do you want to talk about what you're doing in the software area?

Yes. We are fully leveraging the most advanced AI on the product team with the development team, that's AI for software development, AI for review, software that is also automatically looking at accessibility. I've probably become annoying internally just because I'm such a champion of it. And I think it's pretty dramatic in the deployment and how quickly we're moving things. And so that has really opened my eye up, and I think that also builds into our strategy, recognizing how powerful these AI tools are and how we can exist in this world in a very defined way where our strengths are leveraged.

And so I think it's making everyone not just on product, I've done a lot of internal trainings even as of today, but across the team, a lot more productive, a lot more efficient and can really extend from product development, better interactions with publishers, better interactions with customers, and I would say, while I'm really excited by it, and we've made a lot of progress, there's still more to do. And so I think there's a big exciting upside on that just because it has transformed the way that I personally work, our team works, and I think that's starting to permeate throughout the company, which is really great.

I was just going to say having the right tools is very important, and that's exactly what Josh has identified within the organization is having the right tools to be able to produce the analysis from our side in the back end of the business that really prioritizes those areas where we can save time and money. And we're still in the early innings of this. Like Josh said, he's doing training this week, but it's expanding rapidly. And you can see the amazing thing is you can see the use cases immediately. And you could see the results in savings in time and money right away. So early innings, but we're definitely -- we have the right tools at our disposal right now.

Yes. I would just add, AI for us so far is really a productivity multiplier. It's not a cost reduction product. It's being used heavily in development. Josh has been training people internally how to use the new QA tool set to develop sophisticated reports that we previously paid an external data scientist to do. And in fact, a project we talked about in the meeting today related to identifying specifically dollars associated with this B2C2B flow that we talk about from a sales perspective, we'll do all of that in AI using basically quad tools. However, on the other side of the coin, like the 60 rough people we have associated with the DocDel business, they do things that it's hard to replace with AI. So there may be some slight savings opportunity there. But most of what we've seen with AI so far has been improving productivity dramatically amongst highly paid softwaregers, FD&A people, data analysts, et cetera, not necessarily a cost reduction at this point.

And then last for me. If you look at where you rise to the level of sort of infrastructure and it becomes a much higher ASP, can you talk about how large that market opportunity is, whether that's -- or what number of verticals it can apply to sort of give us an idea where that can take for you up to over the next few years if that traction continues to grow?

Well, I think we're still serving the same vertical markets and have a similar TAM that we've had before. I think the ASP potentially is 20%, 30% higher than it is now. But quite frankly, we are going to have to explore and develop some new pricing models because of the different way the products work. When we talk about the MCP product and AI, we talked a little bit about -- it's really a number of calls analysis as opposed to users per week per month analysis. So we've historically always looked at how many users are using the product. Looking forward, some of our customers that have implemented this technology are doing 200,000 calls a month, approaching 300,000 calls a month. and that's going to be a different pricing model and how we share that model with the publishers to represent their contact with the MCP, this is going to be developed and probably fine-tuned over the next 12 months.

So I don't think we fully understand yet the pricing side. We do know it's going to be a lift from where we are today or we believe it will be a lift from where we are today. And frankly, we are excited about the change because user seat licensing is very different from the massive usage you see from these AI engines and people asking questions of them. Anything you want to add to that, Josh?

Yes. I'd just say on the kind of sales cycle and go-to-market, the trial is really kind of a technical buyer and maybe some executives, they're looking at this. It seems shorter anecdotally where it's, hey, here's a week, you test it out, it serves that, it's plugged in, and then it's used by potentially thousands of people. This is different from, say, "Hey, we're going to invite 30 people. We're going to train people. We're going to repeat that training." Maybe one guy doesn't like it, maybe one person does like it, et cetera. And so I think the sales is a lot more straightforward. It's a lot more repeatable. It's a lot more clear also on what we're serving. And then again, if you build it into what they've already invested internally, sometimes in some of these competitive tools, it really scales.

And so I think it's focusing us, and I think that's a good thing. And yes, our tiered model of API calls probably will evolve. We do scale it up with usage. And so that's also a nice thing as it gets used more, it scales up per call. So pretty excited about that.

[Operator Instructions] We'll take our next question from Derek Greenberg with Maxim Group.

I wanted to first touch on just the transaction segment. You had called out you expect further declines in the second half. I was wondering if that's on a year-over-year basis or sequentially from current levels. And I was also wondering in the prior quarter, you had called out that it was due to churn from one large account as well as a pullback from other large customers. I was wondering if you've seen any additional churn or pullback from other customers or if the declines are largely related to that customer set you previously mentioned?

Yes. In terms of transactional business, it is predominantly one large customer churn that we expect to continue. We typically do more DocDel in the second half of the year than we do in the first half. We expect it to do more in the second half looking forward than it has done in the first half of this year. However, we still do expect to post a year-over-year decline because of the churn. I think I mentioned somewhere or maybe it got removed from an earlier draft that if you pull out the churn customer, transactional business was down about 2% year-over-year. That's within what we would consider to be normal seasonality and usage that we see in that business. There are sometimes when we have an unexplained decline because of research intensity. There are sometimes we see an unexplained increase because somebody else has started a research spike into some particular area.

And we look at those 1,300-odd customers that are buying DocDel from us, really the top 10, top 20 customers, again, if you pull out the churn, are the ones that are driving most of that variability, like a couple of our biggest customers, they bought less this year than they bought a year ago. And we call them. In fact, one of them is so big, I call them personally because I deal with them and I'm like, why are you guys down? And he's like, "I don't know. I guess we're just doing less research." So sometimes the customer itself doesn't know why they may be up or down on a year-over-year basis. But anyway, so we do expect second half to be stronger. We do expect to continue to show a year-over-year decline because the churn customer. And we'll see what happens with the kind of the normal seasonality where we fluctuate between a slight loss and a slight gain. We'll see how that works.

And then just on the API business, I was wondering if you could maybe talk about if you see potential threat in, say, like LLM providers going out and licensing similar content to what you guys offer or if you're largely insulated from that and benefit from having the long tail of research? Or if you could just talk about that a little bit more.

I mean just to be clear, think about the API business as it's either an API or it's what's called an MCP, which is basically a connector to an LLM. So it's either the customer developed their own internal tool set and they're using our API or we're connecting to an LLM where you're basically asking questions of Claude that's talking to site to get you a better answer than it would get over the general Internet. And then once you get the fully sided answer, you can again ask Claude to get you the articles, and we will check your entitlements, we will check your rights. We'll go get you the articles. We'll not only deliver them to you wherever you are, but we'll stick them in your corporate library so that you have official "ownership of those" and they're in your company library. So when we talk about API, the API and MCP function very, very similar. Just one is targeted toward customers that have internally developed tools and the other is specifically to work with an LLM.

To your second question about the -- will the LLMs duplicate that capability? None of them do today. They ultimately will start to duplicate some of that capability or a publisher will create their own MCP to connect to an LLM. And so for example, if you pick the top 5 publishers, they will probably all create their own MCP that you can connect to an LLM that's going to get you a better answer than just abstract data, but not the full article. And as we see that happen, -- what -- where we will play is the long tail of medium and small publishers that don't have the capability or the wherewithal to create their own MCP, #1. And #2, people don't want to have 200 connectors into their LLM, just like our largest customers don't want to negotiate 200 agreements with 200 publishers. So they may have direct agreements with the top 5 publishers, we fill the gap of the next 100 or 150 publishers. And I think the same thing will play out in that MCP space, basically the AI space. Josh, anything you want to add to that?

Yes. I think just to be clear, there's 2 endpoints here. There's really calling site, which is to get an answer and then calling Article Galaxy to get access. And so again, most of our platform customers on Article Galaxy, they're not using API today. Our API was a bit outdated. We refreshed that. We've relaunched that, and we're now starting to push that internally. But I think going to where there's massive amounts of users is going to drive usage across both of those. I think Roy touched upon this, but again, just to expand upon it, we're also working with publishers to say, "Hey, we already have your content, we have the infrastructure. We can handle authentication. We can handle tracking of usage." So the AI usage of articles, which is something entirely new. We can sell articles on your behalf. We can enable entitlements via subscriptions, via tokens, all these different things. we can do very seamlessly with these different chatbots, again, Claude, internal Claude, ChatGPT, et cetera.

So we're having active conversations with publishers to try and help them navigate this. They don't have the wherewithal to necessarily build this unless they are one of the big 5. And so we're working as an infrastructure partner. And I think that could open up some potentially interesting new revenue opportunities where maybe there's a cut of subscriptions or there's some other fee. And I think that there's 2 levels there. One is exploring how do we serve current publisher customers and then how do we scale these up to potentially anyone using them, so LLMs from the public. So excited about that progress. And I think, again, very well positioned with our relationships, the technology and then the technology really helps both products. T

he last thing I'll say, there was this assay that stood out to me. People have been talking about killing the "PDF" for a long time. I think the PDF is here, but I think what I found interesting is he described this move to research objects, right? When people want to get answers, they will need a full text article, certainly for compliance reasons, but they're starting to get these objects, right? And then these [indiscernible] statements, the chunks that we have are research objects in many ways. And so again, really excited about kind of where we are, and I think that's proving out internally with the B2B and then there's a lot more room to run. I'd also say this is we've seen some new customers outside of the space and so looking at financial institutions, hedge funds, investment banks, things like this, utilizing APIs for some of their own research.

We'll take our next question from George Melas with MKH Management.

I think my question was largely answered by the previous answers, but maybe let me ask a little bit more, and I don't know if you can answer, but how do you see your relationship developing with that long tail of publishers, so not the top 5. But -- and how many have you had sort of discussion about doing this DocDel with all of them? And how are they reacting? What -- how do you see that developing? And what are the hurdles to getting that new functionality and that new service that would benefit both them and you?

Just to provide a little bit of a foundation. I mean that long tail has been our sweet spot for a decade, right? I mean our larger customers -- largest customers we have, they negotiate rights direct for everybody else, and then we typically walk in with that long tail. And Josh, I'll let you talk about our progress in talking with the long tail about AI rights, DMCP, et cetera.

Yes. I think if you look at how things are sold, this is a real pain for publishers, right? They are having subscriptions, their subscriptions are being challenged. They're like, what do we do, right? They have our content being used in AI, et cetera. And so we do have a sweet spot with these publishers, but I actually think there's opening up relationships with new publishers where we haven't spoke to them. Some of those large ones where I think there's cooperation and then small ones that maybe have just been hold out for other -- some other reason.

So I have had a slate of calls with small, medium and large publishers on this topic specifically. I have more coming up. And I think what is interesting is that, A, we're publisher neutral; B, we've already done a lot of the processing of their articles. So there's no upfront cost to them. We're already building this. We show them, "Hey, here's how you can monitor usage. You want to turn it on for a customer, we'll handle that. You want to turn it on in a specific way through Access Justice Journal or that we'll handle that. And so there's been a real pull, I would say, because we have this long-term relationship. And because I think it also accelerates that other part of that business where it's not just driving AI reads necessarily, which will help them demonstrate the value of their product, but also in the end, drives DocDel. Because at the end of the day, people will still need full text or they'll need full text for the AI to use based off these chunks. And so I think it's been a good evolution, and I hope that evolution continues because I think it strengthens us kind of in this new world.

And do you see some impact on revenue at this point right now? Or is that just way too small to even significant impact?

I think it's still pretty small, but we're starting to get some traction there.

Yes. And I think that's part of this puzzle, right? There's different ways of someone wanting to use AI with an article, right? There's RightFind, which we had talked about previously, and that's AI rights at the individual article level, but people also want AI rights for a group of articles for hundreds of articles or for thousands of articles, right? And I think this MCP and this work that we're doing with publishers really says, hey, you can sell a subscription and you can sell a subscription with AI on top of it. We are the infrastructure for that, the authentication, the tracking, the usage of that, et cetera. And I think that helps publishers tremendously because they're under budgetary impact and the usage is not just DocDel declining, but subscription declining as people are reading less, right, because of this zero-click phenomenon.

And so I think it is too early to kind of tell, but I think we're in the right place. And we do see these big publishers making big announcements with MCPs, but I think we have a somewhat differentiated moat with our citation graph, the DocDel and RightFind aspect of it and the fact that we're not a large publisher that might be off putting for a smaller publisher to work with.

And with the new customers, and you've had real good success with signing up new B2B customers. Are they embracing that in a meaningful way? Do you see sort of a new trend there?

Yes. I think, again, we have a floor for the API pricing, which is significantly bigger than our average contract price already, and that's just the floor. So again, I think some of this wasn't our focus previously. We've shifted a lot of our energy to understanding this. And I think as Roy highlighted, some of the business models around that will evolve, but it's highly encouraging because I do a lot of these API sales because there's some technical aspect of this. But it is almost in some ways, more straightforward because it's easier to test. You kind of see the improvements right away. And then the value is pretty obvious, right? You want to go spend 10 years and talk to every single publisher, go ahead and do that. You want to work with us, get access to patents, get access to research articles, give the ability to buy, rent an article, all in one place, it's significantly easier. And so that pull is there. And I think, again, the larger invoices are going out to reflect that.

At this time, there are no further questions in queue. I will now turn the meeting back to our presenters for any additional closing remarks.

Well, thanks, everyone, for joining us today. As a reminder, we will be attending the ROTH Capital Partners Conference in Dana Point, California on March 23. Looking forward to speaking with you then. Thank you.

Ladies and gentlemen, this concludes today's presentation. You may now disconnect.

Thank you.

Thanks.

Good afternoon, everyone, and thank you for participating in today's conference call to discuss Research Solutions' financial and operating results for its fiscal 2026 first quarter ended September 30, 2025. As a reminder, this conference is being recorded. I would now like to turn the conference over to your host, John Beisler, Investor Relations.

Thank you, operator, and good afternoon, everyone. Thank you for joining us today for Research Solutions First Quarter Fiscal Year 2026 Earnings Call. On the call with me today are Roy W. Olivier, President and Chief Executive Officer; Bill Nurthen, Chief Financial Officer; and Josh Nicholson, Chief Strategy Officer.

After the market closed this afternoon, the company issued a press release announcing its results for the first quarter of fiscal 2026. The release is available on the company's website, researchsolutions.com. Before Roy, Bill and Josh begin their prepared remarks, I would like to remind you that some of the statements made today will be forward-looking and are made under the Private Securities Litigation Reform Act of 1995. Actual results may differ materially from those expressed or implied due to a variety of factors. We refer you to Research Solutions' recent filings with the SEC for a more detailed discussion of the risks and uncertainties that could impact the company's future operating results and financial condition.

Also on today's call, management will reference certain non-GAAP financial measures, which we believe provide useful information for investors. A reconciliation of those measures to GAAP measures is included in today's earnings press release as well. Finally, I would like to again remind everyone this call is being recorded and made available for replay via link on the company's website.

I would now like to turn the call over to President and CEO, Roy W. Olivier. Roy?

Thank you, John. The first quarter continued our progress in improving our B2B new logo sales teams as well as our transformation to becoming a comprehensive SaaS and AI solution for scientific research. It is the strongest organic first quarter B2B results on record. Total ARR for the quarter is up 21%, driven by that strong B2B performance. That performance includes closing some of the largest deals in the company's history, including new platform sales to Real Chemistry, a top 10 pharma company and others. This resulted in the company's ASP increasing and also contributed to driving our second highest quarterly adjusted EBITDA result as well as strong cash flow.

I'd like to pass it over to Bill to walk you through the financial -- I'm sorry, the fiscal first quarter financial results in detail, and then I'll wrap up with some comments and outlook for the remainder of the year, and Josh will discuss our strategy. Bill?

Thank you, Roy, and good afternoon, everyone. Total revenue for the first quarter of fiscal 2026 was $12.3 million compared to $12 million in the first quarter of fiscal 2025. Our platform subscription revenue increased 18% to $5.1 million. The growth was primarily driven by a net increase of platform deployments from last year as well as upsells and cross-sells into our existing customer base.

We ended the quarter with $21.3 million in annual recurring revenue, or ARR, up 21% year-over-year, which consisted of roughly $14.8 million in B2B ARR and approximately $6.5 million in normalized ARR associated with Scite's B2C subscribers. By Q1 standards, this was a strong quarter for ARR growth. Total incremental ARR for the quarter was $375,000 compared to $195,000 in the prior year quarter, which represents a 92% increase.

Moreover, B2B growth was especially strong at $561,000 for the quarter, up from only $128,000 last year. While we did experience a decline in B2C ARR, last year's increase was relatively minimal in what is seasonally a challenging time for B2C growth. Additionally, on a positive note, we are also seeing an increase in B2C leads that are transitioning into B2B business. Please see today's press release for how we define and use annual recurring revenue and other non-GAAP items.

Transaction revenue for the first quarter was $7.2 million compared to $7.7 million in the prior year quarter. You may recall from our fourth quarter fiscal year 2025 earnings call, in which we discussed an 8% decline in transaction revenue that we expected transaction growth to continue to be challenging minimally through the first half of fiscal 2026. As a result, the decline for the quarter was in line with our expectations and was actually a little improved over the decline we experienced in the fourth quarter of fiscal 2025. Our total active customer count for the quarter was 1,326 compared to 1,390 in the same period a year ago.

Gross profit for the first quarter was $6.2 million, up 8% from the prior year quarter. Gross margin was 50.6%, a 270 basis point improvement over the first quarter of 2025. The increase is due to the ongoing revenue mix shift towards our higher-margin platforms business, which was also enhanced by expanding gross margins in the Platforms business. Platform revenue accounted for 42% of the revenue in the quarter compared to 36% in the prior year quarter.

On a trailing 12-month basis, the company blended gross margin now stands at 50%. The Platform business recorded gross margin of 88.1%, a 70 basis point increase compared to the prior year quarter. We have been able to continue to expand gross margins in the Platforms business as our labor and hosting costs continue to increase at a proportionately slower pace than our revenues.

Gross margin in our transaction business was 23.8% compared to 25.7% in the prior year quarter. The decrease was primarily attributable to lower copyright margins and lower fixed cost leverage due to the reduced revenue base. Total operating expenses in the quarter were $5.3 million compared to $5.1 million in the prior year quarter. Higher sales and marketing expenses were partially offset by lower general and administrative expenses and lower stock-based compensation expenses. We have made a concerted effort to keep general and administrative costs contained as we increase investment in other parts of the business.

Net income for the quarter was $749,000 or $0.02 per diluted share compared to $669,000 or $0.02 per diluted share in the prior year quarter. Adjusted EBITDA for the quarter was $1.5 million compared to $1.3 million in the year ago quarter, a 16% increase. This was the second best adjusted EBITDA performance in company history and with Q3 and Q4 typically being our strongest quarters for adjusted EBITDA, we are off to a very strong start for the year.

Turning to our balance sheet. Cash and cash equivalents as of September 30, 2025, were $12 million versus $12.2 million on June 30, 2025. It is important to note that during the quarter, we made our first payment of the Scite earnout, which consisted of $1.3 million in cash and the issuance of approximately 265,000 shares. Despite the cash outlay related to the earnout, we are almost back to where we ended Q4, which means our operational cash flow remains very healthy.

Cash flow from operations was $1.1 million compared to $843,000 in the first quarter of fiscal 2025, a 31% increase. We continue to believe that we can grow our cash balance in fiscal year 2026, while also continuing to service the Scite earn-out from operational cash flow. Further, there are no outstanding borrowings under our revolving line of credit.

As we look ahead, we are off to a good start for fiscal year 2026. As you may recall from our prior earnings call, I discussed some of the seasonality in our business with respect to adjusted EBITDA. Typically, we see a dip in adjusted EBITDA between Q1 and Q2 before rebounding to higher levels of adjusted EBITDA in Q3 and Q4 with Q4 usually being our best quarter of performance.

We think that it is likely that this seasonality will play out again in fiscal 2026, but we think the dip will be less pronounced in Q2 from where it was last year, and there's also a shot at some EBITDA growth sequentially between Q1 and Q2. All that said, our goal remains to experience outperformance to fiscal 2025 in each of the remaining quarters for the fiscal year. To the extent we can execute on that, it will be another record year for the company, and we will continue to experience expansion in cash flows generated by the company.

I'll now turn the call back to Roy. Roy?

Thanks, Bill. A few comments about the transaction business. In last quarter's call, we discussed that we thought the year-over-year decline was driven by 0 click searches. Further research suggests that may not be the case. Our Academic segment is growing and the corporate segment is declining with about 60% of that decline coming from churned account. The remaining churn dollars are primarily 2 customers who are very large and are buying less year-over-year. Both report that this is based primarily on the current economic environment or changes in their research priorities. In short, 3 customers are largely driving the year-over-year results.

As you know, we started investing about this time last year in more of B2B sales resources. We continue to actively monitor these investments to ensure that we are seeing a return on them. In looking at the first half of FY '25, actual results versus our current forecast for the same period in FY '26, we expect to see new ARR growth to be materially higher than the incremental sales investments we made. We expect the new logo teams to generate over $1 in new ARR for every dollar invested. We expect to see the churn upsell team generate a bit under $1 in new ARR for every dollar we've invested. This suggests a payback period of a little over 1 year on products that have a 6- to 8-year lifetime value. In my view, we need to continue to show improvement on the upsell churn teams, but overall, the investments we have made are working.

As previously discussed, the new sales process is resulting in rising ASPs. We continue to set records in terms of total contract ARR and per seat ARR through better sales execution, especially with our AI-based products. While we saw less-than-expected B2C net ARR growth in Q1, we are starting to see more traction toward B2B strategic revenues from B2C.

As you know, B2C are primarily month-to-month subscribers who are attending school or researchers in a corporate account that are trying out the product. We are seeing an increasing number of those users try the product and then report back to their academic or corporate enterprise that it should consider an enterprise subscription. A year ago, in Q1 FY '25, about 50,000 of our total sales pipeline came from this B2C to B2B channel. At the end of Q1 FY '26, over 1 million of our pipeline came from B2C.

Now I'd like to turn the call over to Josh to talk a little bit about our current thinking on product strategy. Josh?

Thanks, Roy, and hello, everyone. I'm going to spend a few minutes connecting what we're seeing in the market to the strategic decisions we've been making over the last few quarters, especially around AI rights, Scite, Article Galaxy and our role with publishers.

A few key trends worth calling out. One is enterprises want to use AI on their articles. Another is publishers are moving into AI licensing. Then a third one is the AI usage of articles is not being tracked. In response to these market trends, we have made various product additions, tweaks to our go-to-market offerings and made sure we're aligned with where the market is going.

On the first point, enterprises wanting to use AI on research articles, we've launched an AI rights offering in Article Galaxy called [ RightsDel ]. RightDel allows the researcher to acquire AI rights for documents they already own. The revenue we charge for those rights is split between the publisher and research solutions in a similar manner to the transaction business we have today. This gives publishers a way to monetize AI usage of their content either on an article-by-article-by basis or across the company's existing library. We believe this will grow platform seat revenues, lead to more cross-sells, increased transactional purchases and ultimately decrease churn.

On the second point, publishers moving into AI licensing, we are working with our publisher partners to offer an AI gateway product based on Scite, in which customers can purchase the rights to use AI with all of that publisher's content to materially improve the productivity of their research teams. This would create an ARR upsell opportunity for publishers in their academic and corporate customers and as a revenue share with research solutions while deepening our role as their AI technology partner. This is a major challenge for our industry, especially the long tail of medium and small publishers, and as we solve this gap, it could make Scite's smart citations a key part of AI infrastructure and a key partner to publishers.

On the third point, AI usage of articles not being tracked. Today, the industry standard is counter-compliant usage metrics. These usage metrics show how many times articles have been downloaded or read, and they're central to how libraries and publishers assess the value and ROI of subscription spending. There is no equivalent for AI usage today. By working with publishers to deploy our AI gateway, we can introduce AI-specific usage metrics that play a similar role to capture for traditional usage, giving publishers visibility, creating a basis for pricing and value discussions and enabling a clear upsell and revenue share opportunity for AI analytics and rights on top of existing subscriptions. This would be a huge value to both the publisher and our corporate and academic users.

Researchers expect tools like Scite Assistant and Article Galaxy to help them answer concrete questions, summarize bodies of work and find key insights across many articles using AI. We believe we're well positioned to do that in a copyright compliant way that delivers real value to customers while generating new sources of revenue for Research Solutions and our publisher partners.

Thanks again for having me, and I'll turn it back to Roy.

Thanks, Josh. Looking forward, we expect to focus on several areas for the remainder of the year. First, improve Scite B2C net ARR growth through product improvements, the pace of delivering those improvements, the marketing and sales messaging around our unique capabilities in this segment.

Second, continue to show improvements in overall ARR growth and ASP growth through better sales execution and improving our products. Third, demonstrate improvements in the retention and upsell part of the business by driving proactive versus reactive customer engagement through better health scoring and product improvements. Fourth, continue to innovate in the transaction or dockdel space to return that business to a flat to slow growth business; and lastly, manage our cost structure to continue progress toward our goal of a weighted Rule of 40.

With that, I'd like to turn the call back over to our operator for Q&A. Operator?

[Operator Instructions]. Our first question comes from Jacob Stephan with Lake Street Capital Markets.

Nice quarter on the B2B growth and on profitability as well. I just want to talk about the attach rate on the AI rights add-on product. Maybe you could help us think through attach rate on kind of new logo deals versus kind of current customer add-ons? Maybe just part B of that question would be, how significant is that in the overall ASP uplift that we're seeing?

Yes. I don't think we have a clear answer to either of those questions. The product is brand new. We've sold it only to some existing customers, and we are currently signing up more-and-more publishers to participate in that product. I think the next quarter or 2, we'll start to get better visibility on an attach rate.

To your second question, there has been some industry chatter recently about what type of uplift on ARR SaaS or AI? Well, vertical SaaS companies expect by adding AI to their vertical SaaS solution. One of those studies suggests the uplift opportunity is about 50% of the ARR. As a reminder, our kind of AG business is about $11-point-something million in ARR, so that could be a material uplift. However, we have a long way to go before we really understand what the real rate is going to be.

Bill or Josh, feel free to add to that.

Yes. I'll just say that it was not a big contributor to the ASP increase for this quarter. That was more the larger new logo deals that Roy talked about earlier in the call.

I'm wondering also maybe you could help us think through some of the overall product strategy shift in B2C. Maybe how are you planning to actually increase the attach rate and the net churn as well? That would be helpful.

Go ahead, Josh.

I think from the product perspective, how we got to success was pretty critical on every single aspect from sign-up to completion of using, say, Scite search or Scite assistant. I think we lost a little bit of that and slowed us down in kind of the velocity, and so we're back on pace and I think rigorously looking at every single metric from sign-up to conversion and rigorously testing, right? A lot of testing to optimize and make sure that we're competitive with that. I think that's maybe a little bit of a reflection of joining a company where it takes some time to kind of get into that fit or swing. I think now the velocity has really hit. I think product is starting to catch back up.

I think just to comment on that further, we have obviously much more competition today than we did a year ago, and that is certainly impacting the conversion rate on the product. I think Josh and the team are doing a lot of really great work now in terms of rolling out product improvements that we think will materially move that conversion rate back up to where we'd like it to be.

Churn on that product has actually been improving. Churn is going up, lifetime value is going up. The issue we're having is new subscriber sign-up conversion from trial, which is not where we'd like it to be and not -- it's below where it was last year. I think that's partly product and partly messaging and being clear about the distinct advantages we have by having access to content behind the paywall.

I'm wondering if -- maybe you could help us think how long is the trial period on that product? Are you seeing that maybe students sign up for just the trial and then cancel it basically to get them through one paper or something like that? Is that kind of what you're seeing?

Yes. churn is improving. In other words, we have materially less churn this year than a year ago. They're not signing up and they're not signing up for 7 days and leaving. Lifetime value is going up. However, in the spring, people cancel because they're going to be out of school for summer and then they resign up in the fall, and so the re-sign-up process or attracting new people in the fall is where we go through a trial, which Josh has mentioned is 7 days, and then we convert that to subscribers, and our conversion rate is not where we'd like it to be.

Sorry, Josh, go ahead.

Yes. I was just going to say, it is 7 days. In many cases, the product has improved significantly in terms of the output you get out of it where it was, you can get a few paragraphs, you can get now a 15-page fully referenced report. To your point, sometimes the success of the product means they're solving their problems quicker, and so we think about all these different aspects, not just conversion rate, but how many queries are they doing, what is the weekly active users, monthly active users. I think, again, we need to be really disciplined in looking at those numbers across the board and then optimizing kind of the product decisions around what gets deployed.

Does that increase the conversion rate? What does that do to the lifetime value of the customer and then also the usage of various KPIs. We are, I would say, in a much better position now. We've also started to really leverage a lot of AI in our own development work, which has greatly accelerated the ability to deploy new features and a lot of UI/UX changes that are necessary for testing. You can easily run a test on this button moving here versus there and what does that do to conversion rates and things like that.

Our next question comes from Richard Baldry with ROTH Capital.

Hoping you could drill a little deeper into the, call it, non-typical or non-seasonal strength you saw in ARR in the first quarter that looked up versus prior years, and it was multiples of what you've seen before. Can you sort of break down where you think that's coming from? Was there any sort of pull forwards that we have to be cautious about looking forward? Or is there something sustainably higher going on there?

Yes, there was no pull forwards. I would say, over the last year, we have upgraded well over half of our sales teams. We've spent well into the 6 figures on training an entirely new sales process where we work with the customer closely to understand the problem we're solving and assess the value of solving that problem for the customer and then pricing accordingly.

I think just having a much more disciplined and focused sales process in addition to marketing is doing a great job driving top of the funnel leads into that sales team resulted in the ARR that we posted.

Bill or Josh, you're welcome to add to that.

I'll just add that the -- some of it also was again, some of the larger deals, but I think that's not a onetime thing. Again, we started talking about it last quarter and then obviously some this quarter. The sales team is focused, and I know we have other deals in the pipeline now that are kind of some similar size to some of what we saw in Q1. That definitely helps the situation when you can land 1 to 2 to 3 of those in a quarter. I think that's something that we have a chance to do each quarter going forward as well.

Switching to the expense side. The G&A is the lowest in almost 2 years, I think. Again, anything onetime oriented there or sort of out of pattern that would reverse itself? Or is this sort of a sustainable level? How do we think about that forward?

Yes. As I said, we had some concerted effort to keep the cost down. I think in some prior quarters, we may have had some legal and stuff. We did have an executive departure at the end of last quarter, so some of that reduction there is that executive no longer being in the business, and we were sort of able to kind of replace that with some resources that are just more efficient from a cost perspective.

I do think it's decent sort of to think about it as a run rate with maybe a little bit of exception here and there as being a little bit low, barring some kind of something that drives legal expense or a onetime recruiting item of some sort, I think we can modestly increase that through the year.

Then last for me, sort of switching gears to AI internally as opposed to talking externally. This seems like the pace of new offerings from the companies is increasing. How much is AI enabling sort of either efficiency gains or productivity gains? How much more do you think you can do with that? We're sort of hearing that a lot of companies are really embracing it internally, not just externally now.

Go ahead.

Yes. I would say on the internal side, we've made some changes. They're not fully kind of deployed across all of the team, but a lot of that is the AI to greatly speed up development, and it's pretty inspiring to watch. So a lot of this copilot and different AI coding tools are now part of a workflow from senior developers to junior developers. We've clearly put in guardrails in place and rules to use this AI, and so it's secure and it's largely done around light UI/UX changes, so not large features.

It's hard to quantify it, but it is dramatic. I think that allows us to shift a lot of things that are important to the business that might seem superficial, but matter a lot, right? Again, that is the workflow of getting that PDF in a second, making sure you're AI compliant using that PDF, asking a question and getting an answer from the literature, all those things built on that foundation of relationships and data that we have needs to be done seamlessly. I think the AI tools that we're now using primarily in development are greatly accelerating the pace. I think we'll continue to see that pace as it rolls out to more teams and more people on the development teams.

Yes. The only thing I would add to that is I think in the next 1 to 2 quarters, we'll be implementing some AI on the support side of the business to improve that. I don't think the development AI impact that Josh talked about or the support team impact is going to result in cost structure reduction. If anything, what it's going to do is free up software engineers to work on the more important stuff or it's going to free up support people to do a better job covering our existing support workload unless we see an unusual percentage of our account base start to use the AI chat tools to solve their issues as opposed to sending in a ticket to us, if that makes sense.

[Operator Instructions]. Our next question comes from Derek Greenberg with Maxim Group.

I wanted to touch on just the transaction segment. You guys had called out already that, that was largely due to 3 customers churning and largely first half will be impacted. I was wondering if you had any visibility into the second half yet? If maybe we'll see potential release due to just lapping when the initial declines had happened.

Yes. Just to be clear, it's 3 customers, 1 churn, the other 2 are simply buying less year-over-year, so they didn't churn. It's just reduced spend on their side. In terms of visibility in the second half, certainly don't think I do. Bill, do you have any comments on that?

We really don't. I think part of the reason we said that we think the second half will be better is we started experiencing the sharp declines in January of last year, and it really sort of accelerated in February. We are seeing a little bit of stabilization. As I mentioned, the decline this quarter was a little bit less than last quarter. We're seeing some things that we're seeing growth in our academic business, and we're obviously adding more platform customers.

When you look at that, that's more hunch than anything at this point that we'll start to see improvement. I'm not saying it will necessarily be growing in Q3 and 4, but hopefully, we're seeing a reduction in that decline just given some of those factors.

Then in terms of second quarter, I was wondering if you saw any impact from the government shutdown regards any of your end markets, your customers?

No, we have not seen material impact in government. Well, in government, corporate or academic.

Then I was wondering if you could just give an update too. I know on the last call, you kind of introduced this concept of a headless strategy plugging directly into customers' workflows. I was wondering if you had any updates there, that would be great.

Really no updates other than we continue to make product changes to be where the [indiscernible] is going. We continue to support many large customers with that strategy today. I would say, I don't know what the percentage is, but a material part of our pipeline is headless work because more of our larger corporate clients are frankly building their own internal LLM or tool set. So what they're looking to do is connect us into the parts of the workflow where they need specific problems solved, whether it's AI rights, document rights, citation information or something else.

Then just my last question was just on M&A. You guys have previously said you were expecting at least one acquisition this year. I was just wondering how the pipeline is looking, how the market is looking, if there are any updates there?

Yes, active pipeline, good discussions. I don't think we have something that will close by the end of the year, but we have a lot of things that are pretty close.

Thank you. This does conclude today's question-and-answer session. I will now turn the call back over to Roy for any additional or closing remarks.

Well, thanks, everyone, for joining us on our call today. Bill and I will participate in the Southwest IDEAS Conference on November 20. Qualified investors interested in participating should contact Three Part Advisors. We look forward to speaking with you in February to discuss our second quarter fiscal 2026 results. Have a great day.

This does conclude today's program. Thank you for your participation. You may disconnect at any time.

[Operator Instructions]

We'll take our first question from Jacob Stephan with Lake Street.

Maybe just first, wondering if you could touch on the nice sequential uptick in ASP. Maybe help us kind of think through some of the drivers of this? Was it more cross-sell, upsells or kind of larger new deal activity?

Bill, do you want to take that one?

Yes, sure. No, we are -- I mean, part of what we've seen with the onboarding of the new CRO and some of the sales training that we've been doing is that we are actually getting larger deals. And so I think Roy mentioned we had -- we've got announced just recently a couple of hundred thousand dollar deals in. And these are in the past few months, we've seen some of the larger deals in our company's history. I'll also say there was sort of a period where we had some churn from Resolute in the past, which was traditionally more of a larger deal where that basically caused a decline in our ASP.

And so that has kind of leaned off and now we're at a place where we can sort of build back ASP. And I think it will be a focus as we do additional sales training, bring on some better salespeople over time and again, continue to see some larger deals. Also just as it relates to some of the API type deals that Josh was talking about.

Okay. Got it. And you kind of ask one question further on Resolute. Obviously, you noted some churn issues kind of starting off there. But how are you using the product? How do you see the Resolute software adapting to your new strategy of being the API provider for LLMs?

Well, Resolute has always had a strong API and has not necessarily had a strong UI in their software. So Resolute works much better in this headless strategy than it works as a product unless we go in and rewrite big parts of the product. which we have not wanted to do. So we haven't talked about Resolute in a number of quarters because it's a product we don't focus on. We focus on a heavy investment in site and heavy investment in Article Galaxy, which, of course, are driving all of our growth. However, as we develop this headless strategy we talked about, being able to plug in the 13 additional databases via API into the workflow kind of resurrected that product in terms of selling that data to customers.

And Josh, you may have a few other comments on that. Go ahead, if you do

Yes. I'd really just emphasize that there are these 13 highly curated databases kind of coming to us for an API to get and access to the article, to get clinical trials, to get research articles, to get news articles, all these different things is a big value add for customers. And so I'm personally excited because it's kind of been right there in front of us for a while, and it's very easy to execute on. The one thing I would also say about the API-first deals is that by embedding ourselves into the infrastructure of some of these large companies, I think those contracts become very sticky. And so I'm personally quite excited by the Resolute databases really coming back to life as a focus for us.

Okay. Maybe just one last one, more on the kind of competitive environment in this headless strategy. Are you aware of anybody else that's kind of doing -- running the same API strategy to kind of plug in with the larger LLMs?

Do you want to take that, Josh?

Yes. I think I think what we're starting to see in the ecosystem is some publishers doing this. And so if you look at why we, I think the third largest publisher, they are directly opening up their articles or segments of their articles to LLM providers such as Anthropic. On their recent earnings call, they talked about leaning more into AI and specifically AI licensing deals. And so I think we're going to start to see this across the ecosystem from publishers themselves. I think publishers will have somewhat of a challenge becoming a pan-publisher source for this, largely because competitors don't want to give their content to other competitors.

And so this is what we're talking about when we say we're pretty uniquely positioned is that we work with virtually all publishers. We're already driving them revenue. And this is really, as Roy and I have said in the past, kind of a shift from [indiscernible] . And so I increasingly see these bits of articles or chunks of articles and specifically the data from articles being something that's valuable that integrates directly into tools, whether that's a hyperscaler or whether that's an internally licensed LLM at a large corporate or even an academic institution.

So I think it's an exciting time, and I think there's a lot of people kind of looking at this and trying to say, how do we bridge this gap between research articles and AI.

We'll take our next question from Richard Baldry with ROTH Capital.

Same question I asked last quarter, the COGS line was actually slightly down on the platform side, while revenues were up pretty good. Can you talk again about sort of the trends there, whether this is sort of getting the peak optimization, I think about it that way? Or is there further cost improvements that can come on the platform side even as the top line is scaling?

Bill, do you want to take that?

Yes, sure. Yes, some of this is effectively using our cash. I mean really where this is coming from is we sort of stabilized the labor base there that grows kind of just was like not a lot of additional headcount, but just cost of living increases, things like that. But we've really tried where we could to lower or limit the increase in the hosting cost. And some of what we've been able to do is take the cash flow that we've had and apply that to some prepayments where we prepay some of our space with Amazon Web Services and other providers. And as a result, we're actually getting it cheaper over time by prepaying. So I'm not sure how much we can do that going forward to sort of see it decrease, but I think we can do that to the extent that it will increase less than at the pace that we're growing the revenue, which again is why I think you're seeing some very high numbers on the gross margin side for platforms. We're also seeing in certain areas, AI becoming cheaper, so as we grow, some of the AI providers we use get cheaper over time. And so that's impacting the number as well.

Okay. Then on the AI-related deals being 4x the growth rate of non-AI, do you think that can continue at this pace? Is there sort of eventually the scale of that base gets big enough that it can't keep up at that sort of delta? How do we think about that headed into the next sort of year or 2 as a driver?

Yes. I think we expect to see similar results in the B2B space. In the B2C space, we don't expect it to grow as much as it did in FY '25 simply because the base is getting bigger and it is getting more competitive. But Josh or Bill, I'd invite you to add any comments you might have.

Yes. I mean I would just again emphasize I think with this headless strategy, this is internal tools or internal companies using internal AI and this allows us to price based on like the usage of this, right, the calls that they're making to our API. And so what we're seeing is as these tools ramp up, it's less looking at here's a 100-person seat license versus here's a company-wide integration into a tool that they're heavily training on. And so I think that will command larger check sizes at B2B. And I think as we started to prove that out, those will continue to grow because we're going into places that companies are already investing a lot of money into.

Great. And last for me would be, can we dig a little deeper into the strength in the deals above $100,000? So are you going after a different type of customer? Or are you going after a different value prop? Are they larger deals per customer? And how are you achieving that on sort of a similar customer base? Or is it different verticals? How are you getting sort of larger deals out of what presumably is a similar customer set?

Yes, there's a few moving parts. One is the new sales process and the new CRO has brought in a number of new people who are not kind of preprogrammed with an expectation of what we should sell a product at. And a big part of the new sales process is spending time qualifying the customer and understanding what their pain points are, what value we can use to address those pain points and what the economic impact to them will be if we do.

And then the products are being priced accordingly. So I think part of it is -- and I think it's probably a big part of it is sales execution and the way we're selling now. Secondarily, we did wholesale change the pricing on the academic segment of the business, not much of that is reflected in FY '25. But what we did do in '25 is we experimented with different pricing points. In other words, when we acquired site, they had a fairly set pricing model for libraries. We sold at that price point. We sold at price points way above that price point, and we kind of played around with pricing in FY '25 until we figured out a new model that we recently implemented.

So some of it is just our standard pricing has changed. And I guess that would be the 2 main drivers that I can think of. Bill, is there any more that you can comment on?

Not too much. I do think it's a sales execution thing. And really, before we frame a proposal to a customer, really trying to understand their pain points and how much value the product is going to deliver for them and then pricing that value accordingly.

[Operator Instructions]

We'll take our next question from Derek Greenberg with Maxim Group.

The first question I have is just on a recent partnership you guys announced with LibKey and the integration there. I was wondering if you could just talk a little bit more about this partnership and the opportunity there.

Yes. As that address -- I'll jump in and Josh, you can add some comments. But basically, in the academic segment, LibKey is a big player in the library, providing a product that's called a Link Resolver. And Link Resolver, what it basically does is when you do a search and you get an answer to your search in terms of a scientific article, it kind of resolves where you go to get to the link to obtain that article.

And they've been doing that for a number of years, private company successful. And we also work with, frankly, 3 other link resolver companies that we worked with for a number of years. And so putting together the Third Iron deal, Third Iron is the company that owns and -- I'm sorry, the LibKey product. We've run a number of webinars in conjunction with them, which introduce us into their libraries. And keep in mind, academics is new to us. I don't think we have more than 200 academic customers. There are 10,000-plus libraries out there that we can sell into.

So we view partnering with Third Iron around LibKey as an opportunity to expand our academic business as well as kind of revisiting the partnerships we have with some other providers that provide a product like LibKey to expand into their academic library business. Josh, anything you want to add?

I don't have too much to add except to say that we look at a variety of different services that Roy mentioned to get our users access, whether that's subscription-based access that they have from their university or whether they're an individual at a university, access to the content as quickly as possible. And so there's really kind of like a hierarchy of needs and looking at how can we make sure we're facilitating access for the end user in the most robust and kind of efficient way possible. And I think leveraging our partnership with LibKey is one piece of that.

Okay. Got it. Turning to the cross-sell between Site and Article Galaxy. I was wondering if you have any statistics you're willing to provide in terms of what percent of Article Galaxy customers are also customers of Site. I recall previously, you said this was single digits and you were looking to get to double digits. I was just wondering how things are progressing on that side.

Yes. We have not disclosed that number. I can tell you that -- and Bill, correct me if I'm wrong, a vast majority of the site sales in FY '25 are to what we call a new, new customer. In other words, we're not doing business with them on the Article Galaxy side. We do some cross-sells and a lot of times, those cross-sells are pretty big from an ARR perspective. But I think if you look at it from a logo perspective, vast majority of the logos are new, new customers. Bill, anything to correct that?

Yes. I would still describe it, excuse me, as low to mid-single-digit penetration on the Article Galaxy customer base.

Okay. That's helpful. My last question is just on margins. We saw some really good improvement this year. EBITDA margins growing 5%, doubling year-over-year. I was wondering, looking towards '26, how we see expansion relative to this year in margins and how you expect, I guess, operating expenses to grow compared to revenue?

Bill?

Yes. I think part of the question for us is how much do we invest back into sales and marketing and tech and product development. As I said, we're trying to basically try to keep investing in the sort of those 2 top lines on our expense base, sales and marketing, tech product development while cutting sales -- excuse me, cutting G&A, things like stock comp where we can. But I will say -- so in other words, I think we'll definitely cross the 10% margin threshold for the year. We want to stay above that. I think next year, we can be above where we are today, but we may temper that a bit. In other words, I think we could run 15 plus, but I don't think we're going to do that. I think we'll invest back into it.

And so we'll kind of be somewhere in between that kind of 10% to 15% range, and that's where I expect we'll kind of end up from an EBITDA margin. I think gross margin will continue to expand. That will be 50%-plus for the year next year. And expense base, tough to say. I mean, again, I think it could -- we'll kind of pull levers where we need to pull levers. But again, could be 10% growth on the sort of SG&A type bucket. But again, I think I'll have more update on the Q1 call as we see our Q1 results come in and as we sort of further define and chart out how we're going to manage expenses and invest in growth for the rest of the year.

I do think transactions are a key element of this. And on our own internal models, as I said, we're modeling those down at least for the first half of the year. And so if you are sort of building models and such, I would do similar from that standpoint until we start to see that turn the other way. But given that, I still think we'll be kind of at the levels I talked about as we look ahead to '26.

I did get one question via e-mail. Can we explain the strategy to stem the decline and resume growth in the transactions business? To address that, what I would say is the current thinking is product improvement to improve conversion percentages. And I think part B of that is understanding what's driving the change. In other words, we're seeing a significant year-over-year increase in monthly average users and weekly average users which is great. But what we're seeing is a big increase in them acquiring free documents and not paying for documents. As Josh mentioned, we recently did a survey that suggested some of our customers, around 10%-ish of our customers are buying less documents because they can get a good enough answer from AI.

So our current thinking is to improve -- we have a massive amount of traffic in site, and we have a massive amount of traffic in Article Galaxy. And so our current thinking is to work to make -- to improve the conversion rate to also take advantage of the opportunity, you just bought this article, here's 3 other articles like it. You just bought this article, here's 5 articles that have a supporting statement in them related to the one you acquired or have a contrasting statement in them related to the one that you acquired. Do you want to buy these?

So it's really -- I use the comment internally, we want to be the Amazon of Docdel. We want to make it super easy. It's not as easy as it could be today. We want to suggest it sell. We don't really do that today at all and do some other things. As I mentioned, we already took action on one barrier and saw a pretty nice improvement, which if it were to continue for all 52 weeks because we look at weekly data, would be a high 6-figure improvement in revenue to that business. And as you know, that's a pretty EBITDA profitable business for us.

So we've got a number of things in the works, but strategically, we focus on SaaS revenue and AI, but we do have a fairly large around 60 people that work on the Docdel business. The leader in that business now is a guy who's very technologically savvy, and he's gone through every internal process, every customer process that we have with the intent of how do we make this more seamless and more suggestive to drive more sales in that business. Back to you, operator.

And there are no further questions on the phone line at this time. So I'll turn the program back to you, Roy, for any additional or closing remarks.

Okay. Well, thanks, everybody, for your time, and I look forward to connecting in November to discuss our first quarter fiscal 2026 results. Have a great day.

This does conclude the Research Solutions fiscal and operating results for its fiscal fourth quarter and full year ended June 30, 2025. Thank you for your participation, and you may disconnect at this time.