NRDY - NYSE

Good afternoon. Thank you for attending Nerdy, Inc.'s Second Quarter 2025 Earnings Call. My name is Megan, and I'll be your moderator for today. [Operator Instructions] I would now like to pass the conference over to your host, T.J. Lynn, Associate General Counsel of Nerdy. You may proceed.

Good afternoon, and thank you for joining us for Nerdy's second quarter 2025 earnings call. With me are Chuck Cohn, Founder, Chairman and Chief Executive Officer of Nerdy; and Jason Pello, Chief Financial Officer. Before I turn the call over to Chuck, I'll remind everyone that this discussion will contain forward-looking statements, including, but not limited to, expectations with respect to Nerdy's future financial and operating results, strategy, opportunities, plans and outlook. These forward-looking statements involve significant risks and uncertainties that could cause actual results to differ materially from expected results. Any forward-looking statements are made as of today's date, and Nerdy does not undertake or accept any obligation to publicly release any updates or revisions to any forward-looking statements to reflect any change in expectations or any change in events, conditions or circumstances on which any such statement is based. Please refer to the disclaimers in today's shareholder letter announcing Nerdy's second quarter results and the company's filings with the SEC for a discussion of the risks. Not all of the financial measures that we will discuss today are prepared in accordance with GAAP. Please refer to today's shareholder letter for reconciliations of these non-GAAP measures. With that, let me turn the call over to Chuck.

Thanks, T.J., and thank you to everyone for joining today's call. For years, we pursued a vision where AI supercharges the human elements of learning, giving tutors, students and teachers superpowers to teach, learn and achieve at levels beyond what humans or AI could do on their own. We made meaningful progress on our Live+AI vision over the past quarter as we work to embed real-time intelligence in each stage of the learning journey, including before, during and after tutoring sessions. As we head into the back-to-school season, these enhancements are fueling momentum with stronger engagement and retention in new cohorts that position us for sustained growth this coming school year and beyond. In the second quarter, Learning Membership revenue returned to year-over-year growth for the first time since Q2 2024, driven by higher frequency and more engaged customers, a series of AI-powered product enhancements and continued improvements in our AI Learner-Expert matching system. Varsity Tutors for Schools bookings grew 21% year-over-year, bolstered by demand for our high dosage tutoring offerings and a positive initial response from school district customers to our Live+AI platform. Our second quarter results reflect improvements to new customer cohorts, combined with operational improvements that are translating into financial progress. Total revenue was $45.3 million, down 11% year-over-year. That was expected and in line with the past guidance we have shared and driven primarily by a non-recurring state-funded program and lower institutional revenue. Excluding these factors, the underlying momentum is becoming visible as we approach the back-to-school season. Consumer Learning Membership revenue grew 4% year-over-year to $37.8 million, representing 84% of total revenue. Active Members stood at [ 30,600 ] as of June 30, with average revenue per member per month or ARPM reaching $348, a 24% increase year-over-year. The ARPM increase is a result of our focus on driving deeper levels of customer engagement through higher frequency Learning Memberships, combined with price increases implemented earlier in the year. Higher frequency Learning Memberships, specifically those with 4 or 8 hours of tutoring per month lend themselves to consistent weekly tutoring habits of once or twice a week, respectively. We're finding that those frequency levels are resonating with customers and leading to a more engaged overall customer base with stronger retention dynamics. Looking ahead, we continue to expect consumer revenue to accelerate in growth each quarter throughout 2025. Gross margin sequentially expanded 350 basis points from the first quarter. We continue to expect to deliver sequential improvements to gross margin each quarter as we move throughout 2025. Our non-GAAP adjusted EBITDA loss of $2.7 million was better than the top end of our expectations. We believe these results, together with the positive inflection in annualized run rate Learning Membership revenue of $127.6 million, which was up 7% year- over-year as of June 30th, keep us on track to deliver profitability in the fourth quarter of this year. During the quarter, we advanced our efforts to make the AI-powered parts of our customer experience more context aware and interactive. We focused on driving operational efficiency and customer acquisition and support functions, and we advanced initiatives focused on continuing to improve customer retention. We continue to enhance the video playback and AI session summary that are sent to customers after each tutoring session to take into account the context of prior tutoring sessions, creating a continuous narrative of student progress that helps families track improvement over time. During the quarter, we also made the session summaries in the UX more interactive and insightful and surfaced more data information that can provide insights that are useful in understanding the learning and interactions happening during the tutoring sessions. This improved experience is driving high levels of engagement and interaction from customers, and it's helping to reinforce the value of the product itself to the customers after each tutoring session. Another product enhancement that we made during the quarter was introducing AI-generated explanations for the AI-powered Learner-Expert matches. This draws on the customer intake data to highlight exactly why a tutor was a great fit and the rationale behind the AI-powered match. The customer feedback has been really positive, which is pulling through to better onboarding and activation metrics, including higher first week session attendance and lower cancellation rates for new customers. We also improved a few other parts of the customer journey that are driving improvements across activation and onboarding metrics. Automated calendar invites for sessions delivered strong activation gains, and it also resulted in improved first week attendance. We've also streamlined the checkout process by incorporating modern payment options that are more mobile-friendly and saw improved checkout rates as a result of those changes. Additionally, we'll be entering this back-to-school period having recently overhauled legacy sections of our website oriented around practice resources. These have historically represented more than 80% of our total traffic. These are now evolving into modern AI- powered interactive learning experiences covering many hundreds of subjects and tens of thousands of academic concepts and users can engage with them with diagnostics, flash cards, worksheets, AI tutoring, video modules and much more. These resources can be highly additive to learning a subject or studying for a standardized test, and we're excited about our ability to drive deeper levels of engagement around a particular subject the student is learning. We anticipate these modern AI-powered practice resources will enhance top-of-funnel awareness, drive deeper levels of engagement among existing Learners and be a value to both consumer students as well as institutional school district customers. Recent advances in AI provide us the opportunity to drive the identification of key processes that will allow us to improve both the customer experience and operational consistency, while further reducing the operating costs necessary to scale our business. Heading into the back-to-school season, we're accelerating our Live+AI approach, combining live instruction with AI to deliver superior outcomes aimed at each primary stakeholder group. For tutors, we're launching a desktop application called Nerdy that serves as a sidekick that leverages real-time computer vision and audio capabilities to be fully context aware in that moment and provides real-time guidance to the tutor on a heads-up translucent display. Using this tool during live tutoring sessions allows them to always have the right answer, spend less time checking students work, get real-time auto suggested ways to better explain concepts to students, and it even reminds them when they haven't checked for understanding yet, while teaching a portion of the lesson. The Nerdy app helps reinforce best practices that are key to driving strong student outcomes and overall customer satisfaction, and it does so in real time during the tutoring session itself. Tutors will also have access to our Tutor Copilot, a chat-based tool, which also gives them access to the latest LLM-powered capabilities and enables them to perform common functions like real-time lesson plan generation that leverages past tutoring session context as well as other common tasks that can improve the session experience. These tools are aimed at making it easier for all tutors to be highly effective with the goal of increasing customer retention through improved session quality and adherence to best practices known to drive customer retention and lifetime value. We also think that these tools can make it much easier to be a tutor on the platform and drive retention among tutors themselves. The Nerdy app is also student-facing. The tool enables students to transcribe and receive summaries of their class discussions in school. And when it hears or sees a complex problem, it can generate step-by-step explanations to help that student better learn in real- time and have a more delightful and less frustrating learning experience. Based upon that context and the other learning interactions, 24/7 AI study agents can then act on the student's behalf and take identified actions oriented around learning, test preparation and practice to save them time and help them better learn and succeed academically. As these tools gain more context and data around student learning interactions, we'll be able to provide them even higher levels of personalization. Lastly, the Nerdy app also has direct relevance for schools, including for both students and teachers. Nerdy can help new teachers feel more confident and be better prepared, bolstered by time-tested best practices and real-time content suggestions, including concept explanations and more, that can help ensure they start a new school year feeling confident as they embark on their new career in teaching, or if they're just teaching a new subject, in that new subject. For English as a Second Language teachers, Nerdy can assist with real-time [ foreign ] language translation and discussion summaries, breaking down barriers and ensuring students get the support that they need. For substitute teachers, no more showing up and having no knowledge of the subject or class. Nerdy can elevate the ability of a substitute teacher to deliver expert instruction in areas where they may not be an expert. That can help ensure students make continued academic progress even if the teacher isn't there for a day. For teachers and school districts, we are launching more than 30 new AI tools this back-to-school season, that are available in our Live +AI platform, that are aimed at helping educators streamline classroom and administrative tasks and better personalize learning for students. These AI tools enable educators to save time by automating tasks like generating individualized education programs or IEPs for students, creating practice problems, quizzes and tests, reviewing and evaluating essays, creating student assignments against rubrics, generating rubrics, translating documents and much, much more. Our teacher tools will free overburdened educators, allowing them to focus more on high-impact instruction. We also believe that these AI tools help elevate our positioning as the next-generation AI- enhanced tutoring and intervention platform for school districts. One thing that we're excited about is introducing agentic practice problems. Agents automatically generate personalized practice content for Learners based on the Learner's most recent tutoring session and other learning interactions on the platform. Rather than a tutor having to assign and create all the content for practice problems so that a student can learn in between tutoring sessions, now our platform does that automatically, and it's hyper-personalized. These various learning modalities ranging from live tutoring in a hundreds of live classes available every week on our platform, combined with the AI tools I mentioned to create something very powerful. This past semester, we partnered with one of the 10 largest school districts in the United States to provide high-dosage tutoring programs in English and Math. The way the school district customer utilized the full capabilities of the platform is indicative of how we see our AI tools deepening the effectiveness of the live instruction. The school district used the Live Learning Platform for live video-based intensive tutoring with 5 students per group and a recurring subject matter expert tutor. The district leveraged our AI-driven student cohort grouping that pairs students of like needs together and utilized our customized AI lesson plans that align state standards. Tutors had access to a chat-based AI copilot tool and AI lesson plan generators that enabled them to be better prepared for sessions and better meet district expectations. The school district administrators received a tutoring session, video playback and AI session summary that included key insights after each tutoring session for each group. That was a product that we enhanced pretty significantly over the course of the semester on a rolling basis. And in general and for this particular customer, saw them engage more and more and we enhanced the product and the insights they could garner. More than 700 students participated in the intensive high-dosage tutoring program, and those participants saw an average 11 point or 5 percentage points improvement from midyear to end of year. That was more than double the gains from the prior control period with all schools and all grades seeing their performance improve, the largest gains occurred in the lowest performing elementary schools and grades. Importantly, the school district was thrilled with these results. In another high-dosage tutoring program in that same school district, students in an elementary school showed gains in their early literacy, reading and their overall scores on the state assessment of 10% to 26%, a huge improvement. These results were remarkable, and they demonstrate our ability to help students across multiple subjects, grades and school buildings, and they speak to what's possible when our AI capabilities for personalization are fully leveraged in concert with live tutoring. In closing, by combining human expertise with Live+AI, we're enabling personalized learning at scale. We're significantly adding to and strengthening these Live+AI capabilities this back-to-school season and believe it can strengthen relationships across every audience we serve. Thank you for your continued support and confidence in our vision. With that, I'll turn the call over to Jason to discuss the financials in more detail. Jason?

Thanks, Chuck, and good afternoon, everyone. As Chuck mentioned, Nerdy is delivering against our commitments with a return to Learning Membership revenue and Varsity Tutors for Schools bookings growth in the second quarter, positioning the company to deliver accelerating sequential revenue growth and gross margin expansion each quarter throughout 2025. Second quarter revenue was in line with expectations, delivering revenue of $45.3 million, within our guidance range of $45 million to $48 million, which represented a decrease of 11% year-over-year from $51 million during the same period in 2024. Consistent with expectations, revenue declined when compared to the prior year period, primarily due to lower institutional revenue in a specific $3 million state-funded consumer revenue program in Q2 2024 that did not recur this year. These impacts were partially offset by higher ARPM in our consumer business as a result of a mix shift to higher frequency Learning Memberships and price increases enacted during the first quarter of 2025. These changes are coupled with higher retention in newer cohorts due primarily to improvements in user experience and new expert incentives. We continue to expect consolidated revenue growth will accelerate sequentially each quarter throughout 2025. Consumer Learning Membership revenue returned to growth with second quarter Learning Membership revenue that was up 4% year- over-year. Revenue recognized in the second quarter from Learning Memberships was $37.8 million and represented 84% of total company revenue. As of June 30th, there were [ 30,600 ] active members and ARPM was $348, which represented a 24% increase year-over-year. We expect Learning Membership revenue growth will accelerate sequentially each quarter throughout 2025 as the mix shift towards higher frequency and higher-priced Learning Memberships is combined with the improvements in the user experience and improved retention in newer cohorts. Our Institutional business delivered revenue of $7.3 million and represented 16% of total company revenue during the second quarter. The integration of Live+AI capabilities into our institutional offerings is resonating strongly with district leaders who need data-driven insights to optimize learning outcomes and determine resource allocation. Varsity Tutors for Schools executed 50 contracts, yielding $4.9 million of bookings, an increase of 21% year-over-year. Our strategy to introduce school districts to the platform and ultimately convert them to our fee-based offerings continues to produce results by delivering 36% of paid contracts and 50% of total bookings value in the second quarter. During the second quarter, gross margin increased 350 basis points sequentially from the first quarter as a result of price increases for new customers enacted during the first quarter of 2025. As mentioned on the earnings call in May, the year-over-year decrease in gross margin was primarily due to investments in our partnership with experts through increased incentives. Following the adoption of new expert incentives, we're seeing faster times to first session, more sessions in the first 30 days, lower tutor replacement rates and higher retention, all of which should continue to strengthen our business over the long-term. As we move throughout 2025, we expect to deliver sequential quarterly gross margin improvements. Sales and marketing expenses for the quarter on a GAAP basis were $13.5 million, a decrease of $2 million from $15.5 million in the same period last year. These decreases in sales and marketing expenses were driven by consumer marketing efficiency gains, coupled with the moderation of our investments in the Institutional business given near-term funding uncertainties. We continue to believe a significant opportunity exists in the institutional space and that the product enhancements we are making to the Live+AI platform will drive growth in future periods. General and administrative expenses for the quarter on a GAAP basis were $26.6 million, a decrease of $6.6 million from $33.2 million in the same period last year. Included in G&A costs were product development costs of $10.7 million. AI-enabled productivity improvements, coupled with new software-driven processes and systems implementations, headcount reductions and other cost reduction efforts have enabled us to generate operating efficiencies and remove significant costs from the business. As Chuck mentioned, recent advances in AI provide us with the opportunity to drive further levels of productivity, including the identification of key processes that will allow us to improve both the consumer experience and operational consistency, while further reducing the operating costs necessary to scale our business. Non-GAAP adjusted EBITDA loss of $2.7 million for the 3 months ended June 30th beat our guidance of negative $3 million to negative $6 million, and compared to non-GAAP adjusted EBITDA loss of $2.1 million in the same period last year. Non-GAAP adjusted EBITDA outperformance relative to guidance was primarily driven by across-the-board P&L improvements via higher gross margin, coupled with lower marketing and G&A spend. We believe these results, together with a positive inflection in new acquired and active monthly recurring revenue in June, keep us firmly on the path to profitability in the fourth quarter. As of June 30th, the company's principal sources of liquidity were cash and cash equivalents of $36.7 million, and we have 0 debt. Today, we're introducing third quarter guidance and updating full year guidance. As a reminder, the third quarter is always seasonally our lowest revenue quarter as 2 out of the 3 months are during the summer when students and schools are out of session. We also have higher sales and marketing expenses ahead of the back-to-school season, resulting in the lowest adjusted EBITDA of any quarter during a typical year. For the third quarter, we expect consolidated and consumer revenues will return to being positive year-over-year due to higher levels of consumer customer acquisition, coupled with improvements in ARPM due to the mix shift towards higher frequency Learning Memberships, coupled with price increases discussed earlier. We also expect improvements to the user experience and investments in tutor pay rates will drive continued retention improvements. For the full year, we expect a return to growth in consolidated and consumer revenues as product innovation accelerates and operational improvement initiatives pull through, leading to accelerating consumer revenue growth each quarter throughout 2025. Institutional revenue reflects the flow through of lower 2024 bookings into the first half of the year, coupled with a cautious federal and state level funding environment. We expect Institutional revenue to return to year-over-year growth in the fourth quarter of 2025. For the third quarter of 2025, we expect revenue in the range of $37 million to $40 million. For the full year, we expect revenue in the range of $191 million to $197 million. Turning to adjusted EBITDA guidance. For the third quarter and full year, non-GAAP adjusted EBITDA improvements year-over- year reflect a return to consolidated and consumer revenue growth, coupled with benefits from AI-enabled productivity and operating leverage improvements, partially offset by investments in AI and expert incentives. We expect price increases for new consumer customers enacted earlier in the year will continue to yield sequential quarterly improvements to ARPM and gross margin. During the fourth quarter of 2025, we expect to be profitable on a non-GAAP adjusted EBITDA basis. For the third quarter of 2025, we expect a non-GAAP adjusted EBITDA loss in the range of $11 million to $13 million. For the full year, we expect a non-GAAP adjusted EBITDA loss in the range of $13 million to $17 million. As mentioned, as we move throughout the second half of the year, we expect to deliver sequential quarterly improvements in consolidated revenue growth rates and gross margin that we expect will culminate in becoming adjusted EBITDA and cash flow positive in the fourth quarter of 2025. This would result in us ending the year with no debt and cash in the range of $30 million to $35 million, which we believe provides us with ample liquidity to fund the business and pursue growth initiatives. In closing, thank you again for your time and for your continued interest in our company. With that, I'll turn it over to the operator for Q&A. Operator?

[Operator Instructions] Our first question will go to the line of Bryan Smilek with JPMorgan.

Can you just talk about the return to revenue growth, obviously, impressive to see. I mean, like what gives you confidence in the improving trends through the back half of the year, whether that be across product, pricing, tutor payment, investments and operational improvements? And I guess within that as well, too, can you just help us better understand the composition of active member versus ARPM growth?

Thanks, Bryan. Happy to. So I would say the last 60 days, in particular, were the single most productive period in our company history by a significant margin. So we've changed how we work and how we approach building AI natively and allow for us to rapidly increase the pace of innovation, and you're seeing that pull through to the products, many of which we announced today that we think are going to have a pretty tremendous impact on growth in the second half of the year. But what we're already seeing is that those improvements that we made to retention, to customer onboarding and activation that we spoke to in the last 2 calls, have continued to hold for new customer cohorts. And I'd say we're kind of ahead of where we expected to be with the new 30-day and sort of leading indicators of retention with a whole host of improvements in flight and shipping right now. So we feel really good about as we approach the beginning of the school year, how conversion is improving, retention is improving and the early indications of all the different waves of schools, also look like they're shaping up to help us have a great back-to-school in school year, but we're kind of looking at the underlying KPIs that are highly indicative of that revenue acceleration. And then separately, the -- on the Institutional side, we're seeing strong leading indicators related to the pipeline growth and interest from schools in the Live+AI platform, and feel really good about some of the kind of the extent to which the high-dosage tutoring product as well as our Live+AI offerings are resonating.

Bryan, this is Jason. Maybe just to add some numbers to that. Importantly, during the second quarter, we returned Learning Membership revenue to growth. It was up 4% year-over-year. That [ sends ] a 4-quarter decline. Also, institutional bookings were up 21% year-over-year in the second quarter, which positions us well as we move into the back half. As Chuck mentioned, we're seeing like continued improvement in new customer acquisition. As we move throughout the rest of the year, we expect ARPM to continue to increase based on, one, the mix shift towards the higher frequency Learning Memberships that are driving higher levels of engagement with our platform, coupled with improvements in Active Members. So from a cadence perspective, we would expect ARPM to move from $348 at the end of Q2, which was up 24% year-over-year to $380 at the end of Q3 and Q4. And then we would expect Active Members to be around 38,000 at the end of Q3, and we would end the year at 40,000 Active Members, which is up 7% year-over-year, returning to both Active Member growth and ARPM growth. And then maybe just a little bit more color on the Institutional business. It's shaping up to be the strongest quarter we've ever had from a bookings perspective in Q3, which gives us that continued confidence as we move into the back half of the year that you'll see sequential consolidated revenue growth acceleration every quarter in 2025, which is exciting to see.

Our next question will go to the line of Jason Tilchen with Canaccord Genuity.

Curious, you shared a ton of information regarding all these AI tools in the prepared remarks and the shareholder letter. Wondering if you could just sort of give a brief overview of what you're most excited about in terms of driving benefits over the next few quarters and then more so over the medium-term, what has the potential to be most impactful for the overall platform?

Of course, good question. So our first kind of foray into the application machine learning was probably like 2016, 2017, where we started using it to drive the student and learner matches. And as we got progressively better at that, we then saw it pull through to better retention out of the gate and ultimately, lifetime value. And of course, that lifetime value was high margin and highly accretive, and we kind of gained confidence in our ability to apply AI and ML to this vertically integrated marketplace business and get leverage out of it. And one of the other kind of similar yields that we got in sort of Gen AI phase that was really material was related to our AI session summaries and video playback where after tutoring sessions, we can allow for parents to not only get a recorded video sent to them and students as well, but also for them to get a summary and insight associated with that session. So we've made that progressively better over the course of the year. A couple of the things that we are rolling out that have been really resonating and I think will get -- drive just continued higher levels of retention and engagements in LTV extension are some of those improvements. So you're starting to see interactive graphics associated with the tutoring sessions themselves. You're starting to see agentic practice problems where we're able to take the transcription from the actual live tutoring session and then use that to generate automatically practice problems and other forms of content immediately after the session, thereby allowing us to extend the value that's provided on an ongoing basis between sessions. That's something that we're seeing terrific feedback on. And we're also then able to take into account the context, not just of the past tutoring session, but now of all of the tutoring sessions to-date. So whether it's 2 or 4 or 10 or 12 or however many there are, we're able to take that into account to have way more informed and thoughtful recommendations, and that required a bunch of work and infrastructure to put in place, our ability to query that context window and pipe around the data in the right way. But now you're starting to see the real benefits of us having all of that content. So I would expect for that to continue to drive retention and engagement and just customer delight for both consumer customers as well as school district customers who will have the ability to see that on a cohort level. But the thing I'm actually most excited about is what's shipping right now with our Nerdy app, which allows for us to influence what's happening in the session in real-time with computer vision and audio where it can see and hear what is occurring and then provide recommendations for both the tutor or for a teacher or for the students. And so, for a tutor, we'll actually be able to elevate the performance of every subject matter expert, allow for them to stretch into other areas, allow for them to always know the right answer, allow for them to remember everything the student said in the tutoring session thus far and ensure that they didn't miss anything, ensure they spend less time checking practice problems, more time and valuable instructional time. So I think we're going to like very significantly positively impact the session experience across millions of sessions per year, and that's something that I think could have a profound impact on the customer experience and ultimately on retention and lifetime value.

Our next question goes to the line of Yi Fu Lee with Cantor Fitzgerald.

Can I start with more of an industry high-level question. So Microsoft Elevate Academy, they announced a $4 million donation to enable Cloud and AI technologies to help communities in school. And yesterday, Google announced another $1 billion. We understand Nerdy pledged with the White House's Pledge in AI education in June as well. Can you expand like will there be any benefit for Nerdy to capitalize on these programs?

We do think that there's a big opportunity to help school districts. We think the White House's Pledge also puts a catalyst behind the need to adopt AI. And I think as we've shared in the past, there was sort of hesitation a couple of years ago related to school district administrators and embracing AI. There was kind of cautious optimism maybe a couple of quarters ago and some interest. And I think what we've seen is that, that dam has sort of broken and now there's broad interest. There's no longer fear of using AI. There's an understanding that not only is it permissible, but it's expected and even required. And in order to best serve your students and your families and your teachers that the school districts need to invest in some of these tools and capabilities so that they can do a better job because ultimately, like the goal isn't, of course, to invest in AI for AI's sake, it's because it allows for districts to solve problems that historically they couldn't and to be more efficient, more personalized and better serve their stakeholders. So what's really exciting is these schools are now to the point where they actually are going to have a material impact on the district and on personalization. We're launching more than 30 new AI tools for schools. Those just launched, and we think that they can really resonate with school districts. So that ranges from kind of dedicated purpose-built tools as well as universal copilots for both chat and computer vision that can really allow for whether it's a teacher solving a specific problem or really any problem to take advantage of modern AI, and then also taking into account all of the different contexts we have on the students, on their classroom and start getting smarter and smarter over time in terms of how we allow for them to better personalize learning, automate work and ultimately serve better interventions for specific students before they need them.

Got it. Thanks for the color on that. The next one is more philosophical in nature. I still remember like kids and teachers are so lucky to have all these AI tools at their disposal nowadays. I still remember when I was a student, I wasn't able to even use a calculator until, like, summer of 8th grade. We used human intelligence back then. So Chuck, do you think this will be the baseline going forward? Because -- I know you guys launched a lot of new tools like AI for Tutors, AI for Students, AI for Teachers and School Districts, right? So would students. teacher be at a disadvantage if they don't have these tools like Nerdy going forward?

Yes. Kind of at the core, we think that our ability to deliver live instructions, surround it with AI that enhances that experience is highly differentiated and allows for us to, both in the consumer world, but also in school districts, be in a position where we can influence in a way that is highly kind of embedded in workflows and sticky and adds a lot of value and takes into account our unique understanding of their operations and how students learn in such a way where we can drive significant impact. So I do think that the tools themselves will certainly continue to be leveraged more and more and more. We think the fact that we have Live, the kind of at the epicenter of that platform and can actually enhance Live now in real-time in addition to just delivering great experts consistently, which is hard to do, but doing that plus actually enhancing that interaction while it occurs during a live session or a live class, that's something where we think the idea of teachers or tutors having superpowers in real-time that augment what they could otherwise do is a really powerful vision that we've been building towards for a long time and the real-time aspect of that with computer vision and audio now is what makes it particularly exciting.

Got it. Can I just squeeze one more in on the number side for Jason. So like high-dosage teaching for Institutional side, we heard great feedback and then you launched it to the top 10 largest school districts. I was wondering what is the time line for the rollout for like nationwide? And how much of an impact do you think it will be to the Institutional segment going forward?

Sure. Great question. Look, we think that the Institutional opportunity is very real for our company, it's nascent. We've only been serving these customers for 2 years. But what you're seeing is we built out the capabilities to be able to serve the largest school districts and the most complex ones over the course of the last 2 years. Service delivery continues to improve. Schools are very happy with the products that we're providing. We've moved schools from a B district -- C district to a B district. The example that we provided in the shareholder letter was very powerful for a large school district over a short period of time. And I think, as I mentioned earlier on the call, the third quarter is shaping up to be the largest bookings quarter we've had to-date, which we continue to feel good about. What's important is that's on like the legacy products that we sold. So high-dosage tutoring, the Live+AI platform, scope tutoring programs. What that number doesn't include on a go-forward basis is all the new tools that we've rolled out and how we'll be able to sell those into school districts in the months, quarters and years to come. So net-net, we feel really good about that business. As Chuck mentioned, schools are embracing AI to an extent they haven't in past school years, which we think bodes well for our products. And quite frankly, children, unfortunately, are still behind post COVID, and we're there to help support school districts get them back up to grade level and excelling again in the classroom.

Our next question will go to the line of Greg Gibas with Northland Securities.

I wanted to maybe see -- are we at the point yet where you maybe are starting to see new activations or early engagement in terms of the earlier kind of first back-to-school cohorts? And if so, could you maybe discuss any early trends you're seeing there? I know it's early in the season, but just if there's any early data that you can share?

Sure. So throughout the spring, we felt like the kind of conversion trends, retention trends were all really healthy. We had higher levels. We had higher frequency customers on average who are far more engaged and leveraging more aspects of our platform. So we think that is a great place to be, and we should continue to lean into those deeper relationships. And those are also the people that sort of pause a little bit more during the summer. Those are -- you need less tutoring in the summer. Those are also people that we think will reactivate at high rates because it's all oriented around academic subjects, and there's a heavy kind of orientation towards the K-12 mix in there. So we feel good about the kind of mix of customers we have coming back. And then as you look at the leading indicators on the new side, conversion is improving, retention is improving, MRR [ growth ] on consumer inflected positive in the quarter and continues to inflect positive, and we like what we're seeing there. And the early waves of school districts that start over the course of about 6 weeks from the beginning of August or end of July up through after Labor Day, those trends look really positive. So it's shaping up to be a good school year. And then we feel that kind of -- that combined with some of the product deliverables we have, that have been very well received in all of our testing, could further accelerate growth.

Great. That's good to hear. And then I guess a couple of modeling-related questions. First, on kind of the dynamics within the quarter, adjusted EBITDA beat your guidance range despite revenue kind of at the lower end of that range. And just wondering if you could maybe discuss what -- any dynamics or items in the quarter that kind of contributed to the variation relative to your guidance? And I guess, secondly, if you could touch on the -- I can follow-up after that.

Greg, thanks for the question. On the Q2 EBITDA outperformance, a couple of things there. One, gross margin continued to improve beyond our expectations. We talked about on the last call, the investments we made in tutor incentives and how those on a year-over-year basis would pull down gross margins. But then when you layer that in with the price increases and new customers coming on, we expect sequential quarterly gross margin expansion every quarter as we move throughout the year. So you saw just that from Q1 to Q2, gross margins expanded 350 basis points as we move through the back of the year. Once you get to the fourth quarter, we think gross margins will be at parity on a year-over-year basis. So that's one lever. And then two, we made a lot of infrastructure investments in AI over the course of the last year, and you're starting to see the fruits of that pull through with greater efficiency, greater levels of automation and internal processes. And essentially, if you look across every P&L line item, you're seeing improvements from an efficiency perspective year-over-year. We would expect that to continue as we move into the back-to-school season and it positions us to be able to scale at levels that won't require commensurate amounts of fixed and variable labor in the future. So net-net, we think we will be profitable in the fourth quarter. If we are, that allows us to be profitable in all of 2026, and we'll continue to go from there. But net-net, on the cost side, we're feeling strong with where we're at and that there's a little bit more opportunity to gain additional leverage.