Talkspace, Inc. (TALK) Earnings Call Transcript

Hello, everyone, and thanks for being here on today's call as we go over the results for our fourth quarter and the whole of 2025. Back when I came to Talkspace near the close of 2022, the strategic shift away from our Consumer model and toward a Payor fee-for-service approach was already underway. Looking back now, I feel proud of how far we've come on the financial side, on the operational side, and in pursuit of our mission of bringing comprehensive, personalized mental health care to everyone. From 2022 onward, our revenue has expanded at a 24% CAGR, powered by roughly 56% annualized growth in Payor Sessions. Over that stretch, operating expenses measured against revenue kept falling, which helped generate operating leverage and better EBITDA margins. Across the entirety of 2025, we brought in revenue of roughly $229 million, up 22% compared with the prior year, with payer growth of 38% fueling that result. On top of that, adjusted EBITDA grew by more than two times, climbing from roughly $7 million in 2024 to $15.8 million in 2025, equal to an adjusted EBITDA margin of 7%. Our Payor growth — and we now reach well over 200 million lives via insurance and employer benefits — comes from 2 drivers: first, the strategic initiatives we've launched to draw people toward Talkspace, which include focused work on boosting awareness and generating high-intent referrals, along with tighter Payor partnerships aimed at smoothing the patient journey and making it simpler to locate care; and second, the broader set of offerings we're building out in the Payor channel to serve new populations and varying acuity levels. Running beneath both of those efforts are the ongoing enhancements we make to the member journey and to our clinical network. Through paid media strategies, search optimization, partnerships, and broader brand recognition, we keep building consumer awareness. Our awareness campaigns have performed very well across the last 3 years, with the Talkspace brand gaining ever more recognition even as our marketing spend fell considerably during that same window. Our work on high-intent referrals has paid off, with volumes climbing month-over-month from Amazon, Zocdoc, and our strategic partners. Thanks to our team's efforts to optimize on and off our website for greater visibility and citations, Talkspace also has a robust and expanding presence in large language models. During the fourth quarter, general purpose LLMs accounted for a rising share of traffic and checkouts as we keep growing this new and expanding channel. Because the Payors recognize the high-quality clinical care we deliver, they have teamed up with us on a number of new initiatives designed to make the patient experience even simpler. That covers directory integrations with several of our Payor partners, some of which use single sign-on so patients can easily log into both platforms. Others are placing Talkspace scheduling directly into their directories, letting patients book sessions while staying on the Payor site. Right now, we're collaborating with a single partner to roll out the ability for their care coordinators to book Talkspace appointments for patients, and we plan to extend this tool to additional partners and Payors. Over the course of the year, we broadened what we offer in the Payor channel as well. We put money into our psychiatry business, increased enrollment in both military and Medicare, and bought Wisdo — the lower acuity, AI-driven social health platform focused on peer-to-peer community and coaching. On the military side, enrollment keeps climbing month-over-month after our January 2025 launch, and patient engagement is rising too through our direct-to-enterprise contract with the Navy. Medicare enrollment is also continuing to grow. And thanks to bringing Wisdo on board, we've noticed heightened interest from Medicare Advantage plans because of Wisdo's demonstrated effects on loneliness and social isolation. Beyond that, Wisdo's partnership with Novo Nordisk to deliver group coaching for patients on GLP-1s creates a fresh opening for us into a pharma partnership category that was previously untapped. Our youth programs, first introduced at the end of 2023 across major markets such as New York, Baltimore, Seattle and North Carolina, keep producing strong measurable impact at scale. In New York City by itself, over 45,000 teens are signed up for our Teenspace program. 66% of enrolled teens demonstrated measurable clinical improvement, with anxiety, depression, relationship challenges and stress management being the most common presenting needs. The program reaches historically underserved communities, with nearly 45% of participants residing in areas of high health and income disparities and 82% identifying as BIPOC. Engagement stays very strong, with over 90% of teams actively texting their therapists and more than half selecting messaging as their sole modality. These program results reinforce Teenspace as a scalable public-private partnership model and Talkspace's leadership in youth mental health solutions. In psychiatry specifically, we grew our network of psychiatry providers to over 400, and we implemented a number of improvements to the patient journey to streamline processes, such as simplifying the medication management workflow so medications can be sent directly to the member's pharmacy of choice. In April, we launched our integration with Amazon Pharmacy, letting members seamlessly fill prescriptions from their Talkspace provider and receive fast free home delivery, which makes for a more convenient patient experience. Additionally, we built an easy pathway for members using Talkspace for therapy to get an internal referral to a Talkspace psychiatrist — an investment where we're seeing strong traction. On AI, we continue to use the technology to enhance business operations and fold AI enhancements into the platform to further improve the patient journey and provider workflow. These enhancements have cut friction in several areas, reducing registration drop-offs and helping more patients successfully start their care journey. Once a member is onboarded, we've also made scheduling their appointments easier, boosting the number of patients who continue care after a first session. The payoff from all this shows up in more checkouts, plus a 49% jump in patients who finish a third session within their first month of care. One more thing driving our session growth higher: Talkcast, the personalized AI-generated podcast I've mentioned before, has performed really well. If a member listens to a Talkcast episode in the gap between session one and session two, that person is 20% more likely to finish both a second and a third session. As of today, we've created more than 76,000 episodes, and the response has been overwhelmingly favorable. Positive feedback came from 95% of provider reviews and 92% of client reviews. On top of that, our approach to network management keeps us focused on refining the clinician network so it matches the particular utilization patterns we're observing, which means we make sure clinicians are on hand in the right place at the right times to match patient demand. Let me now shift to the TalkAI agent we've spent the past year building. General purpose large language models may be in use by an enormous share of the world's population today, but supporting mental health was never their purpose. These models have opened access for millions, and that's a positive development, yet sadly they've produced a wave of documented harmful results. Delivering mental health support calls for a much more specialized and nuanced approach — one that includes confronting distorted thinking, spotting delusions, and detecting risk as it happens. What we've created with the TalkAI agent is meant to be the first safe AI agent built specifically for mental health support, drawing on clinically accepted standards of care, with ongoing human oversight and HIPAA privacy safeguards. This LLM has been trained and fine-tuned using Talkspace's enormous mental health data set; it detects 10 risk areas in real time, enables sound decision-making, and sidesteps the problems already witnessed with general purpose LLMs. It keeps clinicians engaged at all times through explicit escalation pathways that link at-risk users to a licensed human clinician immediately. Rather than substituting for clinicians, TalkAI expands what they can reach, holding to rigorous clinical standards while finding new users who might benefit from human contact. My view is that demand for human care from trained therapists will grow, since millions more people will be identified as needing professional help past what our agent can offer. Beta testing is happening right now this quarter, and we expect to launch commercially late in Q2. To wrap up, it's clear we've made enormous progress over 3 years. A huge opportunity still lies ahead of us, and we're positioned to keep pursuing it aggressively. Beyond the core business, we think we've placed ourselves strategically to lead in applying AI to mental health services amid today's fast-changing landscape. I'm happy with the Q4 results and with how our business performed across the full year. As we look toward 2026, my confidence is high regarding what we can achieve and the chance we have to keep expanding the business and improving profitability, and the robust momentum observed so far in 2026 gives me encouragement. At this point, I'll hand the call to Ian.

Good morning, and thanks for being with us. To start, I'd like to back up what Jon said — we closed out the year carrying some genuinely strong momentum, and we're in a good spot for that to keep going. Before taking you through how we see 2026 shaping up financially, I'll go over what we posted in Q4. Now to the fourth quarter numbers. Revenue came in at $63.0 million in total for the quarter, which works out to a 29.3% rise compared with the same period a year earlier. Payor remained our biggest engine of growth, bringing in $47.7 million, a 41% jump year-over-year. That increase came from higher session volume along with expansion among clients we already had. To put a figure on it, sessions totaled 450,000 during the quarter, up 36.3% from a year prior. On top of that, unique active Payor members reached 124,000 for the quarter, which is a 29.7% year-over-year gain. Over in direct-to-enterprise, we booked $11.6 million, up 21.8% versus the prior year. As we mentioned during our third quarter call, a number of new launches slipped out of Q3 and into Q4, and DTE additionally got a lift from the Wisdo acquisition, which we closed on October 1, plus revenue tied to implementation work for some new accounts. Consumer brought in $3.7 million year-on-year, in line with our deliberate choice to put Enterprise and Payor channels first. Gross profit reached $26.9 million, an increase of 24.4% year-over-year, giving us a gross margin of 42.7% for the quarter. That figure declined 169 basis points from a year ago, largely because of a revenue mix shift toward Payor. Operating expenses came to $23.1 million, up 9.6% year-over-year. Notably, opex as a share of revenue showed a meaningful improvement, landing at 36.7%, which is 660 basis points lower than the fourth quarter of 2024. Adjusted EBITDA was $6.6 million, which represents 147.1% growth year-on-year, with an adjusted EBITDA margin of 10.4% — an improvement of almost 500 basis points over the prior year. Now onto the balance sheet. We closed the quarter holding $92.6 million in cash, which is $25.2 million less than a year earlier, mainly due to our share repurchases — $17.2 million total across full-year 2025 — plus the Wisdo acquisition; for the whole of 2026, here's our initial guidance: we're expecting revenue somewhere between $275 million and $290 million, which would be 20% to 27% growth year-on-year. For adjusted EBITDA, we're looking at a range of $30 million to $35 million, which translates to 90% to 122% growth. If you go back to that three-year plan we introduced at the start of 2024, which extends through the current year, our expectation is a three-year revenue CAGR of about 23%, calculated using the midpoint of the 2026 guidance we've provided. That figure falls within the 20% to 25% objective we set out in that same outlook. Turning to profitability, we anticipate closing out 2026 with EBITDA margins somewhere in the mid-teens, placing us near the top of the 12% to 15% range established in that plan. Let me now cover several items regarding the assumptions that support this outlook. If we break performance down by quarter, the expectation is that revenue will rise steadily through the year, similar to the pattern from a year ago. During the first six months, we'd see a bit less than half of the full-year revenue total, since both active Payor members and sessions build up as the months progress. As for how revenue is composed, we're expecting Payor revenue expansion to keep pace with the Payor growth rate recorded in 2025, driven by the activation approaches Jon walked through previously. Like we've said previously, the Payor segment gives us considerable visibility, since a Payor member remains enrolled for a longer period than a cash-paying customer, and a meaningful portion of the Payor revenue we'll book in 2026 will come from Payor members already using the platform at the close of 2025. For D2E this year, we're again anticipating growth in the low single digits. One thing to remember: renewal volume peaks in the first quarter, which means Q1 experiences the heaviest attrition of any period during the year. Certain implementation revenue also lifted Q4 results. As a result, we'd anticipate D2E revenue declining sequentially in Q1 relative to Q4. Finally, our Consumer revenue is expected to continue shrinking, and that's intentional. Even so, because the base we're starting from in 2026 is less significant, the overall drag is considerably smaller. The midpoint of our 2026 revenue guidance works out to 23% growth compared with last year, but if you look at our Q4 run rate revenue—more than $250 million—the midpoint translates to 12% growth. That gap shows how growth picked up speed throughout 2025. Along with the internal efficiency steps we're still putting in place, these patterns should produce additional operating leverage across the P&L. On the adjusted EBITDA margin side, our expectation is to begin 2026 at a high single-digit percentage and finish the year at a mid-teens level, and that pattern should give us quarterly adjusted EBITDA movement much like what occurred in 2025. To wrap up, we see Talkspace as strongly positioned to deliver durable growth along with further margin improvement, backed by solid momentum in our Payor segment, better operating leverage, and greater clarity around upcoming demand; at this point, we'd like to take questions. Operator?

Nice work on a strong quarter. Regarding the large language model you've got in beta testing right now, what do you view as the main obstacles when it comes to getting folks who currently rely on general purpose large language models to switch over to yours once you launch it?

Appreciate it, Steve; this is clearly still very much a work in progress. Like we said, we're running a beta with people signing up right now to participate in that testing of how this thing performs. On the question of how this product is positioned—as a place for substantive conversations, where your data remains safeguarded and you have both security and safety behind you—I'm not in a position yet to predict the traction we'll get or which user segments will choose it over competing LLMs. It's simply too soon for that. Everything is still very much in flux. So my answer right now is: keep watching. After the beta concludes, we'll be able to share a great deal more. Some initial results have started to arrive. But at this point, what I'd tell you is to keep watching and we'll see how it plays out. This is going to sit in a somewhat distinct spot relative to general-purpose LLMs. I'm not obviously attempting to < >. What I am saying is that things are still at an early stage, and at present we hold quite a bit of intriguing data, though we will definitely discuss this further as the coming months unfold.

Understood, that makes complete sense. Ian, during your remarks about the 2026 guidance, you noted that the bulk of revenue already comes from members on the platform. So my question is whether the top end of that guidance depends on bringing in extra new members who aren't yet on the platform. Put another way, what would it take for us to reach the high end of the guide?

Yes. Steve, to be clear, in my prepared remarks I believe I described a material amount of Payor revenues as coming from existing members on the platform. I want to highlight that because it tends to slip people's minds — under the Payor model, that longer lifetime on the platform and longer tail of revenue gives us, from a visibility standpoint, a far greater level of conviction when modeling out the Payor revenue. So when we begin Jan 1, it isn't a majority, but you can think of it as 30% to 50% range of our Payor revenue actually arriving from folks we already have on the platform. Beyond that, clearly, to echo Jon's comments, we'll keep pushing on both paid marketing work and additional organic work on the marketing front, where there's a lot of really exciting LLM optimization work underway. And then very importantly, the direct integrations we're building with the Payors, becoming more embedded with them to lower that friction for people who find us through their insurance portal. All of that will drive new users throughout the year, just as we've done sequentially throughout 2025, which then obviously carries that long tail of sessions pulling through as well.

Congratulations on an excellent quarter. To build a bit further on those Directory integrations—clearly, you've seen strong results with that initial Payor partner you launched alongside. Could you remind everyone roughly how many more of those deeper integrations you expect to complete with other partners during this year? Also, what lessons came out of that first partner that could be applied again to keep utilization at that high level with you?

Sure, I'll kick things off, and after that I'll pass it to Jon. Regarding that initial Payor, as you mentioned, things have gone really well. They're pleased because they're delivering a far more user-friendly experience to their members, and it's simpler—and honestly, less aggravating—to navigate your care journey. From a CAC standpoint, we clearly see this as highly accretive, since it brings that added conversion and extra traffic originating from the Payor. So it's still early in '26, Ryan, and we're certainly putting in the effort to expand further. Based on what we can see right now, roughly—it depends on how you count—let's say 3 directory integrations are underway in the early portion of '26. So at minimum, 3 for certain. When it comes to materiality compared with a single one last year, the total is likely comparable in scale, in terms of population, perhaps somewhat larger when you combine the 3. That means the opportunity is at least as large as, if not larger than, what we got from the '25 integration. Here's what we're finding, and it's quite interesting: for some of these directories, this marks their first attempt at such integrations. That puts us in a fortunate spot, collaborating closely with them on how the directory gets designed, which clearly gives us influence to help mold that experience, drawing on our own expertise from a decade of running a marketplace business ourselves. They genuinely value the insight we can offer there, and frankly it benefits us too—when it comes to the algorithm, what makes a provider more likely to be hired? Is it quality? Is it scheduling? And how that search algorithm gets structured—we have a voice in those decisions.

That's really useful context, Ian. And clearly, much of what you've accomplished so far with the Payor segment has come from the commercial side, and of course, from the military as well. I'd love to hear how you're thinking about what might be possible in Medicare in 2026, especially now that CMS is launching this access program. Could this represent a chance to really accelerate things, or to push deeper Medicare efforts forward, or to boost utilization in that area? And do you plan to take part in that program?

Yeah. Appreciate the question. So when it comes to the access program, the answer is yes; we're very closely tracking the access program. We've discussed it; we are — class, we are — we have put in our submission. We'd like to be involved. It's built around outcomes. We feel quite at ease with how an outcome-based model would work; so yes is the answer there. We — and also, like you've heard me mention, we're — the Wisdo acquisition in Medicare and MA has gone quite well and keeps expanding. And like we've said before, we increases. It hasn't been a shock, as I've mentioned previously; that market is fairly hard to break into simply because it's so widespread and it spans all 50 states. But we are making — let's say progress is being made; yet with Wisdo, the access program, and the work we're doing on the ground, we remain confident about how it will grow.

Congrats on a solid 12 months and the guidance ahead. Jon, near the close of your remarks, you mentioned something regarding momentum that's already present in '26. I got curious whether you might expand a bit on what you're already observing across nearly 2 months, and what underpins that remark?

I'd say — that remark stems from the fact that as the year kicks off, it genuinely shifts the picture somewhat, since folks returning are reviewing the assurances, and they're starting to engage again across a range of levels. Yet the main thing we use to measure everything is people arriving on the platform and completing sessions. So I chose that remark deliberately, reflecting that as we finish out 2025, we keep witnessing that momentum carry forward in the early part of '26.

Correct. Richard, as you're aware, January gives us a chance to run quite a few marketing campaigns, since it comes after the holidays and after the New Year's resolution period. So there's considerable backing behind our marketing push. The momentum Jon mentions is exactly that — the checkouts we're observing, the CAC conditions we're seeing, putting ourselves in front of people — and all of that is factored into the guide, in line with the guidance.

All right. For my second question, I'd like to focus on behavioral health care costs in general. We've discussed this before, I realize, but when you look at certain benefit brokers, they list rising behavioral health spending among the biggest expense categories. Clearly a portion of that is inpatient, though outpatient also contributes. What I'd like to know is what your discussions with Payors have covered regarding health care costs that keep climbing. You also just brought up Medicare, saying you feel at ease with outcomes-based arrangements and similar models. So I'm wondering how you view possible changes to utilization management or reimbursement, plus the broader market, when it comes to behavioral health?

Sure. This lines up with things we've said before; we're certainly aware of what's happening with health care costs and the fact that premiums are going up for many people. Still, keep in mind that with mental health — and I don't mean only the TAM, the market — the more that individuals get involved with mental health, the more it actually produces savings. As you're aware, an enormous body of data already exists showing that a solid mental health support program reduces costs on the medical side. That's the first point. Second, to repeat, most of the expenses being examined are truly tied to the in-hospital side and in-hospital diagnosis. It genuinely isn't — it isn't related to our work on the outpatient side. Nationally, the outpatient portion of mental health represents a very, very tiny slice of the pie when it comes to health care spending. And then there are some of our other products — once the talk AI agent becomes a reality, it will serve as a cheaper option. The Payors are already aware of that; so I believe there are plenty of positives coming from what we do. That said, I don't believe the global or national problems around health care cost ought to affect us at all. I simply keep seeing this expand, since it's genuinely a win-win for everyone — more individuals receive mental health care.

Following up on that point, do you believe that your capacity to demonstrate results, along with the information you possess, sets you apart when Payors perhaps reduce the pool of providers — or suppliers — they're actually engaging for such services?

Without a doubt. We currently hold a few initial value-based agreements. To be frank, they're fairly basic. It's the duration until a first appointment, the duration until a second appointment, and the count of individuals who attend. But — so every one of those elements is already in place for us; thus, we feel quite at ease with virtually anyone who approaches us seeking a value-based arrangement, since we already have those things established. And one factor behind our strong comfort level is the network. So the curated network matters enormously, which means we track — as you're aware, we track the quality, we examine what therapists are doing; that's extremely important under a value-based contract, since you must be capable of measuring outcomes. And in order to measure outcomes, you need the ability to manage the network. You need the ability to examine and determine the quality that is being provided; so possessing those — possessing all of that in place is truly, truly essential for fulfilling a value-based contract. So we feel quite at ease with all of it.

That dynamic you mentioned, Richard — I believe it's precisely what's unfolding when it comes to the directory side of things, correct? The fact that we're being chosen for these first embedded directories isn't accidental; it stems from many years spent supplying that data to them, exactly like you're pointing to — the clinical oversight, the QBRs alongside the Payors, the audits. So in certain respects, they're giving us a reward through what I'm not sure you'd call it, Richard, but trimming the network down somewhat. Obviously, through completing these directories alongside us, our expectation would be — just as happened with a single Payor last year — that we'd capture a disproportionately large chunk of traffic to their Payor portal. So in an indirect way, they're doing precisely what you're speculating about.

Jon, I'd like to pose something, okay? You made a remark a bit earlier, correct? Many individuals these days are hunting for information, yet they no longer simply visit a search engine, right? They head to ChatGPT or a similar tool and pose their questions there. So you've discussed how one optimizes in order to be surfaced by these LLMs, so that folks can be pointed toward you. And it seems like — is this the fresh SEO? Is even search engine optimization still that significant? Is the real question now how you optimize to be surfaced by LLMs? Since that's something we've been hearing from other companies too, more lately. And tied to that is, how do you then link from that sort of first contact by a patient — someone who is interacting with a chat like an LLM or a ChatGPT or similar — and bring them to your AI bot, right, which would be a more safeguarded setting for patients. How do we connect those two? Or how do we get them to look you up first? Let me begin there.

Right. Great question. Actually, just — I'm not sure whether you saw it, but there's a piece in New York Times this morning entirely about search engine optimization via LLMs. And I have discussed this. We have implemented. Our marketing people have known about this for a good while. So we actually have staff who are solely engaged in LLM search strategies, so that we do show up on whether it's Gemini or Claude or ChatGPT. So we have been optimizing — there's a different name for it. It's termed generic — I believe it's termed generative optimization rather than SEO. But even so, having said that, we aren't merely aware of it — we put in place mechanisms to ensure that people do locate us. And we monitor it as well, incidentally. So we have watched this rise month to month to month. People who are locating us on the other LLMs, the count of such people, it keeps growing month-to-month. So first, that's what we're doing regarding the SEO, quite at ease about where we stand compared with that strategy. As for people locating us, that's an intriguing question since we haven't gone to market yet; we do possess a fairly fully baked initial marketing plan so that people will locate us based on whatever they're seeking; now there are many nuances to that. We lack time here to probably cover all of that. But meaning are they truly seeking therapists? Are they seeking to have a serious conversation. There are many various things that people seek. So, like I said a few minutes back, the way we see it is this: if someone wants a confidential, serious discussion — especially about relationships or other matters that might be troubling them — and their data is safeguarded, and we bring substantial clinical expertise, then people will choose where to take that conversation based on what they're looking for. Obviously, we don't — the truth is, we don't yet know the answer. But we're getting positioned. We'll begin in a somewhat different posture compared with the rest. Still, I can't claim the jury is out, since we haven't even started. So, as I mentioned, we'll learn a lot more when the beta wraps up — why folks are drawn to it, how they're engaging with it — and that will give us a meaningful set of data points about how we'll approach the market relatively after that.

There will be a distinct set of marketing programs and a dedicated budget specifically for the LLM offering, correct? In the guidance, essentially no revenue is tied to TalkAI, which, as Jon mentioned, is going to be released to the public during this summer. However, a distinct marketing push will exist for it as well. Even though we've never shared numbers around traffic or conversion on our main platform, it's enough to say that if you examine even the top-performing e-commerce companies, you can picture a huge amount of visits arriving at our site that never complete a purchase, correct? Therefore, drawing on the research we've conducted, we see the TalkAI offering as something that will absolutely expand our TAM, since many individuals are curious about therapy, they arrive, they look things up. Yet as you're aware, entering therapy isn't the kind of passing choice someone makes simply to purchase a product, correct; so many people do come to — and we believe this will capture, I don't want to label it lower intent, but perhaps a segment of people who aren't yet prepared for a human-to-human therapy appointment, yet are open to engaging with something closer to a GPT-style interface. Thus we'll run a distinct marketing push focused on that from a paid angle. But I'd also like to point out that on the organic side, we believe many folks visiting our site right now generate no monetization whatsoever, and we'll keep them once a TalkAI offering exists; and that — and this stands entirely apart from the GEO, the generative engine optimization efforts, and I'd agree with your concise assessment. It's essentially the new frontier for SEO, and frankly, we're gaining a great deal from our longstanding SEO strength. So it's — as Jon put it, it's a modest yet rapidly expanding channel for us on the core side.