UPWORK, INC(UPWK)Q2 2026 法說會逐字稿

Good day. Thank you for standing by. Welcome to Upwork's Second Quarter 2026 Earnings Conference Call. Please note today's conference may be recorded. I will now hand the conference over to your speaker host, Peter Sanborn, Chief Business Officer. Please go ahead.

Thank you, and welcome to Upwork's discussion of its second quarter 2026 financial results. Joining me today is Hayden Brown, Upwork's President and Chief Executive Officer. Following our prepared remarks, we will be happy to take your questions. But first, I'll review the safe harbor statement. During this call, we may make statements related to our business that are forward-looking statements under federal securities laws. Forward-looking statements include all statements other than those of historical fact. These statements are not guarantees of future performance, but rather are subject to a variety of risks, uncertainties, and assumptions. Our actual results could differ materially from expectations reflected in any forward-looking statements. For a discussion of the material risks and other important factors that could affect our actual results, please refer to our SEC filings available on the SEC website and on our Investor Relations website, as well as risks and other important factors discussed in today's earnings press release. Additional information is also available in our quarterly report on Form 10-Q for the quarter ended June 30, 2026, which was filed today. In addition, references will be made to certain non-GAAP financial measures. Adjusted EBITDA, adjusted EBITDA margin, non-GAAP operating expense, non-GAAP net income, non-GAAP diluted EPS, and free cash flow are non-GAAP financial measures, and all other financial measures are GAAP unless cited as non-GAAP. Information regarding non-GAAP financial measures, including reconciliations to their most directly comparable GAAP financial measures, can be found in the press release that was issued this afternoon on our Investor Relations website at investors.upwork.com. Unless otherwise noted, reported figures are rounded and comparisons of the second quarter of 2026 are to the second quarter of 2025. With that, I'll now turn the call over to Hayden.

Good afternoon, and welcome to Upwork's Second Quarter 2026 Earnings Call. In Q2, Upwork demonstrated strong execution in an operating environment that continues to be challenging. We delivered revenue of $191.7 million at the high end of our guidance range and adjusted EBITDA of $64.1 million, exceeding the high end of our guidance range. This was driven by disciplined cost management across our entire business. Our growth building blocks of SMB, Enterprise, and AI are delivering results. We're driving momentum with larger clients, capturing higher value projects, and scaling our position as the platform of choice for complex AI-enabled work. In SMB, our prioritization of quality and customer value, not volume, continues to pay off. Business Plus continues to pace ahead of plan, with GSV growing 174% year over year as larger customers utilize Upwork for complex recurring needs. GSV per active client reached another record at $5,230, up 5% year over year, marking our eighth consecutive quarter of sequential growth in this metric. The core part of our SMB strategy is helping these businesses tap into AI for practical outcomes. For many of these customers, the constraint is not access to AI technology, but access to people who understand how to apply it to their specific industry, workflow, or business problem. Upwork is uniquely able to solve this at scale, combining a deep pool of AI-skilled talent with the marketplace infrastructure and trust that lets customers confidently tap into the human expertise they need. This is reflected in the 51% year-over-year growth we saw in our AI strategy and consulting category in Q2. SMBs are telling us that they're hiring AI experts to guide them in making their businesses more AI-native so they can grow and scale. In Enterprise, our goals for Lifted remain on track and demand signals are strong. We initiated customer migrations to the new platform on schedule in June. We also saw positive developments with customers advancing through later stages of the sales process. The thesis around serving larger customers with this differentiated offering is working. Our focus on an expanded ability to support and grow larger customers that fit our ideal customer profile contributed to 7% year-over-year growth in GSV per enterprise account, the highest level it's been in more than three years. We also saw 29% year-over-year growth in GSV from employer of record, or EOR solutions, which are made possible due to Lifted's acquisition of Ascen. Moving our employment infrastructure in the U.S. and Canada from third-party partners to Ascen's wholly-owned entities is garnering the improvement to our value proposition we anticipated, accelerating our sales funnel with the added benefit of improving our gross margins. We continue to expect Lifted to achieve approximately 25% year-over-year GSV growth in 2026. Our industry is in a period of transition as AI is changing work and transforming our marketplace in real time. The near-term AI and macro headwinds we identified last quarter have persisted in Q2 with an acceleration in the pace of AI-related automation. We now have greater visibility into the work transition underway and are updating our guidance outlook accordingly. Short-term headwinds do not change our confidence in the enduring value proposition of Upwork. While AI automation shrinks demand for some types of work on our platform, AI creates new demand in other areas. This is evident in the 22% growth in GSV from AI-related jobs in Q2, as defined by jobs stating an explicit AI need. These were up 5% quarter over quarter to reach an approximate annualized $330 million run rate. This quarter we saw indications that the quantum of AI-related work on Upwork is actually much greater. Many clients now simply expect the use of AI tools and capabilities without feeling the need to specify this in their job posts. For example, nearly half of talent in a recent survey reported that their most recent jobs were AI-related, despite only 16% of such jobs explicitly referencing AI in the post, demonstrating how prevalent AI-related work is becoming on Upwork. AI is also changing search. Our client acquisition strategy is to diversify beyond paid and organic Google Search to build a highly durable growth engine spanning emergent channels, including LLM-based search. Google's changes to search have dampened new customer acquisition with accelerated negative impact in Q2. Our SEO rankings have continued to step up, but the channel overall is smaller, underscoring the importance of the ongoing paid acquisition efficiencies we've been driving. Paid search has become our largest customer acquisition channel, and we have grown it in both relative and absolute size, while increasing efficiency. Cost per new contract start improved 22% quarter-on-quarter, and this performance gives us confidence in increasing our paid acquisition investments for the balance of the year. We've also begun tuning our acquisition approaches for new channels that we expect to scale over time, like Answer Engine Optimization, or AEO, and LLM-based referrals. While these channels are still nascent, third-party measures show that our share of AI-based mentions and referrals is 18 percentage points higher than our closest peer. Additionally, our AI integrations are beginning to drive AI-based referral traffic and set us up for growth from these channels as OpenAI, Anthropic, and others enhance their approaches. These levers give us new paths to acquire customers in this unfolding AI and marketing landscape. The expansion of our acquisition channels goes hand in hand with how we're expanding our platform capabilities. With the proliferation of AI, we've been seeing clients trying to use AI agents to hire on Upwork for projects. We are also seeing a growing volume of clients come to Upwork to find people who can complete or fix a project they started with AI. Requests like "turn my vibe-coded concept into a production website" or "humanize the translation output from this AI" are growing, among many others. In the second quarter, we delivered key milestones on our product roadmap to support these use cases at scale. And today we announced the Upwork MCP server, our next capability for enabling agentic interactions on both sides of the marketplace, backed by the talent and job access, trust, guarantees, and quality that only Upwork offers. In Q2, we embedded Upwork directly into both ChatGPT and Claude. And with today's launch, Upwork talent and jobs are now accessible inside any AI tool that connects to our MCP. This lets both humans and their AI agents hire and collaborate with Upwork talent without leaving their own environments, while giving freelancers a way to win and deliver work without leaving their preferred AI tools. The demand for human taste, judgment, and quality assessment is increasingly commanding a premium. We're building these capabilities that let clients reach directly into our marketplace from inside the AI tools where they're already working to find the human expert they need the moment they need them. And as agents become a bigger part of how work gets done, we're building toward a marketplace where clients and their agents can hire not just human talent, but can also hire specialized AI agents and human agent teams. Bringing both sides of this emerging human and agent economy together on one platform is something no other marketplace can do at our scale. These innovations will help position Upwork as a key beneficiary of the AI shift that is underway. Now, I'll walk through our financial performance for the second quarter in more detail, as well as our updated guidance. Our financial results in Q2 demonstrate our capability to execute with discipline, driving profit expansion even as top-line volume reflects the transition underway. GSV was $966 million, reflecting the continued burn-off of lower-value, highly automatable work, the evolving impact of AI on new client acquisition, and a subdued labor market. Total revenue for the second quarter was $191.7 million, representing the high end of our guidance range, supported by strong, high-margin contributions from our ads and monetization levers, including dynamic pricing, Connects, and Business Plus. Revenue from these levers was up 15% year over year, and Connects and talent subscriptions now represent 15% of total revenue. This drove a strong take rate of 19.8% with highly accretive, high-margin revenue streams that also improve high-quality matching. Q2 active client count was 763,000. Active client count continues to be pressured by AI automation and search trends, labor market conditions, and our strategy to focus on client value over volume. Our average spend per contract reached an all-time high over any 12-month period in company history, presenting the sixth consecutive quarter of positive year-over-year growth. GSV per new client grew year-over-year for another consecutive quarter, confirming that we are attracting clients with larger initial budgets and longer project horizons. Non-GAAP gross margin for the second quarter was 77%, remaining near record levels as we manage infrastructure and support costs efficiently. Non-GAAP operating expenses reflected the early execution of our restructuring actions we announced in May, as the annualized $70 million of OpEx reduction we initiated is expected to generate approximately $40 million in realized savings in fiscal year 2026, allowing us to continuously fund our pillars in SMB, Enterprise, and AI while expanding operating leverage. Adjusted EBITDA for Q2 was $64.1 million, exceeding the high end of our guidance range and delivering an adjusted EBITDA margin of 33.4%. Non-GAAP net income and non-GAAP diluted EPS were similarly strong, while our GAAP metrics reflect the impact of our Q2 restructuring charges. While our capital-light marketplace model consistently yields exceptional free cash flow conversion, this quarter's free cash flow of $35.9 million included the impact of one-time cash payments related to our restructuring. During the quarter, we closed a new $150 million revolving credit facility with a $50 million accordion. This positions us to repay the August 2026 convertible notes at maturity while supporting our disciplined capital allocation strategy, opportunistic M&A, and our share repurchase program. We repurchased approximately 164,000 shares in Q2 while we closed the revolving credit facility. Year-to-date, we have repurchased approximately 8.3 million shares. Turning now to our updated outlook, as I mentioned at the top of this call, the near-term AI and macro headwinds we identified last quarter have persisted throughout Q2 and into Q3, with an uptick in the pace of AI-related automation and impacts from SEO. We are adjusting our full year guidance on both the top and bottom line accordingly. For the full year 2026, we now expect revenue in the range of $730 million to $750 million. This assumes a heightened pace of AI-related automation and no improvements in the labor market. As planned, we will continue to expand take rate this year as part of our overall growth story. We expect full year 2026 adjusted EBITDA of approximately $225 million to $235 million, which represents an adjusted EBITDA margin of 31% at the midpoint. This is driven by the flow through of our reduced revenue outlook alongside planned marketing spend. We expect full year 2026 non-GAAP diluted EPS to be between $1.38 and $1.43. Our updated guidance also assumes stock-based compensation of approximately $60 million to $65 million for the full year 2026. For the third quarter, we expect to generate revenue in the range of $176 million to $184 million and adjusted EBITDA in the range of $50 million to $54 million, which represents an adjusted EBITDA margin of 29% at the midpoint. We expect Q3 non-GAAP diluted EPS to be between $0.31 and $0.33. Our underlying metrics, including record GSV per active client and continued Business Plus and AI work growth, confirm Upwork's strong value proposition, even in a changing market. Our product innovations and tailored offerings for larger customers are deepening our moat, and we continue operating with financial discipline. We're navigating the near-term impacts of this significant evolution in how work gets done, while continuing to build Upwork to capture the demand, talent, and economic opportunity that this new era of work offers. Thank you for your continued support. Operator, we are now ready to open the call for questions.

And our first question comes from Ron Josey of Citi.

Hayden, I had two questions, one on operations. You talked about the challenging operating environment continuing. We'd love to hear more about the view and the visibility, and then more insights on the SEO challenges that you mentioned. And then some interesting data came out today around the MCP server. I wanted to understand directly how that might impact the business going forward given the AI benefits.

Thanks, Ron. From a visibility standpoint, we are really in the midst of a transition of work broadly. Our decisions, including our growth strategy and our effective cost management, are enabling us to navigate this situation from a real position of strength. The trend we saw this quarter is a continuation of last quarter, but with an acceleration in the AI automation of very low complexity work and further deterioration on Google SEO, especially notable in the non-brand channel. That's understandable given that demand for less complex work is shrinking as customers embrace AI, and this is flowing through to our platform. The important thing is this is an evolution and the situation is fluid. We are already seeing encouraging signs that our strategy is working, and this is showing us what Upwork will look like as this transition unfolds. Upwork will be home to a lot of new work. AI-relevant work is already a big and enduring part of Upwork. We're seeing this with a $330 million GSV run rate in our AI categories. As I mentioned earlier, this is really an undercount because there's a lot more AI usage de facto across the platform. We can also see that on the other side of this transition projects are becoming much bigger. This is how we've seen eight consecutive quarters of increased GSV per client. It's not just a mix shift; we are actually seeing increases in things like the length of hourly work contracts, which hit a record of 100 hours on average this quarter. A hallmark of the new Upwork will be these new channels and the new participants in those channels, which speaks to the MCP. With the launch of the MCP, we are stepping into new AI-first client acquisition strategies, which started working already with integrations into Claude and ChatGPT. This now enables Upwork to integrate agents and agent participants on both sides of our marketplace at scale. We know how to navigate these changes; it's in the DNA of our company. Regarding SEO, these are industry-wide trends that are affecting the business similarly across peers. Google has changed SEO referrals and the layout of the search page so that many fewer searches result in any kind of referral to businesses. That's weighing on new client acquisition. It's compounding with a subdued labor market and AI-driven changes. The good news is we've done a lot to counter this, including rebuilding our marketing funnels. Paid acquisition performed well in Q2, giving us confidence to increase marketing spend by approximately $5 million to $10 million incrementally in the back half of the year. These new acquisition channels around AI are the future for lower-cost organic acquisition and let us integrate where work is happening. On the MCP specifically, we built this not just to create a new demand channel, but because we observed agents already coming to our site trying to do work on behalf of clients. Previously they couldn't get very far due to restrictions on agent activity. Now, with the MCP server, Upwork can be called by and embedded into any AI tool that a client or freelancer is using. This enables clients to post jobs and hire from inside their AI tools and lets freelancers manage proposals, projects, and communications within their preferred AI environments. It allows freelancers to operate as human supervisors to trusted AI tools integrated into Upwork, giving both clients and talent full visibility and auditability of tool usage. This launch brings Upwork to where customers already are and opens broader access. Because clients working in AI tools expect immediacy, human impact, and people on demand, our scale and talent breadth allow us to offer humans at scale on demand through these solutions, which few others can do.

And our next question comes from Eric Sheridan of Goldman Sachs.

Maybe two if I could. One that will dominate the debate is how do you think about the duration to get to the end state? Is this a 6- to 12-month transition or could it be more than 12 months? I'm curious for any framing you think about what the countervailing factors in the business might lead to in terms of a mixture of these headwinds and tailwinds before we get back to more normalized levels of growth. Second, among the active client base you have today, how many of those active clients do you think will eventually transition to higher-end, higher-quality AI-driven work versus how much is about rebuilding your active client base to be centered around where the nature of work on your platform is going for the longer term?

Thanks, Eric. Timing is hard to predict. It would be disingenuous to claim exact knowledge of how long this will take. The important point is that this is an evolution, and on the other side of these changes, Upwork is structurally positioned as a beneficiary because our platform can serve where the market is going. We're focused on executing the strategies that are working. We see enterprise growth, Business Plus up 174% year over year, and AI tailwinds in parts of our business that speak to AI work. These will continue to scale and compound as we work through this. That makes me optimistic about where we'll land. Regarding the active client base, we see this as an expansion, not a replacement. Very small business customers on Marketplace Basic are shifting workloads to be more AI-oriented. We see strong resonance with SMBs and larger customers who want Business Plus combined with AI transformation. This is why our AI strategy consulting categories were up significantly in the quarter. Customers are using new tools to do different types of work than a few months ago. We see this across customer segments: they're leaning into AI workloads and changing what they come to Upwork to do.

And our next question comes from Bernard McTernan of Needham & Company.

Two for me. First, on the AI category work: we've seen a slowdown over the past two quarters — growing 50% year over year in Q4, 40% last quarter, then 22% now. I know there are caveats you gave in your prepared remarks. Can you dig in deeper there? Second, the marketplace take rate — we were expecting it to be down sequentially, but it came in higher sequentially. Was that just Business Plus, or was there anything else to call out in the quarter and how should we think about take rate for the rest of the year?

On the AI category, it's pacing at a $330 million run rate. I want to emphasize this is an undercount of AI work on Upwork because customer behavior is changing. Clients who previously specified AI in job posts now often assume AI usage and don't mention it. In a recent survey of talent, almost half said their most recent project was AI-related, whereas only 16% of job posts explicitly mentioned AI. AI is permeating all types of work, and the metric reflects that transition. Regarding marketplace take rate trends, we had success with ads and monetization in Q2 and will continue to expand these in Q3 and Q4. Expansion of dynamic pricing on freelancer fees and other levers contributed to the over-performance in take rate. Our outlook is for take rate to continue to step up in the back half of the year.

And our next question comes from John Byun of Jefferies.

This is John Byun on behalf of Brent Thill. Question on Lifted: you mentioned it's on track for approximately 25% GSV growth and that the employer-of-record category did very well. Can you share more about the migration there, the project categories that are contributing, and how that is building in the pipeline?

John, in Q2 we saw great progress with both customer migrations and pipeline building for Lifted, which gives us confidence for the approximate 25% GSV growth in full year 2026. June was when we started customer migrations in earnest, and that activity is ramping up because it's going very well. Customers are providing very positive feedback. As we run them through the new systems and workflows, things are proceeding smoothly. The pipeline remains incredibly healthy and customers are converting. We talked about pipeline expansion last quarter, and now we're seeing those customers move through the funnel. One example: we recently won a multimillion-dollar RFP with a very large global company for a staff augmentation contract that we would not have been invited to participate in before Lifted and before we rounded out these capabilities. We're seeing proof points between that and the EOR improvement I shared earlier: these capabilities are helping us close deals and improve margins versus using third parties. We're excited about what's ahead through the rest of the year and into 2027.

And a quick follow-up on the SEO change. Does it look like this is stabilizing or is it ongoing? What could the continued impact be from the search changes?

We started seeing these impacts accelerate in Q2. There is volatility in the industry around this, and our outlook is that they could continue to get worse through Q3 and Q4. That's baked into our current guidance. We haven't seen stabilization yet, and to the extent they do stabilize that would be better than our current outlook, which assumes further deceleration.

And our next question comes from Josh Chan of UBS.

Two quick ones. Could you give a sense of the monthly cadence on when things started to weaken? You said they're not stabilizing yet, but when did the weakening begin? Second, previously there was thought that only a portion of Upwork's GSV might be susceptible to AI. Has your thinking changed on that?

At the time of our last earnings call we had seen some of these trends, particularly around AI automation, stabilize. Subsequent to that, in this past quarter, we've seen those trends accelerate. This is an acceleration of erosion in the part of the business we had identified as more exposed to AI automation — it's a pull forward of activity versus our previous expectations. Regarding susceptibility, this acceleration affects the portion we had already flagged as exposed; it hasn't fundamentally changed the framework, but it has accelerated the timing of that exposure.

And our next question comes from Matt Condon of Citizens JMP.

Hayden, we hear enterprise clients are leaning into AI and reducing full-time headcount. What are your thoughts about AI potentially impacting freelancer work in that segment as well?

Good question. The reason we built Lifted to address the full $650 billion enterprise market was based on the idea that enterprises spend heavily on contingent labor, but only about 10% historically went to freelancers. Previously Upwork was eligible for only part of that. Now, with Lifted, we can serve customers across many work types, which helps insulate us from AI risk in any single work category. We're not saying freelancing won't be automated in parts, but being able to diversify how we serve customers across every type of work gives us more opportunity to bring AI talent to bear and helps protect us if one part of a customer's business automates faster. This expansion is healthy for the business and is why we're focused on driving Lifted's growth into the back half of the year and into 2027. On go-to-market, with SEO changes not stabilized and LLM investments still nascent, we are leaning into different channels. We're seeing strength in international SEM where we're under-penetrated and have gotten very good returns in Q2 testing. We also have headroom across channels including Meta, Reddit, podcasts, and connected TV. These nascent channels are ramping and getting more sizable, and we now have confidence to scale incremental spend efficiently in the back half of the year.

And our next question comes from Marvin Fong of BTIG.

You mentioned investing more in marketing. How did you arrive at the $5 million to $10 million range? Is there potential to do more to drive growth? Second, on monetization levers like Connects and pricing, how are you managing those levers in the current environment? Connects is growing faster, but how should we think about pulling those levers to drive revenue?

On the $5 million to $10 million incremental marketing spend, we arrived at that figure through extensive testing in the first half of the year across the channels I mentioned. Based on performance and degradation curves in those channels, this is an optimal starting point where we can deploy capital efficiently and potentially see some benefits in Q4. It's a dynamic space, and we'll monitor results closely. On monetization levers, as we shift from customer volume to customer value, we balance how much we charge on the talent side for job access with a focus on match quality and fill rate. In areas where the marketplace is more congested with talent relative to client demand, we can use pricing levers that benefit take rate while preserving match quality. Our North Star remains match quality and fill rate. In the back half of the year we will pull certain levers as marketplace dynamics evolve, always within guardrails that preserve match quality.

This concludes our question-and-answer session and today's conference call. Thank you for participating and you may now disconnect.