Yelp suspended its share repurchase programme on August 6, 2026, telling investors it will direct cash toward retiring the $100 million drawn on its revolving credit facility. The pause arrives in a quarter where advertising revenue fell 3%, restaurant and retail advertising dropped 10%, and the company's non-advertising line nearly doubled.

Yelp Inc. (NYSE: YELP) reported second quarter net revenue of $375.5 million on August 6, 2026, an increase of 1% from the $370.4 million recorded a year earlier and $8 million above the top of the range the company had guided to. Net income fell 28% to $31.7 million. Adjusted EBITDA declined 9% to $91.4 million. Both profit measures came in above the company's own outlook, and both were lower than the prior year.

The single most consequential disclosure for shareholders was not in the press release. It surfaced on page 11 of the shareholder letter and again in the chief financial officer's prepared remarks: the buyback is off. According to the shareholder letter, Yelp repurchased $50 million of stock during the second quarter at an average price of $24.92 per share, then bought roughly $25 million more in the third quarter, bringing repurchases for the year to approximately $200 million. With $339 million left under the authorisation as of July 31, 2026, the company stopped.

"With $339 million remaining under our repurchase authorization as of July 31, 2026, we have now paused our program as we work to pay down our revolving credit facility," the letter states. "We expect to resume repurchases in 2027."

That single sentence reframes the quarter. Yelp entered 2026 with $216.1 million in cash and $103.3 million in short-term marketable securities. It ended June with $94.1 million in cash, no marketable securities, and $100 million outstanding on a revolver that carried no balance at the end of 2025. The cash went into two places: $263.6 million net of cash received for the acquisition of Hatchify Inc., and $174 million of stock repurchased in the first six months of the year.

The quarter in numbers

Advertising revenue, still 91% of the total, came in at $342.5 million for the three months ended June 30, 2026, down 3% year over year. Within that, the two category buckets moved in opposite directions and at different magnitudes.

Services advertising revenue was $240.96 million, effectively unchanged from $240.80 million a year earlier. Restaurants, Retail and Other advertising revenue, which Yelp abbreviates as RR&O, fell to $101.53 million from $112.90 million, a decline of 10%.

Other revenue reached $33.02 million, up 98% from $16.70 million. That is the fastest growth rate on any line Yelp reports, and it lifted the non-advertising share of net revenue to 8.8% from 4.5% a year earlier.

Total costs and expenses were $331.79 million, up 5% year over year and rising faster than revenue for the second consecutive quarter. Income from operations dropped 18% to $43.73 million, compressing the operating margin to 11.6% from 14.4%.

Other income, net fell 87% to $733,000 from $5.70 million. According to the shareholder letter, the decrease reflected lower interest income on smaller average cash and marketable securities balances at lower rates, compounded by interest expense on the credit facility borrowings. That $4.96 million swing accounts for roughly 40% of the $12.4 million decline in net income, a point that is easy to miss when reading the operating lines in isolation.

The provision for income taxes was $12.81 million against pre-tax income of $44.46 million, an effective rate of 28.8%. A year earlier the rate was 25.3%. According to the letter, the increase was driven by a higher discrete tax expense related to stock-based compensation, partially offsetting the effect of lower pre-tax income.

Diluted earnings per share came to $0.57, down from $0.67. The decline in net income was cushioned by a 15% reduction in weighted-average diluted shares outstanding, which fell to 55.7 million from 65.7 million. Without that buyback effect, the per-share decline would have been considerably steeper.

Six-month view

The half-year figures strip out some of the quarter-to-quarter noise. Net revenue for the six months ended June 30, 2026 was $736.98 million against $728.93 million, growth of 1.1%. Advertising revenue for the period fell 3.0% to $674.98 million. Other revenue rose 86.6% to $61.99 million.

Net income for the half was $49.39 million against $68.48 million, a 27.9% decline. Adjusted EBITDA was $170.78 million against $185.43 million, down 7.9%, with margin at 23% versus 25%.

Free cash flow, which Yelp defines as operating cash flow less purchases of property, equipment and software, was $61.07 million in the second quarter against $45.01 million a year earlier, an increase of 36%. For the six months, however, free cash flow fell to $106.23 million from $132.47 million, a decline of 20%. The quarterly improvement and the half-year deterioration both trace to working capital timing rather than to any structural change in the business.

Services: flat is the new growth

For most of the past three years, Services advertising was the line that carried Yelp. In February 2025 the company reported that Services had delivered its fifteenth consecutive quarter of double-digit year-over-year growth, and Yelp's full-year 2024 results showed Services advertising revenue of $879 million, up 11%.

That engine has now stalled. Services revenue of $240.96 million in the second quarter of 2026 is 0.1% above the second quarter of 2025. It is also below the $243.81 million recorded in the third quarter of 2025, meaning Services advertising revenue has not set a new quarterly record in a year.

The composition underneath the flat line is worth reading closely. According to the shareholder letter, Services ad clicks increased slightly year over year, paying advertising locations in the category were flat, and average revenue per location was flat. Because click volume rose while revenue did not, average cost-per-click in Services fell.

That is a meaningful signal for anyone buying local search inventory. Yelp is describing an auction in which supply of qualified consumer intent grew, demand from advertisers held steady, and the clearing price consequently declined. It is the mirror image of the dynamic Yelp reported through 2025, when click volume shrank and CPCs rose by 9% to 14% year over year.

The company attributes the increased click supply to consumer-side product work. "On the consumer side of our business, we saw encouraging traffic trends in the quarter, with improvements in app installs and page views," chief executive Jeremy Stoppelman said on the earnings call.

RR&O: four straight years of contraction

The Restaurants, Retail and Other category tells a longer and less ambiguous story. The quarterly series Yelp publishes in its shareholder letter runs back to the second quarter of 2023, when RR&O advertising revenue was $121.70 million. Every subsequent second quarter has been lower: $118.38 million in 2024, $112.90 million in 2025, $101.53 million in 2026.

Against the second quarter of 2023, RR&O advertising revenue is down 16.6% in three years, without adjusting for inflation.

The second quarter of 2026 figure does represent a sequential improvement. RR&O revenue rose 2.9% from the $98.70 million booked in the first quarter of 2026, which was the lowest quarterly figure in the published series. According to the shareholder letter, ad clicks, average CPC and paying advertising locations in RR&O all improved compared with the first quarter.

Yelp gives two causes for the year-over-year decline. The first is macroeconomic. The second is competitive, and the letter names it directly: "competitive pressures from food ordering and delivery providers."

That phrasing has appeared in Yelp's disclosures before. When Yelp reported third quarter 2025 results, it cited the same combination of macroeconomic pressure and competition from food delivery platforms for a 2% RR&O decline. The deterioration from a 2% decline to a 10% decline over three quarters suggests the competitive component has grown, not receded.

The competitive context is documented. DoorDash repositioned its advertising unit as a commerce media platform on June 4, 2026, reporting more than 400,000 active advertisers across DoorDash, Wolt and Deliveroo, a base that includes the long tail of independent restaurants running self-serve campaigns. Uber Advertising introduced Deal Drops and Reorder Rewards for restaurant advertisers four days later. Deliveroo cut local rate cards by 20% and opened programmatic inventory across four markets in early August 2026.

Every one of those platforms is competing for the same restaurant marketing budget that funds Yelp's RR&O line, and each offers something Yelp historically has not: a direct attribution path from ad impression to completed order.

The metrics that price the auction

Yelp publishes three operational series that determine advertising revenue: paying advertising locations, ad clicks and average cost-per-click. All three moved in the second quarter, and the direction of each carries different implications for media buyers.

Paying advertising locations totalled 510,000 on a monthly average basis, down 1% from 515,000. Within that, Services locations were 259,000 against 260,000, and RR&O locations were 251,000 against 255,000.

The year-over-year comparison understates what happened. Sequentially, total paying advertising locations rose from 485,000 in the first quarter of 2026 to 510,000, an increase of 25,000 locations or 5.2%. RR&O locations rose from 235,000 to 251,000, a gain of 16,000 or 6.8%, and the highest RR&O figure since the third quarter of 2025.

Chief operating officer Jed Nachman addressed the reversal on the call. "On restaurants, we saw kind of the best performance that we've seen from a PAL perspective in a few years," he said. "On the services side, we stayed flat on the PAL side. Overall, PALs were down 1% year-over-year, which is a big improvement over Q1."

He then qualified it. "We don't expect any massive trajectory turn in terms of the broader local economy, but it is certainly encouraging to see our traffic trends, as well as the Paying Advertiser Location trends."

Ad clicks declined 5% year over year. That is the shallowest decline in six quarters. The series since the start of 2025 reads: minus 3%, minus 7%, minus 11%, minus 8%, minus 10%, minus 5%. According to the letter, the decline was driven by fewer clicks in RR&O categories and partially offset by a slight increase in Services.

Average cost-per-click rose 1%. That is the smallest increase since the fourth quarter of 2024, when CPC was flat. The prior five quarters recorded increases of 9%, 11%, 14%, 6% and 8%.

Yelp attributes the 1% figure to mix rather than to pricing power. Services clicks generally carry higher CPCs than RR&O clicks, and Services clicks represented a larger share of the total than a year earlier. Strip out that mix effect and the underlying picture is of declining prices in Services and declining prices in RR&O, with the blended average held marginally positive by the shift in composition.

For agencies buying Yelp inventory, the practical read is that the platform got cheaper per click in both categories during the quarter while delivering slightly fewer clicks overall. Whether cheaper clicks convert at the same rate is not disclosed.

Average revenue per location, which Yelp defines as category advertising revenue divided by category paying advertising locations, decreased slightly overall. Applying the reported figures, RR&O revenue per location fell from roughly $443 per quarter to roughly $405, a decline of 8.6%. Services revenue per location was approximately $930, effectively unchanged. The company describes both Services metrics as flat.

Other revenue and the $250 million target

Other revenue is the line Yelp wants investors watching. At $33.02 million it grew 98% year over year, accelerating from the 75% growth recorded in the first quarter of 2026, when the line reached $28.97 million.

According to the shareholder letter, the increase came from three sources: the addition of revenue from Hatch, higher revenue from the Yelp Places API programme, and growth in food ordering through the partnership with DoorDash. The call description was slightly different, citing Hatch, data licensing and food orders. The Places API and data licensing overlap but are not identical categories, and Yelp does not break out the individual components.

Annualised, second quarter other revenue represents a $132 million run rate. The shareholder letter sets that against a stated target of $250 million by the end of 2028.

"We are targeting an annual run rate of $250 million in other revenue by the end of 2028," chief financial officer David Schwarzbach said on the call.

Reaching $250 million from $132 million over roughly ten quarters requires compound growth of approximately 29% a year. That is a materially slower pace than the 98% the line just delivered, which means the target is achievable on the current trajectory but does not depend on it continuing. It also means that if other revenue hits $250 million and advertising revenue merely holds flat at the current annualised $1.37 billion, total revenue in 2028 would be around $1.62 billion, roughly 11% above the 2026 midpoint guidance over two and a half years.

That arithmetic frames the strategic question the results pose. Other revenue is growing quickly from a small base and carries different unit economics. Advertising revenue is declining slowly from a large base. The crossover point where the smaller line offsets the larger one is not close.

Hatch: what $270 million bought

Yelp agreed to acquire Hatchify Inc. in January 2026 for approximately $270 million in cash plus $30 million in employee retention payments, and closed the transaction on February 2, 2026. The cash flow statement shows $263.6 million paid net of cash received.

According to the shareholder letter, Hatch annual run rate revenue grew 59% year over year to $35 million in June 2026. On that basis Yelp paid roughly 7.7 times forward run-rate revenue for the business.

The company attaches an unusually explicit caveat to that growth figure. According to a footnote in the letter, Hatch revenue includes overage fees, and growth rates are calculated against pre-acquisition results drawn from Hatch's internal, unaudited financial records that were not audited or reviewed by an independent accounting firm. Yelp describes the year-over-year growth rates as approximate and states that they may not be directly comparable to current-period figures prepared under its own accounting policies.

That disclosure matters when reconciling the numbers. Other revenue rose $16.3 million year over year in the second quarter. A $35 million annual run rate implies roughly $8.75 million of quarterly revenue from Hatch, or about 54% of the increase. The remainder would come from data licensing and food ordering. Because the run-rate methodology differs from Yelp's revenue recognition, that split is an approximation rather than a disclosure.

The integration has been expensive. According to the letter, Yelp significantly increased the size of the Hatch team during the quarter to accelerate the product roadmap, and this "resulted in an adjustment period in the quarter," with trends improving in July. Acquisition and integration costs of $3.48 million were excluded from adjusted EBITDA in the quarter, and $7.90 million for the half year. The company expects to exclude approximately $3 million in the third quarter and approximately $12 million for the full year in accrued acquisition and integration-related payments for continuing Hatch employees.

Goodwill on the balance sheet rose to $354.71 million from $135.85 million at the end of 2025, an increase of $218.86 million. Intangibles net rose to $92.09 million from $49.04 million. Depreciation and amortisation expense increased 47% year over year to $18.12 million, which the letter attributes primarily to higher amortisation of intangible assets from the acquisition alongside higher capitalised website and internal-use software development costs. That amortisation is a permanent drag on GAAP net income that does not appear in adjusted EBITDA.

On the call, Schwarzbach described the state of the asset in blunt terms. "Really, when we acquired Hatch in February, it was a startup, and now we obviously want to scale operations there," he said. "We're bringing a lot of folks as well as process and metrics to the go-to-market side, and we were able to deliver quite a few product improvements throughout the course of the second quarter."

Asked about the structural margin difference between subscription revenue and advertising, he said: "In terms of over the longer term, the margin profile for this business we expect can be similar to other subscription businesses. In the near term, though, we definitely want to continue to invest, to scale and again, to capture market opportunity."

Yelp does not disclose Hatch's gross margin, its customer count, or its net revenue retention. The letter states that Hatch manages tens of millions of distinct leads representing billions of dollars of value for its customers per year, without defining the measurement basis for either figure.

According to the letter, product work shipped in the quarter included conversational analytics, outbound voice, enhancements to the Google Local Services Ads and ServiceTitan integrations, and optimisations to the core voice model. A test integration inside the Yelp for Business platform that routes interested service professionals to the Hatch sales team generated more than 1,000 new leads.

Yelp Host and the economics of a phone call

Yelp Host, the artificial-intelligence call answering product for restaurants, reached an annual run rate of 2.4 million calls handled in July 2026, more than triple the January level. That implies roughly 200,000 calls a month, and an implied January run rate below 800,000 calls annualised.

The product launched in October 2025 as part of a release that included Yelp Assistant expansion, Menu Vision and Yelp Receptionist. Yelp Host had handled more than 1 million cumulative calls by June 2026, growing at an average of 38% month over month since launch.

Three changes to the product were disclosed in the quarter. An integration with OpenTable lets consumers book, modify and cancel reservations automatically through Yelp Host. Food ordering functionality with full point-of-sale integration lets restaurants take phone orders for pickup without commission fees. And 16 new languages were added, with automatic detection of a caller's language.

The commission point is the commercially significant one, and Stoppelman made the argument on the call.

"Food ordering, we've recently added. That's great for restaurants. Obviously, it takes a lot of time to take down food orders, but also if they're taking in orders primarily through food delivery services, that can really add up a lot of fees and expenses in getting that food out," he said. "Whereas food orders coming in over the phone is just pickup, it comes without a commission, our restaurants really love that."

That positions Yelp Host directly against the delivery marketplaces whose advertising expansion is pressuring Yelp's own RR&O line. It is a coherent strategy and an admission of the same competitive dynamic the RR&O numbers already show.

Asked whether Yelp Host has driven cross-selling into the advertising product, Stoppelman was candid about the limits. "On the ad side, we do have a dedicated group to working with restaurant partners. They're selling Yelp Guest Manager, they're selling Yelp Ads, they're selling Yelp Host. We do see some positive synergies there. It's early days. I don't think it's really changed the game for us on the restaurant ad side."

The shareholder letter includes a testimonial from a named operator. "I was just blown away by people having these conversations with the AI. It's the voice, the inflection, the way Yelp Host can build on the information and extrapolate and can speak words you never typed into the backend. We've had guests walk up to us in the restaurant and compliment us on it," said John Garbiel, owner of Metropolitan Diner, according to the letter.

Yelp does not disclose Yelp Host pricing, subscriber count, or the revenue the product contributes. Calls handled is a usage metric, not a monetisation metric, and the gap between the two is where the $250 million other revenue target will be won or lost.

Yelp Assistant and the share of projects

The updated Yelp Assistant, rolled out across all categories earlier in 2026, generated approximately 10% of Request-a-Quote projects in the second quarter, according to the shareholder letter. Overall project growth was approximately 10% year over year.

Those two figures are frequently conflated and they measure different things. The first is the share of a funnel step attributable to a specific interface. The second is the growth rate of the funnel step itself. Yelp's own footnote defines projects as those created by users through a Request-a-Quote flow or through Yelp Assistant, and states that year-over-year changes in Request-a-Quote projects are rounded to the nearest multiple of 5%.

The rounding convention is worth noting. A reported 10% growth figure under that methodology could represent anything from 7.5% to 12.5%.

The letter describes qualitative gains: repeat usage, faster response rates, increased personalisation and more complete conversations. It does not provide engagement figures, conversation volumes, or a conversion rate from Assistant conversation to submitted project.

The April 2026 spring release moved transaction completion inside the Assistant interface, adding booking, ordering and scheduling through partners including Vagaro, Zocdoc and Calendly. According to the letter, further partner integrations are in progress that would enable users to take more actions through Yelp Assistant in both Services and RR&O categories.

Data licensing, OpenAI and the distribution question

The clearest strategic shift in the quarter sits in the third of Yelp's three stated initiatives: extending reach across the artificial-intelligence ecosystem through licensing.

Yelp licensed reviews, photos and business information to OpenAI for use inside ChatGPT on July 24, 2026. According to the shareholder letter, Yelp ratings and reviews recently began powering ChatGPT's local experience in relevant categories, and a Request-a-Quote integration with ChatGPT is described as coming soon.

Stoppelman characterised the arrangement on the call without overstating its current contribution.

"We're really excited about working with OpenAI, obviously. It has a big consumer footprint with ChatGPT. You can now see our integration starting to happen there. It's still kind of the first inning, but you can see ratings. It links back to Yelp. You can see review snippets that also can link back to Yelp. In fact, we just started to see rolling Request a Quote," he said.

He continued: "Services represents a huge majority of our revenue, so having that interconnectivity, I think, is pretty exciting. It's super early days in terms of traffic impact, but we see this as a potentially significant distribution channel over time."

Asked directly whether leads arriving through artificial-intelligence platforms convert differently from other performance channels, Stoppelman declined to claim a result. "I think it's quite early to really have a strong opinion on that. The way that it's being merchandised is just at the bare minimum, as you can see, I don't think that's the eventual steady state. I really haven't seen any data on conversion just because it's just been starting more recently."

That answer is more informative than it appears. Yelp has been licensing local data for years, to Apple, Amazon, DuckDuckGo, Bing and Yahoo, and it treats the OpenAI relationship as an extension of the same model rather than as a new category of arrangement. Stoppelman said as much: "We do have a history here in licensing our data and cooperating with other big players. For example, our data can be found in places like Amazon, DuckDuckGo, Bing, and Apple has been obviously a huge partner that we've worked with for a long time. When our content shows up there, I think it's a win-win."

The commercial terms of the OpenAI agreement are not disclosed. Neither is the revenue contribution. Data licensing sits inside the other revenue line alongside Hatch subscriptions and food ordering commissions, and Yelp does not break it out.

There is a structural tension in the position that the results do not resolve. Yelp is selling its content into an interface that answers queries without requiring a visit to Yelp, while simultaneously arguing that the interface will drive meaningful traffic back. Both can be true. Which effect dominates determines whether licensing revenue is incremental or cannibalistic, and Yelp does not yet have the data to say.

The citation study and its boundaries

To support the licensing argument, the shareholder letter cites a commissioned study finding that Yelp was the most organically cited home services discovery platform by large language models in the fourth quarter of 2025, receiving 3.4 times as many citations as the next closest platform.

The underlying figures, published in the letter, show Yelp with 512,700 citations against 149,700 for the Better Business Bureau, 145,600 for Angi, 56,000 for Thumbtack, 33,600 for HomeAdvisor and 10,300 for Nextdoor.

The methodology footnote defines the boundaries precisely, and they are narrower than the headline. The analysis measured citations across ChatGPT, Google Gemini, Perplexity and Google AI Mode during the fourth quarter of 2025. It covers the United States only. It counts links appearing in response to prompts regardless of whether the link was clicked. It excludes advertisements and native content produced by the artificial-intelligence search provider. And the comparison set is five home services discovery platforms, not the broader web.

That last constraint is the important one. Being the most-cited platform within a comparison set of six local directories is a different claim from being a dominant source in artificial-intelligence answers generally. Detailed analysis of the same dataset showed Google AI Mode accounted for approximately 340,721 of Yelp's citations, or about 66% of the total, with Perplexity contributing 146,196 and ChatGPT and Gemini together supplying just over 25,700.

In other words, Yelp's citation lead is concentrated in the two platforms where it performs strongly, and thinnest on ChatGPT, which is precisely the platform where the new licensing agreement applies. The agreement addresses the gap rather than extending the lead.

The study was commissioned by Yelp. It is not independent research, and the letter identifies it as commissioned.

Google traffic, core updates and AI Overviews

The most closely watched exchange on the call concerned search traffic. Ishan Patel of Raymond James asked whether Yelp could distinguish paid from organic tailwinds following Google's May and June updates, noting that internet peers have reported AI Overviews traffic headwinds.

Stoppelman's answer was direct. "Certainly, we've been monitoring folks as they talk about what's going on in the space with AI Overviews and traffic patterns. It does seem that we bucked the trend. I think our area has been different when it comes to AI Overviews. Our categories that we get a lot of traffic in are highly monetizable. While you do occasionally see AI Overviews, they're usually not right at the top. I think there's some element of Google trying to preserve its own revenue there that is a bit protective."

He added that Yelp's traffic is largely organic. "From a paid traffic standpoint, largely, we are organically driven. We do have some limited paid traffic that we try to optimize. We've talked about some of that with multi-location and so forth. It's relatively de minimis in terms of overall traffic impact."

Earlier in the call he had attributed the traffic improvement to four sources: search engine optimisation work, higher app downloads, gains from the existing partner network of non-artificial-intelligence licensees, and an algorithmic shift at Google.

"We also saw some positive changes on the algorithmic side for Google, where we've seen they've started to really lean into user-generated content in a big way, I think because of the onslaught of spam and low-quality content. That's benefited us as well," he said.

Two Google releases fall inside the second quarter and the weeks that followed. The May 2026 core update ran from May 21 to June 2, completing in under twelve days. The June 2026 spam update was released globally on June 24, 2026, at 09:03 Pacific time.

The claim that commercial query categories carry lighter AI Overviews coverage is consistent with the argument that Google preserves advertising inventory on monetisable queries, though Yelp offers no data to support it and Google has published none. The transcript of the call renders the phrase as "May and June court updates," which is a transcription error for core updates. Yelp did not quantify the traffic gain in either percentage or absolute terms.

The broader publisher picture is less benign. Organic Google search traffic from United States users to USA Today's national paper fell 18% between June 2025 and June 2026, and Reddit and USA Today have weighed cutting Google's crawler off entirely as search referrals decline. Yelp's position, if it holds, is an exception rather than evidence that the trend has reversed.

Where the money went: expenses line by line

Total costs and expenses of $331.79 million rose 5% year over year while revenue rose 1%. The composition explains the margin compression.

Cost of revenue was $41.78 million, up 18% and the fastest-growing expense line. According to the letter, the increase came from higher website infrastructure expenses tied to maintaining and improving infrastructure, including the integration of artificial-intelligence products, plus additional infrastructure expense from the Hatch acquisition. As a share of revenue, cost of revenue rose to 11% from 10%.

That is the compute bill. Yelp is running conversational interfaces, a voice agent handling 200,000 calls a month, and language model inference across review analysis and search. Those workloads cost money per query in a way that serving a static business page does not.

Sales and marketing was $153.56 million, up 6%, and rose to 41% of revenue from 39%. The letter attributes the increase to investments in acquiring Services projects, increased business owner marketing spend, and higher workplace operating costs, the last partly because a rent abatement received in the prior-year period did not recur. Employee-related costs in sales and marketing were relatively flat, with Hatch headcount offset by lower average headcount elsewhere.

The Services project acquisition spend is the line that connects to the paid search strategy discussed below. It is a customer acquisition cost that Yelp is choosing to incur to feed its Services marketplace.

Product development was $70.47 million, down 10%, and fell to 19% of revenue from 21%. This is the only major expense line that decreased. According to the letter, the decline came from lower employee-related costs on lower average headcount, a higher proportion of employee work directed toward infrastructure enhancements that shifts costs into cost of revenue, more employee costs being capitalised, and lower cost of labour, partially offset by Hatch headcount.

Three of those four drivers are reclassifications or capitalisation rather than genuine cost reduction. Work moved from product development into cost of revenue, and costs moved from the income statement onto the balance sheet, where they emerge later as amortisation. Property, equipment and software net rose to $99.18 million from $91.69 million, and purchases of property, equipment and software were $27.41 million for the half year against $23.56 million.

General and administrative was $47.87 million, up 3%, holding at 13% of revenue. The increase came from Hatch acquisition and integration costs and higher legal costs, partially offset by a lower provision for credit losses on reduced customer delinquencies. That last item is a small positive signal about the health of Yelp's advertiser base: fewer local businesses are failing to pay their bills. The provision for credit losses fell to $18.01 million for the half year from $22.56 million.

Depreciation and amortisation was $18.12 million, up 47%, and rose to 5% of revenue from 3%.

Adjusted EBITDA excludes depreciation and amortisation, stock-based compensation, indemnifiable expenses and acquisition and integration costs. Adding back $18.12 million of D&A, $25.75 million of stock-based compensation, $352,000 of indemnifiable expenses and $3.48 million of acquisition costs to net income, taxes and other income produces the reported $91.43 million. Roughly half the gap between GAAP operating income and adjusted EBITDA is now non-cash charges tied to acquisitions and capitalised software rather than to equity compensation.

Stock-based compensation and the dilution arithmetic

Stock-based compensation fell to $25.75 million from $34.78 million, a decline of 26%. As a percentage of revenue it dropped to 6.9% from 9.4%, a difference of 2.5 percentage points that Yelp reports as a three-point improvement to 7% after rounding.

The largest reduction was in product development, where stock-based compensation fell to $10.67 million from $17.85 million, a 40% decline. Sales and marketing fell to $6.08 million from $7.30 million. General and administrative fell to $7.75 million from $8.56 million. Cost of revenue rose slightly to $1.24 million from $1.07 million.

Yelp reiterated the target of reducing stock-based compensation below 6% of revenue by the end of 2027, from 11% in 2024 and 9% in 2025. The company's outlook table guides to approximately 6% in the third quarter of 2026 and approximately 7% for the full year.

The dilution effect is the point. Weighted-average diluted shares fell 15% year over year to 55.73 million. Combined with the reduction in equity compensation, that produces the mechanism the letter describes: "We expect that the impact of our efforts to reduce SBC, coupled with continued share repurchases, will stack over time and benefit GAAP profitability in the years to come, particularly earnings per share."

The buyback pause interrupts one half of that stacking mechanism for at least the remainder of 2026 and possibly longer.

The balance sheet and the revolver

The balance sheet is where the quarter's strategic constraints become visible.

Cash and cash equivalents fell to $94.14 million at June 30, 2026 from $216.06 million at December 31, 2025. Short-term marketable securities went to zero from $103.29 million. Combined liquid assets therefore fell from $319.35 million to $94.14 million, a reduction of 71% in six months.

Against that, the revolving credit facility carried $100 million at the end of June, having been drawn to $165 million and repaid by $65 million during the half year. Total liabilities rose to $346.04 million from $247.63 million. Total stockholders' equity fell to $644.53 million from $710.85 million, and accumulated deficit widened to $1.42 billion from $1.29 billion, reflecting buybacks charged against equity in excess of net income earned.

The uses of cash in the first half are itemised in the cash flow statement. Operating activities generated $133.64 million. Investing activities consumed $183.27 million, of which $263.60 million was the Hatch acquisition net of cash received, offset by $109.29 million of marketable securities sales and maturities. Financing activities consumed $72.21 million, comprising $174.00 million of share repurchases and $17.57 million of taxes on net share settlement of equity awards, against $165.00 million drawn on the revolver less $65.00 million repaid and $19.48 million from employee stock plans.

Put plainly, Yelp funded a $270 million acquisition and $174 million of buybacks in a half year that generated $133.64 million of operating cash flow. The gap was covered by liquidating the securities portfolio, running down cash and drawing on the revolver. The buyback pause is the arithmetic consequence.

Capital allocation, restated

The letter sets out a three-part framework: invest in future growth, pursue strategic acquisitions through a combination of cash and financing, and return more than 50% of free cash flow to shareholders each year through share repurchases.

The second element is new in emphasis. Yelp acquired RepairPal for approximately $80 million in November 2024 and Hatch for approximately $270 million in early 2026, and it is now explicitly signalling that future acquisitions may be financed rather than funded from cash. The revolver draw was the first practical demonstration.

The third element carries a threshold that the current pause does not obviously breach. Free cash flow for the first half was $106.23 million; repurchases were $174.00 million, or 164% of half-year free cash flow. Full-year repurchases of approximately $200 million against a plausible full-year free cash flow figure would still comfortably exceed 50%. The pause therefore reflects balance sheet repair rather than a retreat from the policy.

The capital allocation chart in the shareholder letter records cumulative repurchase authorisations of approximately $2.5 billion across three periods: July 2017 to November 2022, February 2024 and February 2026. Of that cumulative total, $339 million remained available as of July 31, 2026.

Guidance implies a second-half decline

The outlook is where the quarter's optimism meets its constraints.

For the third quarter of 2026, Yelp guided to net revenue of $365 million to $370 million and adjusted EBITDA of $70 million to $75 million. For the full year, it narrowed net revenue to $1.460 billion to $1.470 billion, from a prior range of $1.455 billion to $1.475 billion, and narrowed adjusted EBITDA to $315 million to $325 million from $310 million to $330 million.

The third quarter guide is the number to read carefully. Yelp reported net revenue of $376 million in the third quarter of 2025. Guidance of $365 million to $370 million therefore implies a year-over-year decline of between 1.6% and 2.9%, with the midpoint around minus 2.3%.

The full-year figure carries the same implication. First-half 2026 net revenue was $736.98 million against $728.93 million in the first half of 2025, growth of 1.1%. Full-year 2025 net revenue was $1.465 billion, which places second-half 2025 revenue at approximately $736.1 million. At the midpoint of full-year 2026 guidance, second-half 2026 revenue would be approximately $728.0 million, a decline of about 1.1%. Even at the top of the range, second-half revenue would be roughly flat to slightly down.

Adjusted EBITDA guidance implies a sharper second-half compression. First-half adjusted EBITDA was $170.78 million. Full-year guidance at the $320 million midpoint leaves approximately $149.2 million for the second half, against $183.6 million in the second half of 2025, a decline of roughly 19%.

The company is explicit about why. "We expect expenses will increase sequentially in the third quarter as we invest in our AI transformation, Hatch and consumer marketing," the letter states. The outlook table guides depreciation and amortisation to approximately 5% of revenue in both the third quarter and the full year.

Nitin Bansal of Bank of America pressed the point on the call, framing his question around the fact that Yelp "lowered the upper end of the revenue guide by $5 million." The narrowing raised the floor by $5 million and cut the ceiling by the same amount.

One operational change deserves separate attention because it inverts the usual direction of spend for a publisher.

Yelp is buying leads through paid search and routing them to multi-location Services advertisers. According to the letter, the company directed leads acquired via paid search to multi-location Services businesses, and has already seen substantial budget growth from many of these advertisers. The associated spend appears in sales and marketing under investments in acquiring Services projects.

Bansal asked what had changed relative to earlier attempts. Nachman answered.

"We have certainly experimented in the past on buying leads for the overall ecosystem. I think the difference now is that we're specifically targeting multi-location services businesses. We've talked about that opportunity for a while now. They're obviously able to commit a lot of budget, if you're able to drive leads. Our ability to go out and source quality leads from the broader ecosystem just enforces and allows us to kind of get through more budget," he said.

He continued: "Ultimately, these businesses are looking for scale. They want the ability to buy a ton of leads that make a difference. We've been very targeted in the way that we go out and do that and are really confident on the ROAS of that spend for our customers."

Structurally, Yelp is operating as an arbitrage layer: buying intent in one auction, packaging it as leads, and reselling it into its own marketplace at a margin. The model works only while the spread between acquisition cost and lead price holds, and Yelp does not disclose either side of that spread or the volume involved.

The strategy also runs into a scheduled change in the underlying auction. Google will retire the standalone Local Services Ads dashboard and migrate campaigns into Google Ads as a pay-per-lead Performance Max campaign type, with the first phase beginning in August 2026 for selected home and storefront service advertisers in the United States. Manual bidding ends and historical performance reports do not transfer. For a company buying leads in that auction on behalf of multi-location advertisers, the loss of manual bid control is a live operational variable.

Yelp also implemented whole-page optimisation for Services categories on its mobile website during the quarter, which the letter says increased Services ad clicks from that platform.

Market reaction

Yelp shares closed the regular session on August 6, 2026 at $25.04, down 2.53%, then rose 1.96% in after-hours trading to $25.53, according to the earnings call transcript published by Investing.com. The prior close had been $25.69, meaning the after-hours price remained below where the stock had traded before the results.

The transcript records a consensus revenue forecast of $366.7 million against reported net revenue of $376 million, a beat of approximately 2.5%. The stock's 52-week range is given as $19.60 to $34.64.

Source inconsistencies worth flagging

Three discrepancies separate the shareholder letter from the published earnings call transcript, and each affects a figure a reader might otherwise cite.

The transcript quotes Schwarzbach saying Yelp "repurchased $15 million worth of shares at an average price of $24.92 per share" in the second quarter. The shareholder letter states $50 million at the same average price, and the highlights page states 2.0 million shares at an aggregate cost of $50 million. Two million shares at $24.92 equals approximately $49.8 million. The cash flow statement shows $174 million of repurchases for the half year, and the letter's cumulative figure of approximately $200 million including third-quarter activity is consistent with $50 million rather than $15 million. The letter figure is the reliable one.

The transcript renders Yelp Host's scale as "a $2.4 million call run rate." The letter states an annual run rate of 2.4 million calls handled. The transcript inserts a dollar sign that changes a volume metric into an implied revenue metric.

The transcript refers to Google's "May and June court updates." The context makes clear these are core updates.

The transcript carries a disclosure stating that the article "was generated with the support of AI and reviewed by an editor," which is a plausible explanation for the transcription errors. Where the letter and the transcript conflict on a number, the letter has been used.

Why this matters for the marketing community

The results describe a local advertising market that is not recovering, and a platform response that has moved from optimising the auction to replacing it.

For media buyers, the pricing signal is the most immediately usable disclosure. Average cost-per-click rose only 1% year over year after five quarters of increases between 6% and 14%, and the underlying movement in both Services and RR&O was downward. Click supply in Services increased. Paying advertising locations rose 5.2% sequentially, with RR&O locations up 6.8%, the strongest quarterly move in that category in years. Cheaper clicks and a broader advertiser base in the same quarter suggest an auction that has loosened rather than tightened.

For restaurant marketers specifically, the 10% RR&O decline sits inside a documented squeeze. Sponsored listings in Google's local pack rose from under 3% to roughly 22% of tracked mobile keywords between November 2025 and January 2026, a 733% increase measured across 1,200 mobile ranking reports. Google Maps has become a Demand Gen placement. Delivery marketplaces have converted their merchant bases into advertising businesses, with DoorDash reporting more than 400,000 active advertisers. Every one of those developments adds an auction competing for the same local budget.

For services advertisers, the operational change is the migration of Local Services Ads into Google Ads beginning in August 2026, which removes manual bidding and ends historical reporting continuity. Yelp is simultaneously buying leads in that ecosystem and selling lead management software into it through Hatch, whose product set includes a Google Local Services Ads integration.

For publishers and platform owners weighing licensing agreements with model providers, Yelp's disclosure is instructive precisely because of what it does not contain. The company has an agreement with OpenAI, content live inside ChatGPT, and a chief executive who describes the arrangement as potentially significant distribution. It has no conversion data, no traffic attribution and no disclosed commercial terms. That is the state of the art for one of the most-cited local content sources in artificial-intelligence answers, four months after the citation study and two weeks after the licensing agreement.

The consumer research Yelp has published complicates the picture further. Its April 2026 Morning Consult survey of 2,202 United States adults found that 65% had used an artificial-intelligence search tool in the prior six months, but only 15% said they trust the information those platforms provide "a lot." Follow-up research published in July 2026 found that 43% of Gen Z respondents trust artificial-intelligence summaries for restaurant decisions. A platform whose commercial argument rests on the trustworthiness of human-generated content is licensing that content into interfaces its own research says consumers do not fully trust.

The financial framing is the plainest read. Yelp is spending to build revenue streams that are growing quickly from a base of $132 million annualised, while an advertising business running at roughly $1.37 billion annualised declines slowly. The company has committed to the transition through the balance sheet, borrowing against a revolver to fund an acquisition and pausing buybacks to repay it. Adjusted EBITDA margin guidance of roughly 21% to 22% for 2026, against 25% in each of the prior three years, is the price of that transition stated in advance.

Whether the transition works will be visible in a specific number rather than in narrative. Other revenue needs to compound at roughly 29% annually for ten quarters to hit the stated target. Advertising revenue needs to stop falling. Neither condition was met in the second quarter of 2026, though the first was comfortably exceeded and the second showed its first sequential improvement in a year.

Timeline

Summary

Who: Yelp Inc. (NYSE: YELP), the San Francisco local discovery platform founded in 2004, reported the results. Co-founder and chief executive Jeremy Stoppelman, chief financial officer David Schwarzbach and chief operating officer Jed Nachman presented on the earnings call, with Kate Krieger, director of investor relations, hosting. Analysts asking questions were Alex on behalf of Eric Sheridan at Goldman Sachs, Sergio Segura at KeyBank, Zach on behalf of Colin Sebastian at Baird, Nitin Bansal at Bank of America, and Ishan Patel on behalf of Josh Beck at Raymond James. The results affect local advertisers in services and restaurant categories, agencies buying local search inventory, and platforms weighing content licensing agreements with model providers.

What: Second quarter net revenue of $375.5 million, up 1% year over year and $8 million above the top of guidance. Net income of $31.7 million, down 28%. Adjusted EBITDA of $91.4 million, down 9% but $16 million above guidance. Services advertising flat at $241.0 million, Restaurants, Retail and Other advertising down 10% to $101.5 million, and other revenue up 98% to a record $33.0 million. Paying advertising locations down 1% year over year to 510,000 but up 5.2% sequentially. Ad clicks down 5%, average cost-per-click up 1%. The share repurchase programme is paused with $339 million remaining under authorisation while Yelp repays $100 million drawn on its revolving credit facility, with buybacks expected to resume in 2027.

When: Yelp released the results after market close on August 6, 2026, covering the quarter ended June 30, 2026, and hosted a webcast at 2:00 p.m. Pacific time the same day. The repurchase authorisation balance is stated as of July 31, 2026. Third quarter guidance is $365 million to $370 million of net revenue and $70 million to $75 million of adjusted EBITDA. Full-year 2026 guidance was narrowed to $1.460 billion to $1.470 billion of net revenue and $315 million to $325 million of adjusted EBITDA. The other revenue target of $250 million annual run rate is set for the end of 2028, and stock-based compensation below 6% of revenue for the end of 2027.

Where: Yelp operates primarily in the United States market. The commissioned citation study covering ChatGPT, Google Gemini, Perplexity and Google AI Mode is United States only. Yelp Host gained 16 additional languages during the quarter. The Local Services Ads migration that affects Yelp's lead acquisition begins with United States advertisers in August 2026.

Why: Yelp is reallocating capital from share repurchases toward debt reduction after funding a $270 million acquisition and $174 million of first-half buybacks from a balance sheet that generated $133.6 million of operating cash flow over the same period. The underlying commercial reason is that advertising revenue, 91% of the total, declined 3% year over year while the company is attempting to build non-advertising revenue streams through Hatch subscriptions, Yelp Host, food ordering and data licensing. Guidance implies second-half 2026 revenue below the prior year and adjusted EBITDA roughly 19% lower, reflecting planned investment in artificial-intelligence products, Hatch scaling and consumer marketing at a point where the local advertising market is not recovering.