Yelp's Local-Ad Engine Is Cracking. The AI Pivot Won't Save Q2.

I was three tabs deep in a comp spread — Yelp against the local-directory peers, the review-marketplace orphans, and a handful of restaurant-tech adjacents I keep on a watchlist — when the Q1 print landed Wednesday after the close. I had the model rebuilt by Thursday morning. By Thursday afternoon I’d run it three different ways. And I keep landing on the same uncomfortable read: Yelp is no longer a public-market story. It’s a take-private waiting for a sponsor to finish the math.

Let me be clear up front, because this column lives or dies on calling things straight: the take-private thesis is my editorial interpretation, not consensus, not reporting, not something a banker whispered to me at a bar. It’s a judgment based on what’s in the numbers, what’s in the buyback cadence, and what’s missing from the strategic narrative. Treat it as the opinion section of The Bottom Line. The facts I’m about to walk through are the facts. The conclusion is mine.

Here’s what Yelp told us on May 7. Net revenue of $361.5 million, up 1% year-over-year. Net income of $18 million, or $0.30 a share — down 27%. Adjusted EBITDA of $79.4 million, down 7%, a 22% margin. Services advertising revenue of $234 million, up 1%. Restaurants, Retail and Other (RR&O) advertising revenue of $99 million, down 11%. “Other revenue” — and we’ll get into what that actually contains — of $29 million, up 75%. Paying advertising locations of 485,000, down 6%. Ad clicks down 10%. The company repurchased 5.1 million shares for $125 million in the quarter. Full-year guidance was reaffirmed: revenue of $1.455 to $1.475 billion, adjusted EBITDA of $310 to $330 million.

Now let me tell you why those numbers, in that combination, set off the alarm I’m hearing.

The core engine is melting

The RR&O segment is the part of Yelp that most consumers think is Yelp. The restaurants. The retail storefronts. The dentists, the dry cleaners, the home services that aren’t already on the Services side. That segment did $99 million in Q1 — and it shrank 11% year-over-year. That isn’t a soft quarter. That’s the third consecutive quarter of negative growth on the surface area that, by my count, still represents close to a third of the ad business when you adjust for mix.

The line that mattered more was paying advertising locations: 485,000, down 6%. Six percent fewer SMBs are writing Yelp a check than were a year ago. And the clicks those advertisers paid for? Down 10%. So fewer customers, paying for less inferred attention. That is the exact combination — declining unit count, declining unit economics — that, in any other vertical, would be described as a structurally impaired ad surface.

The Street wants to call this cyclical. Restaurants are tight. Retail is mid-tier-grim. SMB budgets are under pressure from card fees, payroll, and the slow grind of margin compression. All true. Where I disagree with the Street: cyclical revenue declines don’t ship with permanent location declines. When advertisers leave a directory and the directory doesn’t pull them back in the next up-cycle, what you have is share loss, not seasonality. Yelp has been losing share to Google’s local pack, Instagram, TikTok geo-tagged content, and — for restaurants specifically — to the OpenTable/Resy/Toast triangle that now owns the discovery-to-booking handoff. None of those are getting weaker in Q2.

The Services side held up — $234 million, up 1%. That’s the dentist-and-plumber business, which Yelp re-architected over the last two years around request-a-quote, AI-assisted lead routing, and the Yelp Assistant product. It’s working. It’s not exciting. It’s a low-single-digit grower that funds the rest of the company. My base case: Services grows 2–4% for the year. RR&O declines 8–12% for the year. Other grows 50–70%. That gets you to the bottom of the guide on revenue and somewhere in the middle on EBITDA, which is exactly why the guide was reaffirmed and not raised.

The AI pivot, sized honestly

This is where I get unfashionable.

Yelp’s narrative for the last four quarters has leaned hard on what management calls “the AI-led re-platforming” — Yelp Assistant, the generative review summaries, the AI matching layer on the Services side, and the productization of Hatch (the AI-driven SMB communication suite they acquired in 2024). Stoppelman opened the call by anchoring on it. Schwarzbach reinforced it. Nachman, on the sales-org side, talked about it as a multi-year compounding tailwind for advertiser ROI.

Fine. I’m prepared to believe most of it. But let’s size it honestly.

The “Other revenue” line — which is where the AI-adjacent revenue actually shows up on the income statement — was $29 million in Q1. Up 75% year-over-year. Genuinely impressive growth rate. But $29 million is less than 10% of total revenue. And “Other revenue” is not a clean AI line. It contains, by Yelp’s own segment commentary: Hatch subscription revenue, third-party data licensing, and food ordering rev share. The data-licensing piece is the one with the AI-adjacent narrative — Yelp’s been quietly licensing its review corpus and structured local data to model providers — but it’s bundled. We do not know the inside-the-bundle split. I would not be surprised if Hatch is the biggest piece, with data licensing second, and food ordering rev share a steady but smaller stream.

So the headline “AI revenue up 75%” the Street is going to write this weekend is doing real work. It’s true on a directional basis. It is also describing a sub-10% segment that would need to roughly double again, while RR&O held flat, just to offset one more year of mid-single-digit RR&O decline. Where I disagree with the Street: the AI pivot is real, the growth rate is real, but the base is too small to bend the curve on a $1.4-billion company before late 2027 at the earliest. That’s not a knock on the strategy. It’s arithmetic.

If you want to see what an AI pivot actually rescuing a directory looks like, you’d want to see Other revenue at 20%+ of mix, growing 40%+, with RR&O stabilized at flat. Yelp is at <10%, growing 75%, with RR&O at -11%. The mix is moving the wrong direction faster than the new mix is filling in. (My earlier piece on Yelp’s AI stack walks through the product layer in more detail.)

What the buyback tells you

$125 million of share repurchases in Q1. 5.1 million shares. At a roughly $24–$25 average, by my back-of-envelope on the volume-weighted print.

Now, buybacks are a Rorschach test. Bulls will tell you it’s the board signaling conviction that the stock is undervalued. Bears will tell you it’s a management team that’s run out of growth investments worth making and is liquidating the float to defend EPS. I lean bear here, but with a more specific read: the cadence of this buyback, combined with the reaffirmed (not raised) guide, looks to me like a balance-sheet preparation, not a return-of-capital story.

Here’s what I mean. If you’re a board that thinks the stock is dramatically undervalued and you think the operational story is about to inflect, you accelerate the buyback and you raise the guide. Yelp did the first half of that and not the second. If you’re a board that thinks the stock is undervalued and the operational story is going sideways, you keep the buyback steady to defend per-share metrics and you give yourself optionality on either (a) holding the public-company flag through a multi-year re-platforming, or (b) being approached.

The float is now smaller. The cash balance is still healthy. The EBITDA-to-enterprise-value math is, to a private buyer, more attractive every quarter that buybacks continue and the multiple doesn’t recover. That is, structurally, what setting up an LBO target looks like — whether or not anyone inside the building is thinking about it that way.

Who would buy this, and at what level

I’m going to keep this categorical. I don’t have buyer names and I’m not going to invent them. But here are the buyer profiles I’d run an LBO model against if I were on the sponsor side this weekend:

A private-equity rollup focused on local-commerce infrastructure. The thesis here is simple: Yelp’s traffic is still real, the review corpus is a durable moat that improves with age, and the cost structure has room. A sponsor takes it private, accelerates the cuts on the consumer side, sells off or carves out non-core data assets, and either re-IPOs in five years or merges into a larger local-services platform. The number that gets this done is probably in the $32–$38 range — a 30–55% premium to where it’s trading post-print, an EV that puts you somewhere in the 7–9x forward EBITDA zone. That’s expensive for a declining ad business and cheap for a moated data asset, which is exactly the spread that makes deals happen.

A strategic with a directory ambition. This is the harder profile to draw because most of the obvious strategics already have their own local-commerce surfaces. But there’s a category of platform companies — particularly on the payments and SMB-software side — that have been wanting a discovery layer for years and have never built one well. Yelp would be a forward-integration play: own the top of the funnel, route into your own checkout, monetize the SMB on subscription instead of CPC. The math here is fuzzier because the strategic value of the directory varies wildly by buyer, but I could see a strategic stretching to the $40s if the synergy story checks out.

A consortium with a data-licensing thesis. This one’s narrower. If you believe the model-training data market becomes more important and more contractualized over the next 24 months, Yelp’s structured local review corpus is one of the more attractive licensable datasets in existence. A consortium that takes Yelp private to monetize the data asset more aggressively, while letting the ad business decline gracefully as a cash-flow ballast, is the most contrarian read. It’s also the most interesting one. I’d put a non-trivial probability on it, but I wouldn’t bet the model.

Across all three, my base case take-out range is $32–$40, with a midpoint around $36. That’s a deal that prints in late Q3 or Q4 if it prints at all. (The broader restaurant-tech M&A roundup I ran in March is the relevant comp set for sponsor appetite in this category.)

The Q4 scenario

Let me walk through what I think is the most likely sequence between now and year-end, with the caveat that I’m describing a base case, not a prediction:

Q2 prints in early August. RR&O decline accelerates modestly — call it -12 to -13%. Locations cross below 480k. Ad clicks down another 10%+. Other revenue grows another 60–80%. The headline is “AI revenue accelerating.” The under-the-hood story is “core business decelerating faster than AI business is filling in.” The stock fades 5–10% on the print and another 5% over the following two weeks as analysts re-cut their RR&O assumptions.

Q3 prints in early November. The full-year guide gets trimmed on revenue — probably to the $1.43–$1.45B zone — while EBITDA is held in the original range through opex discipline (read: layoffs, which I expect to be announced in late Q3 in a “re-org around AI” framing). Buyback continues. Stock trades into the high teens, possibly a 16-handle, on a forward EBITDA multiple that’s now in the 4–5x zone.

Late Q4: that’s the window. A 4–5x forward EBITDA multiple on a moated-but-declining ad business with a real (if small) AI growth engine, a clean balance sheet, and a buyback program that’s chewed through 10%+ of the float over 12 months — that’s an LBO model that prints. My base case probability on a take-private announcement before year-end 2026: 35–45%. Not a sure thing. Not even a majority. But materially higher than the Street is pricing.

The risk to this view runs in two directions. To the upside: the AI pivot accelerates faster than I’m modeling, RR&O bottoms in Q2 instead of grinding through Q4, and the stock re-rates on a “growth-restored” narrative. I’d put that at 20–25%. To the downside, the business simply melts and the take-private window doesn’t open because no sponsor wants to catch a falling knife in local advertising — which is the scenario where shareholders fare worst of all. I’d put that at 30–35%. The remaining probability is the boring base case where it muddles through, gets neither bought nor saved, and trades sideways in the $22–$28 range while the buyback slowly drains the float.

My base case isn’t that Yelp is doomed. It’s that the public-market vehicle is the wrong wrapper for what Yelp is becoming. A levered private holder can run the playbook — cut harder, license the data more aggressively, re-segment the sales org around Hatch and away from CPC — without the quarterly humiliation of explaining a -11% segment to people who don’t care that the -11% segment is being deliberately deprioritized.

The AI pivot, to be clear, is the right strategic move. I’m not arguing the strategy. I’m arguing the venue.

What I’d watch between now and August

Three things, in this order.

First, the Q2 segment commentary on Hatch specifically. If Yelp starts breaking out Hatch ARR — or even hints at the magnitude on the call — that’s the strongest signal that they’re preparing to tell a SaaS-shaped story to a SaaS-shaped buyer. Right now Hatch is buried in “Other revenue” with two other line items. If it gets disclosed standalone, pay attention.

Second, the cadence of the buyback. If Q2 repurchases run at $125M+ again, the float-shrink thesis is intact. If they slow to <$75M, the company is either preserving cash for an acquisition of its own or building dry powder for a balance-sheet event. Either is interesting.

Third, the language around “strategic alternatives” — which, to be clear, has not appeared in any of Yelp’s commentary, and I’m not suggesting it has. But if it appears in Q2 or Q3, even in the standard boilerplate, the market will sniff it instantly and the re-rating begins that day.

Until then, the Q1 print stands as the cleanest evidence yet that the public-company chapter is approaching its last act. (And if you want the broader argument for why the AI-premium narrative is doing too much work in this sector generally, that’s the piece I ran last month.)

The math doesn’t work at current multiples without an M&A floor. The buyback is building that floor a brick at a time. And somewhere between now and Q4, my model says the floor gets tested.

Don’t bet the farm on it. But don’t ignore it either.

— Marcus writes The Bottom Line. Tips: tips@tabletransfers.com.