Snap Fell Over 10% on a Half-Percent Revenue Miss — the Answer-Engine Print Was Reddit's, Six Days Earlier

The arithmetic of the miss is unusually clean. Consensus of $1.73 billion on 499 million expected daily users implies ARPU of about $3.47; Snap delivered $3.48. The shortfall traced to four million missing daily users rather than weaker ARPU — worth roughly $14 million of revenue at the ARPU actually delivered, partly offset by ARPU landing a cent above the implied consensus level.

Everything else beat. Adjusted EBITDA of $383 million came in 13.5% above the $337 million consensus, gross margin reached 62% from 55% a year earlier, and fourth-quarter guidance of $1.950 billion to $1.985 billion carried a midpoint above the $1.95 billion street number, with adjusted EBITDA guided to $490 million to $525 million against $383 million just delivered. This was a quarter that beat on profit and guided up.

The first hard evidence that answer engines were eating discovery did not come from Snap. It came from Alphabet on Oct. 28 and Reddit on Oct. 29.

Alphabet reported record search growth — year-over-year search growth at its highest level in five years, with AI Mode searches running three times longer than traditional ones — while stating that the growth in AI Mode searches came with lower monetization, that paid clicks rose 8% against an 11% average for the first nine months of the year, and that cost per click rose 4%. Search and other advertising revenue of $63.35 billion missed the $64.96 billion estimate, and the shares fell more than 10%. Then Reddit disclosed that Google-related search referrals declined meaningfully and accelerated into the fourth quarter, with October traffic down more than a million referrals a week on average, and warned daily uniques could fall by a low double-digit percentage from a 130 million base. Reddit shares fell 29%, its worst day ever.

Why would answer engines hit Snap first? Because what an assistant answers is a question, and a question is the moment an advertiser buys an impression to shape preference. The upper funnel is the only inventory where the buyer gets a click of attention and nothing else: no purchase, no install, no proof. Once a consumer can get the same comparison from an answer engine, that impression loses its value without losing its supply.

Management did not say that. Chief commercial officer Ronan Harris attributed the brand decline mainly to lower consumer-goods spending driven by macro pressures in the U.S. and the Middle East, and noted the second quarter benefited from easier comparisons and World Cup-related spending. That explanation is plausible and it is the proximate cause. The deeper question is why the brand budgets that exist are not coming back to Snap, and here the demand-side disclosures are blunt.

The chain behind Snap's $3.48 of ARPU runs through three gates. Upstream, the company runs the vast majority of its computing, storage and bandwidth on Alphabet's cloud unit and Amazon's AWS, and its own annual report warns that hosting costs will keep rising and that competitors can integrate search engines and artificial intelligence services into products they control. Alphabet's Gemini models on Vertex AI already power the in-app assistant. So the supplier of the compute and models is also the party taking the top of the funnel — a structural conflict that is usually priced as a risk, not a discount.

Downstream, the budget is set by consumer-goods advertisers. And the mix inside the audience decides which budget any given impression can reach.

The regional arithmetic is the part investors should keep in view. North America holds 19% of daily users but produces about 59% of revenue at $11.07 of ARPU; Rest of World holds 305 million of the 495 million daily users and monetizes them at $1.13. So Snap's ARPU line is a North American and subscription statistic wearing a global label, and the audience that AI assistants are most likely to intercept first is the audience with the least revenue per user attached.

The most consequential disclosure in the release was not a number. From Q1 2027, global monthly active users become the primary audience metric, global MAU-based ARPU becomes the primary monetization metric, and free cash flow per share joins the scorecard. Management said the reason plainly: it does not intend to optimize the product around daily use at the expense of long-term value.

That is a defensible product philosophy. It is also an awkward coincidence of timing. Daily-user growth went from 8% to 4%, monthly-user growth from 7% to 4%, and ARPU growth from 2% to 10% — the growth mix inverted in a single year — and then the metric that showed the inversion was retired. The filing also discloses that from Q1 2027 monthly users will be calculated on activity in only the final 30 days of a quarter, which it warns may not capture engagement across the full quarter. That is a narrower measurement window on the metric that replaces the broader one.

The honest framing is that the answer-engine question is not settled by this print, and Snap's own disclosure cannot settle it. The company publishes no brand-versus-direct-response revenue split in dollars, no referral-source mix for its audience, and no price-per-impression data. Those are exactly the disclosures that would let an outsider separate a discovery-budget reset from a normal cyclical quarter, and they are not disclosed.

What is disclosed does support one conclusion with reasonable confidence: the discovery budget was repriced, not taken, and Snap sits on the repriced side of that market while the beneficiaries — Meta on measurable outcomes and Reddit on ARPU — sit on the other. The evidence that AI answer engines compress discovery traffic is hard and dated within the last three weeks of the quarter; the evidence that they compressed Snap's brand budgets is management's own attribution to consumer-goods spending. Both are true, and only one of them is a durable structural change.