Perplexity's Sponsored Answers Are the First Real Test of AI Advertising. Here's What the Model Looks Like.

Perplexity launched a paid advertising product in 2025 that does something no programmatic platform has done before: it places brand content directly within an AI-generated answer, delivered at the moment a user submits a question with purchasing intent.

The product is called sponsored answers. It is early, it is expensive relative to traditional digital advertising, and it is the most important new advertising format in years.

How Sponsored Answers Work

When a Perplexity user submits a query that matches an advertiser's target intent — "best running shoes for marathon training," "which CRM is best for a 50-person sales team," "compare business travel credit cards" — Perplexity may insert a sponsored unit within the AI-generated response. The sponsored unit appears as a clearly labelled brand response adjacent to the organic AI answer.

The mechanic is different from search advertising in several important ways.

In search, the user sees a list of results and decides which to click. The ad is one result among many, competing for attention. The user brings their own judgment to the selection. In sponsored answers, the AI has already synthesized and presented a recommendation. The user is receiving an answer, not a list of options. The sponsored unit appears within that answer context — not as an alternative to the AI's recommendation, but alongside it.

This changes the intent signal. A search query is a statement of topic interest. A Perplexity query is typically a statement of decision intent — the user is further along in the purchase process and is using the AI to help make a choice. The CPM premium Perplexity charges for sponsored answers (reported to be 3-5x traditional search CPMs for comparable categories) reflects this signal quality.

The Intent Signal Advantage

The strongest argument for the sponsored answers model is the quality of the intent signal attached to each impression.

Traditional programmatic advertising matches ads to audiences and contexts. Audience targeting uses demographic and behavioural data to infer who is likely to be interested in a product. Contextual targeting uses the content of the page being viewed to infer intent. Both are proxies — approximations of the moment and context in which a user might be receptive to an advertising message.

An AI query is not a proxy. It is a direct statement from the user about what they are trying to decide, expressed in natural language, at the moment of decision. "What is the best project management software for a remote team of 20 people" is more specific, more contextual, and more decision-proximate than any audience segment or contextual signal available in the programmatic ecosystem.

The conversion rates reported by early Perplexity advertising partners reflect this. LLM-referred traffic converts at approximately 1.5x the rate of other referral channels, according to data from Criteo's pilot with Perplexity. The sample size is small and the category coverage limited, but the directional signal is consistent with the intent quality argument.

The Supply Problem

The limitation on the sponsored answers model at current scale is supply: Perplexity has tens of millions of monthly active users, which is significant but substantially smaller than Google Search, YouTube, or Meta's combined audience. Advertiser budgets that can run efficiently on Google or Meta cannot fully deploy on Perplexity without hitting reach constraints.

This supply problem is temporary. The trajectory of AI assistant adoption — driven by improved model quality, integration into operating systems and browsers, and growing consumer comfort with AI-generated answers — points toward AI answer interfaces absorbing a growing share of the queries that currently go to search. The supply will grow.

The strategic question for advertisers is whether to develop the capability to run sponsored answers campaigns now, when the format is new and the learning curve is manageable, or later, when the inventory has scaled and the auction is more competitive. Early movers in Google's auction had a structural advantage for years. The same dynamic may apply here.

What This Signals for the Broader Advertising Stack

Perplexity's sponsored answers model is the clearest signal yet that the AI answer layer will be monetised, that the monetisation model will resemble search more than display, and that intent signals from AI queries will be the premium inventory of the next advertising era.

Several implications follow for the programmatic ecosystem.

Search budgets are at risk before display budgets. If AI answer interfaces absorb search intent — the user who would have searched Google asks Perplexity instead — the search advertising category faces direct competition in a way that display does not. Google has recognised this and is incorporating AI-generated answers into Search (AI Overviews), effectively competing with Perplexity on the same format within its own platform.

Publishers face a structural monetisation problem. Sponsored answers generate revenue for Perplexity, not for the publishers whose content Perplexity retrieved to generate the answer. The publisher contributed to the AI answer that the advertiser is paying to be adjacent to, without receiving a share of that advertising revenue. This dynamic will not persist indefinitely — either through publisher licensing arrangements, new formats that drive traffic back to publishers, or regulatory pressure — but it is the current state.

The SSP layer needs an AI query signal. Supply-side platforms that can enrich bid requests with AI query intent signals — what LLMs are recommending, which brands appear in AI answers for relevant categories, which publishers are being cited — have a capability that the current programmatic auction cannot see. The intent enrichment thesis is that this signal, passed to DSPs at auction time, produces higher bids and better campaign performance for the advertisers who understand what they are looking at.

Perplexity's advertising model is not the endpoint of AI advertising. It is the first commercially viable proof of concept. The format, the intent signal, and the supply dynamics it has demonstrated will inform every AI advertising product built in the next three years.

The auction for AI-mediated purchasing decisions is being designed right now. Understanding it while the architecture is forming is worth significantly more than understanding it after the defaults are locked in.