Advertising has always run on attention. Agentic advertising or, content built for an AI agent to read, retrieve, and act on, rather than for a person to click, runs on something scarcer and much harder to buy: trust.
That’s not a branding platitude. It’s showing up in the data, and the data tells an uncomfortable story. Across 2025 and 2026, consumer trust in AI search moved in the opposite direction of AI search usage. More people are relying on these tools every month. More of them are growing skeptical of what those tools tell them, at exactly the same time. Understanding why that gap opened, what the numbers say about it, and how the biggest AI platforms are already responding differently is a real opportunity for agencies right now.
TL;DR
- Agentic advertising means content built for an AI agent to retrieve and potentially use in its answer, not content built for a human to click.
- Usage and trust are diverging. A Fractl and Search Engine Land survey of over 1,000 U.S. consumers found the share who call AI search more helpful than traditional search fell from 82% in 2025 to 54% in 2026, even as 70% of the same group said they use AI search more than a year earlier.
- Paid placement rarely buys citations. Independent analyses of AI-generated answers estimate that roughly 84% of what AI engines cite is earned media, while paid or advertorial content accounts for about 0.3%. Money mostly can’t buy its way into the answer, only credibility can.
- Platforms are drawing the line differently. OpenAI launched ads in ChatGPT in February 2026 but keeps them structurally separated from the generated answer. Perplexity walked away from advertising entirely that same month, for the same underlying reason. Google is blending ads further into its AI Overviews. There’s no shared industry standard yet.
- The practical response isn’t caution but a different kind of investment: earned authority, third-party corroboration, transparent sourcing, and content built to hold up under machine scrutiny, not just human reading.
What Is Agentic Advertising, and Why Does It Need Its Own Playbook?
Traditional digital advertising was built around a page of results. A person typed a query, saw a row of links plus a few clearly separated ads, and did their own weighing: which source looks credible, which snippet actually answers the question, which ad to ignore. Advertising sat next to the user’s judgment. It never replaced it.
An AI agent removes that step entirely. Instead of ten sources to compare, the user gets one synthesized answer, often with no clear signal of which sentence came from a publisher’s reporting, a brand’s content, or the model filling a gap on its own. The user isn’t filtering sources anymore. They’re trusting the agent to have already done that filtering honestly, on their behalf.
That’s the structural reason agentic advertising can’t just be traditional advertising wearing a chatbot costume. The channel changed what trust has to do, and changed who’s responsible for doing it.
Is Consumer Trust in AI Search Declining in 2026? What the Data Shows
This isn’t a hypothesis. It’s measurable, and the numbers are stark.
Fractl and Search Engine Land’s Q2 2026 survey of 1,008 U.S. consumers and 150 marketers repeated a study first run in 2025, asking the same questions a year apart. The result is a genuine reversal: the share of consumers who found AI search more helpful than traditional search dropped from 82% to 54% in a single year, while the share who rated it less helpful grew nearly sixfold, from 3% to 17%. Over the same period, 70% of respondents said they were using AI tools for search more than the year before, and just 3% said they were using them less. The full underlying data set is broken out on Fractl’s research page.
The same survey found the “AI use hurts brand trust” number climbing fast, from 20% of consumers in 2025 to 39% in 2026. The effect is strongest with younger consumers: 54% of Gen Z respondents said heavy AI use would lower their trust in a brand, compared with roughly a third of Gen X and Baby Boomer respondents. A separate WordPress VIP “Future of the Web” survey of 2,000 respondents, also covered by TechCrunch, found that 60% of U.S. consumers say the word “AI” in a brand’s own messaging is actively a turnoff, 86% don’t fully trust AI-generated content and still want to check the original source, and 42% rank unattributed AI answers as less trustworthy than airline fees or medical bills.
None of this means people are abandoning AI tools — quite the opposite. Klaviyo’s 2026 AI Consumer Trends Report found only 13% of consumers say they completely trust AI in general, with 36% saying they somewhat trust it. But 85% say they have at least some trust in AI for shopping recommendations specifically, and 39% report having bought a product with an AI system recommended in the past six months. Trust in the broad idea of “AI” is shaky. Trust in one specific, well-sourced recommendation is much higher. That distinction is the whole game for anyone building agentic campaigns: broad trust is nearly impossible to manufacture, but narrow, well-evidenced trust in a specific answer is achievable and it’s the thing agencies can actually influence.
Why Paid Placement Struggles to Buy What Trust Requires
Here’s the number that should reset how agencies think about budget allocation in this channel: independent analyses of what AI engines actually cite when generating answers put earned media at roughly 84% of citations, a brand’s own website at only 5–10%, and paid or advertorial content at approximately 0.3%.
Separately, research from Ahrefs and Semrush on which pages large language models actually cite found that most of those citations don’t come from pages ranking in Google’s own top 10 results for the same query. In ChatGPT’s case, only about 6–8% overlap with Google’s top 10, compared with roughly 29% for Perplexity. Nearly 90% of the pages ChatGPT cites rank at position 21 or lower organically, or don’t rank in the traditional sense at all. Different engines also draw from meaningfully different sources: Wikipedia is the single most-cited domain on ChatGPT, while Reddit is the most-cited domain on Perplexity and Google’s AI features — accounting for as much as one in five citations on Perplexity specifically.
Put together, this is the clearest evidence that agentic channels reward a fundamentally different kind of asset than paid search or paid social ever did. You can’t outbid your way into an AI-generated answer the way you can buy a top search ad slot. What gets cited is disproportionately community discussion, third-party editorial coverage, and structured, verifiable information.
How the Major Platforms Are Already Drawing Different Lines
Three of the largest AI platforms are running three different experiments with the same underlying trade-off — advertising versus trust — and watching them diverge tells agencies more about the durable rules of this channel than any single campaign result will.
OpenAI launched ads inside ChatGPT for Free and Go tier users on February 9, 2026, expanding to a self-serve platform on May 5, 2026. The design choice is deliberate: sponsored cards appear below the answer, clearly labeled, and per OpenAI’s own stated principles, never inside or altering the answer itself. Paying subscribers on Plus, Pro, Business, and Enterprise tiers see no ads at all. The bet is that a hard structural wall between the paid unit and the generated answer lets monetization and credibility coexist.
Perplexity took the opposite bet the same month. It phased out its own sponsored placements entirely and told the Financial Times it had no near-term plans to bring advertising back, framing the decision around a simple idea: users need to believe they’re getting the single best answer, not one shaped by who paid for visibility.
Google, meanwhile, continues integrating advertising more directly into its AI Overviews and AI-powered search features, leaning on the trust it has already built over two decades of clearly separating “ads” from “results” in a format users understand intuitively.
None of these bets has been proven right or wrong yet. They’re genuinely different theories about where the value actually sits, which means a strategy agencies build around any one platform’s current rules is fragile, because those rules are still being actively contested by the platforms themselves.
What Earns Citations in AI Search Results?
Fractl’s research surfaces something useful here: a “validation cycle.” Consumers increasingly treat the AI answer as a starting line, they check an average of roughly 2.4 platforms before making a purchase decision, often following an AI answer with a second search, a review site, or a forum thread. A brand that shows up in the AI answer and gets corroborated in that follow-up check wins the cycle. A brand that only shows up in the AI answer, with nothing to back it up anywhere else, tends to lose the buyer’s confidence at exactly the moment it matters most.
That maps directly onto the citation data above. The sources AI engines cite most, like, Wikipedia, Reddit, YouTube, LinkedIn, independent trade press, are exactly the sources a skeptical buyer would check next to verify an AI-generated claim. Earning a place in that layer isn’t a side effect of good PR. It’s now a direct input into whether an AI agent surfaces a brand at all, and whether a human trusts the answer once they see it.
How Can Agencies Build Trust for Agentic Advertising? A Practical Playbook
None of this is a reason to sit out agentic channels. It’s a reason to compete on a different axis than the one most media plans are currently built around.
- Fund earned visibility deliberately, not opportunistically. With paid and advertorial content capturing a fraction of a percent of citations, PR, analyst relations, community engagement, and third-party editorial coverage aren’t adjacent to an agentic strategy but close to its foundation.
- Show up where the validation happens. Community platforms, review sites, and forums do outsized work in both AI citation share and in the human follow-up checks that come after an AI answer. Treat presence there as core distribution, not an afterthought.
- Be careful with “AI” in the brand’s own voice. With six in ten consumers saying the word “AI” in a brand’s messaging is a turnoff, and younger consumers the most sensitive to heavy AI use, disclosure and tone matter more than most teams currently budget for.
- Build content that reads the same to a person and a machine. The most durable way to earn a citation is content that would be exactly as accurate and useful if a person read the raw source. That’s the actual product now, since agents are increasingly the first audience a piece of content meets.
- Diversify the platform bet. Because OpenAI, Perplexity, and Google are running three different philosophies on ads and trust, don’t build a strategy that only works under one of them. Aim to be genuinely citation-worthy everywhere, and treat any single platform’s ad product as a bonus, not the foundation.
- Track citation share the way you’d track share of voice. Multiple firms now publish or sell tools for measuring how often a brand is cited across ChatGPT, Perplexity, Gemini, and AI Overviews. Treat it as a standing metric, not a one-off audit.
Agentic Advertising FAQs
What is agentic advertising?
Advertising built for an AI agent to retrieve, evaluate, and potentially use when generating an answer, rather than content designed primarily for a human to see and click.
Why does trust matter more in agentic advertising than in traditional digital ads?
Because an AI agent typically hands the user one synthesized answer instead of a page of comparable sources. Users can no longer filter sources themselves, so the credibility of the answer depends almost entirely on the system and the sources behind it, and survey data shows trust in that system is currently declining even as usage rises.
Can a brand simply pay to be cited in an AI-generated answer?
Not reliably. Independent studies of AI citations put paid and advertorial content at roughly 0.3% of what gets cited, against about 84% for earned media. Credibility, not spend, is the dominant input.
Do all AI platforms treat advertising the same way?
No. OpenAI runs ads in ChatGPT but keeps them structurally separate from the generated answer. Perplexity has avoided advertising entirely since February 2026. Google is integrating ads more directly into its AI Overviews. Each platform is setting its own policy independently, with no shared industry standard yet.
How can agencies build campaigns that hold up well in agentic channels?
Invest in earned and community visibility, keep owned content equally accurate for humans and machines, avoid strategies built around a single platform’s current rules, and track citation share the way you’d track any other visibility metric.
Key Takeaways: Trust and Agentic Advertising in 2026
The data tells one consistent story: usage of AI search is climbing fast, trust in it isn’t keeping pace, and the money that used to buy visibility in a search result barely moves the needle on visibility in an AI-generated answer. That’s a real constraint but it’s also the clearest opening in the channel right now. Agencies that treat credibility as something to build deliberately, through earned coverage, community presence, and content that holds up under machine scrutiny, are the ones that will stay visible no matter which way any individual platform’s ad strategy eventually breaks.
