AI Agents Won't Fix Bad Audience Data, They'll Amplify It
AI agents run on audience data at machine speed, so bad inputs come out louder, says Skydeo's Mallory Gray. Rising AI citations don't equal conversions — data quality still decides.
Updated
The brief
- Skydeo, drawing on 1.4 trillion data points across more than 320 million people, says AI agents amplify audience data quality — good or bad — instead of fixing it.
- "A brand can increase its visibility in AI answers substantially without seeing the same improvement in qualified traffic or conversions," said Mallory Gray, creative director at Skydeo.
- The DMP era of the early 2010s showed that scaled third-party data compounded errors faster — the same lesson now applies to AI agents.
AI agents run on audience data at a speed and scale no human team can match — and whatever quality that data had going in, good or bad, comes out the other end louder.
That is the core argument made by Mallory Gray, creative director at Skydeo, an audience data company that says it draws on 1.4 trillion data points across more than 320 million people. Skydeo sells exactly the kind of data the argument puts at the center, so it has an obvious stake in the answer. The argument still stands on its own.
"A human researcher might look at several sources, identify patterns, form a hypothesis, and build an audience from there," Gray said. An agent works across "thousands of behavioral, purchase, interest, and intent signals" simultaneously, continuously revising as new information arrives. Humans still decide what matters and what the brand should do about it. The agent just expands how much raw material can realistically feed that decision.
Being mentioned isn't the same as being chosen
The point cuts against most of what passes for AI marketing strategy in 2026. Marketers have spent the year racing to get cited by ChatGPT and Gemini, treating a mention as the finish line. Showing up in an AI answer and showing up for the right reasons are not the same accomplishment. Treating them as interchangeable is how a brand ends up automating its own blind spots.
The distinction sharpens once you separate GEO — generative engine optimization — from what Gray calls AI visibility. GEO, as most marketers practice it, makes content easy for a generative engine to extract and cite: clean structure, clear answers, credible mentions. Gray does not dispute that any of that works. Her argument is that citation frequency measures the wrong thing on its own.
"A brand can increase its visibility in AI answers substantially without seeing the same improvement in qualified traffic or conversions," she said. The fix is not more optimization. It is asking who is actually being served up your brand and whether that matches who your business needs.
Output up, results flat — the warning sign
The warning sign Gray flagged is one many SEO teams have likely already tripped: output keeps climbing — more content, more variations, more campaigns — while engagement or conversion quietly flattens or slides. Volume looks like progress because it is easy to measure and automation makes it cheap to produce.
The harder question, the one Gray says teams stop asking, is whether anyone on staff can still explain why a particular audience was targeted or a particular message went out. Once the honest answer becomes "the AI chose it," the feedback loop that used to catch a bad assumption is gone. A strategy can run itself into the ground for months before anyone notices the number that mattered was never the one going up.
We have been here before: the DMP era
There is a parallel from the last time marketing got obsessed with a data platform. The DMP era of the early 2010s promised that enough third-party data, stitched together at scale, would out-target anyone relying on first-party relationships. It mostly didn't. The third-party data was frequently wrong, and scale meant the wrongness compounded faster. Cookie deprecation from Safari and Firefox forced the industry back toward first-party and declared signals, even after Chrome dropped its own plan to deprecate them.
Gray's argument about AI agents is the same lesson with a new engine. Access to a powerful model is getting increasingly common. What a brand feeds into it isn't — and that is still where the advantage sits.
Three checks to run before you scale
Audit your audience signals. Separate what customers declared, what you observed them do, and what a model inferred about them. Be honest about how much of your targeting rests on the weakest of the three.
Stop measuring AI visibility as a citation count. Track whether the prompts you appear for match what your actual customers ask, and whether those mentions convert at a rate that justifies the optimization spend.
Build in a standing explanation check. Someone on the team has to explain, in plain language, why a given audience or message was chosen. If the honest answer is "the AI decided," that is the moment to slow down, not speed up.
AI agents are turning audience data quality into the thing that separates a brand that gets mentioned from a brand that gets bought. The agent isn't the differentiator. What you fed it before it ever started working was.
Based on linkedin.com
Filed under ai-agents, audience-data, geo, ai-visibility, data-quality
More from Amara Osei
Show full bio
News editor covering industry trends and analytics at Marketing Herald.
57 articles