AI is taking over media buying faster than most teams have adjusted for. Gartner expects more than 70% of global ad spending and 80% of U.S. ad spending to flow through self-serve platforms where AI "materially influences" buying, costs and outcomes by 2028. That leaves one uncomfortable question: AI ad platform bias.
Eric Schmitt, VP analyst in Gartner's Marketing practice, is blunt about how much autonomy to hand over. "You wouldn't give your 13 year old your credit card, send them to the grocery store and tell them to make good buying decisions," he said. "Similarly, you don't want to give these ad platforms unfettered access to your media budget without a human in the loop."
The reason is structural. The AI was built in service of the platform, not the buyer. When maximizing price for the platform meets minimizing cost for the buyer, it's obvious which way the model leans.
"It's kind of unsurprising that a lot of the AI wizard recommendations that I see surfacing in these user interfaces start with, 'You should spend more with us,'" Schmitt said. "At the end of the day, the core mission for these algorithms is to generate the numbers necessary for the platform provider to hit its quarterly revenue and profit results."
The platforms also "don't often play nice with each other," which leaves multichannel measurement and interoperability exactly where they were.
Schmitt's prescription is to narrow the field. "Ask yourself: 'What's the overall role of the media relative to what I'm trying to accomplish?' Don't let the shiny objects and glitz of AI distract you from investigating the core physics of the platform."
Beyond audience size and usage, evaluate whether the platform lets you map spend to the outcomes you actually want. And treat reducing platform count as strategy: Google, Meta and Amazon capture over half of U.S. paid media spend and are approaching 60% globally excluding China.
Schmitt suggests pulling in other stakeholders — finance executives among them — to assess whether AI buying platforms are working. "Use this as an opportunity to find common ground with your own internal stakeholders and make sure that your strategy is in line with theirs, as much as possible, so you can have consensus on the risk versus reward of these decisions."
AI remains genuinely useful, and the barrier to entry has dropped. "You don't need a tech background to start to troubleshoot or understand what's happening," Schmitt said. "You can take more junior people or repurpose existing staff and put them in front of these consoles. Even if they have never done a regression analysis or put together a campaign brief in their lives, they can now do that. And that's great… until it's not."
Which is why experienced oversight and independent measurement providers still matter. "You have to play the game. But judge the performance and outcomes of the campaigns through a lens that's maybe a little different from the fox in the henhouse reporting that you will get from the individual platforms."
Turn automation levers on one at a time and document what changes each time. Platforms are conceding some control — Performance Max is testing channel prioritization — and every exposed lever is worth taking. And stay suspicious of metrics the platform finds easy to optimize, the pattern behind cheap clicks that look like efficiency.
No. Read them knowing the objective function may be platform revenue rather than your return. Accept the recommendations, but keep a human review step before budget moves.
Cut variables so causes stay traceable. Google, Meta and Amazon already hold over half of U.S. paid media spend, so working a few platforms deeply beats spreading thin across many.
Because platforms grade their own homework. Platform reports and independent measurement can diverge sharply on the same campaign, so you need a separate baseline for budget decisions.
To apply what you just read to your own site, start with a free audit of where things are now.
A strategist replies within 24 hours on business days.
Most business plans fail on structure, not content. Five principles for writing a business plan evaluators can actually judge: problem first, evidence, revenue logic, purposeful charts and counter-questions.
Startup marketing should start while the product is still being built. Three pre-launch principles for founders: start early, go where your audience already is, and budget for experiments.
Korean alcohol shipments fell to a 27-year low, but receipt data shows 45–51% of shoppers still buy alcohol at retail each month. What that means for channel, timing and seasonal campaigns.