Google made a change to search terms reporting for AI queries this week, and barely anyone noticed. Tucked away in an update that received minimal coverage, Google altered how search terms are shown to advertisers when those searches come through AI Mode, AI Overviews, Lens, and autocomplete. Instead of showing you what the person actually typed, you might now see an "interpreted" version of the query.

This is worth paying attention to, not because it's catastrophic, but because it's a clear signal of the direction we're heading.

What Actually Changed

When someone searches using AI Mode—Google's conversational search interface—the query they type might be long, messy, or phrased in a way that doesn't map neatly to a keyword. Google's system interprets that and serves the relevant ads. Fair enough.

The change is that advertisers now see the interpreted query in their search terms report, not necessarily the original one. So if someone asked Google AI "what's a good gift for someone who likes cooking but already has all the equipment", you might see something cleaner in your report—rather than the full rambling question.

On the surface that sounds reasonable. In practice, it means you have less visibility into exactly what people were thinking when they found your ad. And for anyone who spends time cleaning up search terms reports and adding negatives, that's a meaningful change.

The Bigger Picture: AI Is Reshaping Where Your Customers Come From

At the same time, there's good news for e-commerce businesses. Adobe's Q2 data this week showed a 393% increase in retail conversions from AI-referred traffic. That's not a rounding error—that's a genuine new channel that's starting to deliver results.

Google Analytics has also just added AI assistants as a default channel group. ChatGPT, Gemini, Perplexity—these are now tracked separately in GA4 rather than lumped into referral traffic. Which means you can finally start to see, properly, whether any of these tools are sending buyers your way.

If you haven't checked your GA4 recently, it's worth a look. You might find you're already getting a trickle of AI-referred traffic that was previously invisible to you.

What "Google-Agent" Means for Online Shops

There was another quiet development this week that I think will become a bigger story over the next year. Google has introduced something called Google-Agent—essentially an identifier for AI systems acting on behalf of users. Not a search crawler, not a human visitor: an AI agent browsing, comparing, and potentially buying on someone's behalf.

This is early days. But think about what it implies. If someone sets up an AI assistant to "find me the best price on X and order it", that agent will be visiting your product pages, reading your descriptions, checking your stock levels, and making a decision.

Is your product page set up to speak to a human browser? Almost certainly. Is it set up to give a machine agent exactly what it needs to make a confident purchase decision? Probably not yet—and neither is most of your competition.

This is going to matter. Clear pricing, structured product data, accurate stock information, specific specifications. The pages that win in AI-mediated commerce will be the ones that give clear, unambiguous information—not the ones with the cleverest copy.

What Your AI Ad Strategy Actually Needs

There was a useful reminder in the industry press this week that AI in advertising magnifies whatever you feed it. Strong inputs—good product data, clear conversion signals, useful audience information—produce better results. Weak inputs produce worse results, faster.

Performance Max is the obvious example. The campaign type is essentially an AI system that takes your assets, your audience signals, and your conversion data, and tries to find buyers. If your product feed is patchy, your conversion tracking is shaky, or your assets are generic, the AI doesn't have much to work with.

The practical upshot: the basics matter more than ever. A clean product feed, accurate conversion tracking, and well-structured assets aren't just good housekeeping—they're the inputs your AI campaigns depend on to perform.

What to Do This Week

Nothing here requires urgent action. But here are three things worth checking:

Check your GA4 for AI channel traffic. Look under Traffic Acquisition for the new AI assistant channel group. If you're already getting visits from ChatGPT or Gemini, note whether those sessions are converting. It's early data, but it's worth knowing.

Glance at your search terms report. If you run Search campaigns alongside Performance Max, keep an eye on your search terms for signs that the query types are shifting. More conversational queries, more question-based phrases. If you see them, consider whether your landing pages actually answer those questions.

Don't panic about the search terms reporting change. It affects a relatively narrow slice of queries right now—the AI Mode traffic that's still a small percentage of overall search. Your existing negative keyword work and search term hygiene still applies; you just need to be aware that some of the data may be interpreted rather than verbatim.

The overall direction is clear: AI is changing how people search, how they discover products, and how they ultimately buy. The businesses that do well in this environment are the ones keeping their fundamentals solid—good data, accurate tracking, clear product information—rather than chasing the latest feature.

That's always been true. It just matters more now.