There's something shifting in paid search this week that's worth paying attention to, and it's happening quietly rather than with a big product launch. Google has started letting ordinary Search campaigns — including ones with tight, restrictive match types — serve inside AI Mode. Alongside that, Search Console's AI reporting has gone worldwide, and the industry conversation has moved firmly onto whether AI tools recommend your shop at all. Here's what's going on and what it means for you.
Your Exact Match Keywords Are Now Showing Up in AI Mode
Google has confirmed it's running what Ginny Marvin, Google's Ads Liaison, called "a small experiment" — serving regular Search campaigns with restrictive match types (exact and phrase) inside AI Mode results.
Why does that matter? Up to now, the working assumption in the industry was that AI Mode traffic would mostly be fed by broad match and AI Max-style expansion, because AI Mode queries are long, conversational and messy. Someone doesn't type "leather weekend bag" into AI Mode. They type "what's a good leather weekend bag that fits in an overhead locker and won't look scruffy after a year". That's not a query your exact match keyword list was ever built to catch.
So if Google is now matching those sprawling conversational queries to your tight keywords, the safety net you thought exact match gave you is looser than it was. Your carefully controlled campaign is being asked to answer questions you never wrote keywords for.
What this means for you in practice: check your search terms report more often than you used to, and actually read the long, sentence-like queries. Those are the AI Mode ones. Some will be brilliant — genuine buyers describing exactly the problem your product solves. Others will be research questions from people nowhere near a purchase. You can't opt out of AI Mode separately, but you can absolutely add negatives, and you should be doing that weekly rather than quarterly right now.
Going forward, I'd expect this experiment to widen. Google rarely runs a "small experiment" that stays small. Assume that within a year, a meaningful chunk of your Search spend is going through AI-generated answers rather than a classic blue-link results page.
Search Console's AI Reporting Just Went Worldwide
Google has confirmed its Search Console AI reports are now available globally, giving you visibility into how your site performs in AI-driven search surfaces. John Mueller also weighed in on recovery timing after ranking drops, markdown formatting for AI crawlers, and cache-busting your sitemaps.
For an online shop, the useful bit is the reporting. For most of the past year, "how are we doing in AI search?" has been answered with a shrug. Now there's actual data, and it's free, and it's already sitting in an account you probably have.
What this means for you: if you sell products online, your product and category pages are being read by AI systems that then summarise them for shoppers. If those pages are thin, badly structured, or bury the specs in an image, you're invisible in that summary. Mueller's point about markdown for AI crawlers is a nudge in the same direction — clean, structured, readable content wins. That's not a new SEO trick; it's the same discipline that makes a good product feed. Clear titles, real descriptions, proper attributes.
Worth spending twenty minutes in Search Console this week just to see what's there. You may not like the numbers, but you can't fix what you can't see.
The New Question: Does AI Actually Recommend Your Shop?
This piece tackles something I'm paying close attention to — how to find out whether AI tools recommend your business, and how to close the gaps when they don't.
Here's the shift in plain terms. For twenty years, the game was ranking. Get to position one, get the click, make the sale. Increasingly, the game is being mentioned. When someone asks an AI assistant "where should I buy X", you are either in that answer or you're not, and there's no page two to console you.
What this means for you: go and ask the AI tools directly. Type "best place to buy [your product category] in the UK" into a couple of them and see what comes back. If your competitors appear and you don't, that's a gap with a real revenue number attached to it. The fixes tend to be unglamorous — comparison content, genuine reviews on third-party sites, clear specification pages, being listed where the AI is reading from.
None of this replaces your paid search. But it does mean the traffic that used to arrive free from organic search is increasingly getting intercepted, which puts more weight on the paid side of the equation. Budget accordingly.
AI Agents Are Starting to Do the Interviewing
An interesting one for anyone producing content: teams are now using AI agents to interview subject matter experts and turn those conversations into published material.
Sounds niche, but think about the bottleneck in your own business. You know your products better than anyone. The reason your category pages are thin isn't a lack of knowledge — it's that nobody has the time to sit down and write it out. An agent that can interview you for twenty minutes and produce usable content changes that maths considerably.
What this means for you: the constraint on good product content has always been time, not expertise. If that constraint loosens, the shops that fill out their content properly will pull away from the ones that don't — and given what's above about AI visibility, that content is exactly what determines whether you get recommended.
Proving Paid Media Actually Made You Money
This one's framed for B2B — how to produce a pipeline number from paid media that you can defend to your finance team — but the underlying point applies to every e-commerce business I've ever worked with.
The question is always the same: what did the ad spend actually deliver? Platform-reported conversions are generous. They include people who would have bought anyway, and they don't always survive contact with your actual sales figures.
What this means for you: your Google Ads conversion number and your bank balance need to be in the same conversation. If Google says you made £40,000 last month and your accounts say £26,000, that gap needs explaining before you make a single budget decision. Reconcile the platform numbers against real orders. It's dull work and it's the most valuable hour you'll spend on your account all month.
What I'm Watching
There's a thread running through all of this, and it's not really about AI at all. It's about control. You had control over which queries triggered your ads — that's loosening. You had control over how shoppers found you — that's moving into systems you can't see inside. What you still control completely is the quality of what you're selling, how clearly you describe it, and whether you know your real numbers.
That's where I'd put the effort over the next few months. Watch your search terms weekly, get into Search Console's AI reports, find out whether AI tools mention you, and reconcile your ad platform's conversion figures against actual orders. The businesses that come through this shift well won't be the ones that chased every new feature. They'll be the ones whose fundamentals were solid enough that the algorithms had something worth recommending.