AI search visibility is moving to retailers

Google AI Mode just told you who it trusts to answer for it, and it isn't the publishers who spent a decade building traffic on Google's back. Kevin Indig's Growth Intelligence Brief #22, published August 6, 2026 on Growth Memo, tracked citation behavior across 20 verticals and found AI Mode cut publisher mentions by nearly 50% while tripling how often it names Home Depot and quadrupling how often it names Wayfair. That pattern showed up in 15 of the 20 verticals Indig tracked, not a fluke confined to home goods. If you've been treating AI search visibility as a fight over mention rate and citation rank alone, this report adds a second dimension: which category of domain gets to exist in the answer at all. Retailers are winning that fight right now. Publishers and community platforms are both losing it, at the same time, inside the same surface.

The AI search visibility shift, in one report

Indig tracks AI Mode citation behavior the way an analyst tracks a stock: same panel of prompts, same methodology, run on a cadence, so a move this size isn't noise from one bad sample. Growth Intelligence Brief #22 checked citation behavior across 20 verticals and found the retailer shift in 15 of them, which means three out of every four categories he checked showed the same directional move. When a pattern shows up in 75% of the verticals you're tracking, it stops being a vertical story about home goods or hardware. It becomes a platform decision about who counts as a trustworthy source, applied broadly across categories that have very little else in common.

Put a number on what tripling and quadrupling actually looks like inside a single vertical. Say a home-goods category handed out 20 citations to Home Depot and 20 to publishers in Indig's earlier tracking window, our illustrative math, not Indig's own per-vertical figures. A 3x increase for Home Depot alongside a roughly 50% publisher cut moves that split to something like 60 Home Depot citations against 10 publisher citations. The content that used to anchor the category's answer is getting displaced by a retailer's product and category pages, and the retailer is walking away with several multiples of what it held before. That's the shape of the shift, even before you know the exact per-vertical counts.

Why AI Mode search is consolidating toward retailers

AI Mode isn't just an answer engine with a chat interface bolted on anymore. It's increasingly a shopping assistant with a search box wrapped around it, and that changes what counts as a trustworthy source for the queries it targets. A publisher's roundup post can describe a product well. A retailer's page can show the live price, current stock, reviews, and a purchase path in the same response. When AI Mode is optimizing for a user who might buy something inside the session, a retailer's page answers more of the query in one shot, and Google's own citation behavior in this report reflects that bias plainly.

This is also why treating AI Mode like a single monolithic engine is a mistake we've argued against before. Treating each engine as its own investment case matters here specifically, because AI Mode's retrieval preferences are shaped by Google's commerce ambitions in a way ChatGPT's or Claude's currently are not. An engine that increasingly wants to close the transaction, not just answer the question, is going to keep favoring domains built to close it.

Home Depot and Wayfair aren't arbitrary winners here. Both run large, schema-rich product catalogs, both already feed Google Shopping, both carry deep review volume, and both have the domain authority and crawl budget to get every SKU indexed cleanly. For our retail and ecommerce clients, this is good news dressed as a warning: the retailers already investing in clean product schema, verified reviews, and live inventory feeds are the ones AI Mode is now naming several times more often. For everyone else, publishers, review sites, and comparison content that doesn't sit on a transactional domain, it's the opposite. AI search visibility that used to come from being the best explainer of a category is eroding in favor of being the place where the category actually gets bought.

Reddit's collapse shows this isn't just a publisher problem

Reddit is the control group here, and it failed too. Indig's brief found Reddit lost its entire June 2026 citation recovery, dropping 14.0% to a mention index of 2,312.5 by July 15. That recovery had been read by some in the GEO community as proof that Reddit's earlier citation collapse was temporary, a licensing-driven dip that would correct itself once search and AI platforms settled their access agreements. Growth Intelligence Brief #22 says the correction didn't hold. Whatever ground Reddit clawed back in June, AI Mode gave right back within six weeks.

Run the arithmetic backward and you can see the size of what got erased. If a 14.0% drop lands the index at 2,312.5, the pre-drop figure Indig's panel was tracking in June sat close to 2,690, our math, derived from the reported percentage, not a number Indig published directly. That's roughly 375 points of citation index gone in six weeks, on a platform that had just spent the prior month telling the industry its citation problem was behind it.

We've made the case before that Reddit's citation math was already breaking down, and Brief #22 is a second data point in the same direction, not a new argument. What's different this time is which engine is doing the cutting. AI Mode isn't just deprioritizing Reddit the way ChatGPT's crawlers reportedly do. It's cutting Reddit and publishers at the same time it's expanding retailer citations, which means the freed-up citation slots aren't going to community content or independent editorial either. They're going to a narrower list of large, transactional domains. Community platforms and publishers aren't competing with each other for what's left. They're both losing to the same third category.

What the shift means for AI citation tracking

This isn't happening on a marginal surface. AI Mode crossed 1 billion monthly active users earlier this year, which means a citation swing inside it reaches a genuinely mass audience, not a beta cohort of early adopters. Any AI citation tracking setup that treats AI Mode as one line item next to ChatGPT and Perplexity is under-weighting a surface with real transactional intent behind it. The queries AI Mode increasingly answers skew closer to "buy" than "learn," and that's exactly the query type where this report shows citation share moving hardest toward retailers.

The other trap right now is reading a shrinking mention-rate number as bad news across the board. AI Mode's total mention volume contracted 12.9% over the period Indig studied. If your dashboard reports a raw citation count and that count is down, the instinct is to treat it as underperformance. But a domain-type breakdown tells a different story: publishers and Reddit are absorbing nearly all of that 12.9% contraction, while retailers are gaining share inside a shrinking pool. A flat or declining citation count for your brand could mean you're being squeezed exactly the way publishers are, or it could mean nothing at all if you don't compete in a vertical AI Mode treats as transactional. You can't tell the difference without segmenting by who's gaining and who's losing, which most mention-rate dashboards still don't do.

“A citation count without a domain-type breakdown is a vanity metric wearing a spreadsheet.”

How to protect AI search visibility when retailers own the answer

None of this means publishers or community sites are finished in AI Mode. It means the default advantage has flipped toward domains built for transactions, and a program that doesn't account for that is optimizing for a version of AI Mode that stopped existing on August 6. Four moves, in order of how fast they pay off.

Do this in the next two weeks, not the next planning cycle: pull your AI Mode citation data specifically, split it by publisher, community, and retail/transactional domains, and check whether the ground you've lost matches the shape of Brief #22. If you're a retailer or ecommerce brand, this report is your cue to double down on the generative engine optimization work that makes your product pages the kind AI Mode already prefers to cite: the schema, the reviews, the live pricing. If you're a publisher or comparison-content site, the fix isn't panic, it's specificity. Our research on where citations actually concentrate argues the winners inside any given format are still a small, ownable list, and this report just tells you retail is the format gaining ground fastest right now. Either way, stop reporting a single mention-rate number to your leadership team and start reporting who's winning inside it. That's the number that would have caught this shift before Indig did.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.