Why is AI showing your competition?

July 24, 2026

The Speed Read

Google AI Mode now shows text ads on 33% of commercial keywords, but the brands cited as sources inside the AI answer are almost never the same advertisers whose ads appear below it. Two auctions, two different winners.
New analysis finds 70% of top retailers can’t be transacted with by AI shopping agents. Meanwhile, agents drove 20% of holiday retail sales in 2025. The gap between agent-ready and everyone else is about to show up on a P&L.
Amazon’s biggest third-party sellers are organizing like a trade association, comparing fee stacks and pushing back on ad take rates that have climbed for three straight years.
Retail media networks are shifting from placement sellers to content producers. DTC brands still handing Amazon a static image and a headline are leaving performance on the table they already paid for.

Google Turned Your Search Auction Into Two. Most Brands Are Only Playing One.

New data shows Google AI Mode now serves text ads on roughly 33% of commercial keywords. That sounds like good news for advertisers. It isn’t, not exactly.

The harder number is buried in the same analysis: the brands cited as sources inside the AI answer above those ads are almost never the same brands whose ads appear below it. That’s two surfaces running on the same query with two different winners.

Your Performance Max campaign is optimized to win the bottom slot. Your competitor’s product feed and structured data are winning the citation in the answer above it. The shopper reads the AI summary first, clicks the brand it mentioned, and your ad technically served. It lost the impression that mattered.

Most performance teams aren’t catching this because attribution doesn’t see it. You get an impression, possibly a click, and the model reports a conversion. What you don’t see is how often you won the auction and lost the consideration.

The fix isn’t complicated, but it requires checking something your dashboard currently doesn’t track. Audit your top 20 commercial queries directly in AI Mode, not in Google Ads reporting. The question isn’t whether your ad is showing. It’s whether your brand is being cited in the answer above it.

Two operational moves are worth making this week. Clean up your product feed and PDPs if they’re running on 2022 copy and incomplete structured data. AI Mode citations pull from the same signals as Shopping, and a messy feed means you’re invisible in the answer even when you’re bidding. And start tracking PMax alongside AI Mode visibility separately. The campaign’s reported performance looks normal right now because the clicks that do happen get credited. The impressions intercepted before they hit your ad are invisible in your current reporting.

The auction changed in the spring. Most advertisers haven’t adjusted yet, which means this is still a gap worth closing before Q4 drives commercial query volume up.

If you want a second set of eyes on how your accounts are performing inside AI Mode versus standard Search, that’s what our media audits cover.

70% of Top Retailers Can’t Be Bought From by an AI Agent. Q4 Is Coming.

A new analysis found 70% of the top 100 retailers are effectively invisible to AI shopping agents. Not underperforming. Invisible. The agent hits the site, can’t parse the product data or complete a checkout, and moves to the next option on its list.

This matters right now because agents aren’t a 2027 story. Salesforce reported they drove 20% of retail sales during the 2025 holidays. That share will be higher this year.

The bar to be in that pool is surprisingly low: machine-readable product data, agent-friendly checkout, correct schema markup on your PDPs. Brands clearing it aren’t doing anything exotic. They’re doing fundamentals most teams keep deprioritizing because the payoff doesn’t show in this quarter’s ROAS report.

The uncomfortable math: a $500,000 creative overhaul does nothing if the ChatGPT shopping agent can’t add your product to a cart. The creative optimization that’s been sitting on your roadmap for six months is lower-ROI than one sprint cleaning up your product feed and testing checkout completion for non-human sessions.

Three checks worth running before Q3 ends. Pull your product catalog and confirm every item has complete, accurate pricing, inventory status, and description in a format an API can consume. Run a synthetic test of your checkout flow with an automated session and see where it breaks. And run your top 20 PDPs through Google’s Rich Results Test and fix anything flagged as incomplete.

The brands that are agent-ready going into Q4 didn’t do anything remarkable. They just did it first.

Quick Takes

Amazon sellers are organizing, and the margin math has changed. The Million Dollar Sellers group met in New York during Amazon’s own Seller Growth Summit last week and spent their time comparing fee stacks, not celebrating. Ad take rates are climbing. Sponsored placements now appear in slots that used to be organic. Amazon Marketing Cloud fees are compressing the same margins that made the channel attractive three years ago. Before Q4 budgets lock, run your fully-loaded Amazon margin math with 2026 costs. Brands quietly rebalancing toward Walmart Connect, Shopify Audiences, and their own DTC channels are doing it because the arithmetic told them to.

Retail media networks are asking for your content budget, not just your media budget. Amazon, Walmart, and two other major RMNs are actively pitching shoppable video, sponsored series, and brand-built commerce formats instead of pure placements. The reason: placement-only revenue caps out, and they want more of the production line too. If your agency hands Amazon a static image and a headline in Q4, you’re leaving performance on the table inside inventory you already paid for. Ask your agency this week whether they can produce retailer-native creative in-week. If the answer involves a two-month studio process, that’s a structural problem going into the highest-spend quarter of the year.

Dollar Shave Club looked at the AI tool bill and didn’t know what it was paying for. The brand’s new chief brand officer spent her first month running everything through Claude, ChatGPT, Higgsfield, and Gemini. Then she looked at the cost. Her quote: “I have no idea what that bill looks like.” Within weeks she built a triage system: cheap models by default for routine tasks, expensive models only when the work actually demands it. Most marketing teams don’t have that documented yet. If you can’t explain to your CFO which AI tool is doing which job and why, you’re a few budget conversations away from having to. Build the triage rules before the review, not during it.

They texted. They DM’d. They moved on.

Over 3.5 billion people open WhatsApp, Instagram, or Facebook Messenger every day. Your customers are already there, asking questions and deciding who to buy from.

Wati puts your business across WhatsApp, Instagram DM, Facebook Messenger, SMS, RCS, and web chat in one AI-powered inbox.

Automations respond instantly, track every conversation, and ensure nothing gets missed.

Meet customers where they already are, before your competitor does.

Meet your customers where they already are. Before your competitor does.

See All Wati Channels

The Last Word

Three stories, three different platforms, one signal: the performance stack most brands built in 2023 is being tested against an opponent it wasn’t designed for.

The brands going into Q4 with an edge already ran the audits. Feed hygiene, AI Mode visibility, Amazon margin math. The window to do that without time pressure is closing.

That’s the only real head start available right now.