What happened
Google's July Demand Gen drop, published July 29, expands Checkout Links to nine new markets: Switzerland, Australia, South Korea, Indonesia, Mexico, France, Poland, Israel, and Argentina. Supply a checkout URL with your ad and shoppers route straight to purchase instead of wandering your site first. Google's pitch number: advertisers who provided checkout URLs boosted conversions by 6% on Demand Gen — measured across a single month, September 2025, in the US only.
The quieter change is bigger. Target ROAS bidding got retuned so campaigns stop being over-cautious at launch and hit their targets faster. That is Google conceding what advertisers have grumbled about for years: new Demand Gen campaigns spent their opening weeks hedging, and everyone priced that hedge into their test verdicts.
The drop also deepens product-feed integration, adds an affiliate-partnership option for amplifying creator content, and — US only for now — wires Universal Commerce Protocol checkout into YouTube ads, so viewers can buy without leaving the ad. Google backs the retail push with an Ipsos holiday-shopping stat: 94% of holiday shoppers using YouTube reported taking further purchase steps after watching related video content.
Why this matters
Every Demand Gen verdict you reached under the old bidding behavior is now suspect. If you killed a test in week two because tROAS was sandbagging delivery, that test was graded under conditions that no longer exist. That doesn't mean the campaign would have worked. It means you don't actually know — and the retune just made finding out cheaper.
The 6% figure deserves your skepticism. It is Google's own measurement, one month, US only, drawn from advertisers who opted in — a group that skews toward stores with their checkout flow already in order. Treat it as a direction, not a forecast. Your account produces the only number that matters.
UCP checkout from YouTube ads is the tell about where this heads. Google is laying rails for purchases that never touch your website. In that flow, your product feed does the selling — titles, images, price, availability. Stores that treat the feed as an afterthought are handing those transactions to whoever didn't.
What to do about it
Retest what you shelved
Pull the Demand Gen campaigns you killed during the learning phase and rerun the best one or two under the retuned Target ROAS. Keep budgets and targets identical to the original test so the comparison stays clean. Annotate the account with the drop date so future you knows why performance shifted.
Run Checkout Links as a controlled test in the new markets
If you sell into Switzerland, Australia, South Korea, Indonesia, Mexico, France, Poland, Israel, or Argentina, split traffic between Checkout Links and your normal landing flow. Hold Google's 6% at arm's length and let your own conversion data set the benchmark.
Audit your feed before UCP-style checkout spreads
Universal Commerce Protocol checkout is US-only today, but the direction is set: buying happens inside the ad, off your site. Check that feed titles front-load the identifier, images meet spec, and price and availability sync cleanly. In that flow, the feed is your product page.
Map the holiday timeline now, not in October
Google shipped this drop with a 2026 Demand Gen Retail guide attached and a 94% YouTube shopper stat doing the selling. Ignore the cheerleading, keep the calendar logic: if Checkout Links are going to earn a place in your Q4 setup, the proving run happens in August and September, not November.