Alcohol ads land on YouTube in October, and AI Mode is quietly picking a different seller than your carousel
Seven verified moves this week, including a hard October 30 activation date for personalized alcohol ads on YouTube, a study showing Google AI Mode rarely shows the same seller as the shopping carousel, and Search Console's AI reports going global so you can finally measure the leak.
YouTube opens personalized alcohol ads on October 30
Google is updating its personalized advertising policy for YouTube to allow alcohol, alcohol-related products, and alcohol alternative beverages starting October 30, 2026, in markets where local law permits. Egypt, India, Indonesia, and Poland are carved out for now. Age restrictions, sensitive category rules, and the ban on personalizing to minors all stay in place.
Who this hits: alcohol brands, their retained agencies, and any DTC beverage operator who has been running unpersonalized YouTube inventory because the policy blocked audience work. From October 30, competitors who prepped audiences and creative in advance will bid against you with a targeting advantage on day one.
The move this week (under an hour): open Google Ads, draft one Demand Gen campaign shell against your priority market, and stage two audience segments (existing customer match plus a lookalike off your highest-LTV cohort). Do not enable it. Have the creative brief for age-gated variants written by Friday so production is not the bottleneck on the 30th.
Search Engine Land: YouTube personalized alcohol adsGoogle Ads policy update (official)
ChatGPT Ads is now self-serve across 31 European markets
OpenAI opened self-service access to ChatGPT Ads through Ads Manager across 31 European markets as of August 31, 2026, and added targeting, measurement, pacing, and reporting tools. Custom audiences now support mixed identifier types in a single request and can exceed 5 million members. Ads only show to users on the Free and Go plans.
Who this hits: performance advertisers in the UK, EU, India, and MENA who have been watching ChatGPT Ads from the outside. Early CPMs on new inventory reward whoever shows up first with a clean measurement setup. If you wait until Q4 planning, the auction density will not be what it is today.
The move this week (under an hour): request Ads Manager access from your OpenAI rep or the self-serve queue, and pull the audience identifier spec into your CDP or CRM ticket queue so ops knows what format to export. If you already run Meta or Google Customer Match, the identifier work is mostly a mapping exercise, not a new build.
Search Engine Land: OpenAI expands ChatGPT Ads globallyOpenAI: ChatGPT Ads expands across Europe
AI Mode is showing a different seller than your carousel, at a higher price
Productrise monitored over 2 million product listings from August 9 to August 31, 2026, running the same product searches in AI Mode and standard search on the same day in the US and UK. Only 1.28% of products in the carousel also appeared in AI Mode for the same query. Among the products that did appear in both, the first seller listed was different 49.6% of the time, and the AI Mode price was 21.6% higher on average. Google told Futurism it has not verified the report and said both surfaces draw from the same Shopping Graph.
Who this hits: retailers and brands that assumed AI Mode inherits their carousel ranking. If a reseller or marketplace is winning the AI Mode surface for your own SKU, you are losing the click and, given the price gap, possibly training the buyer to expect a higher anchor from a competing seller.
The move this week (under an hour): pick your top ten revenue SKUs, run each query in AI Mode logged out, and record who the first seller is and at what price. Anywhere a competitor or marketplace outranks you, flag the merchant feed and product page for review. This is a measurement task before it is an optimization task.
Search Engine Journal: AI Mode prices differ from carouselProductrise study
Search Console AI reports went global on August 31
Google rolled out the Search generative AI performance reports in Search Console to all sites worldwide as of August 31, 2026, alongside the global opt-out toggle for AI Overviews, AI Mode, and AI Overviews in Discover. The report shows impressions in AI features for Search and Discover. It does not show clicks generated by those appearances. Sites without enough data will not see the report yet.
Who this hits: any operator who has been watching organic sessions slide without a way to attribute the loss to AI surfaces. You now have a first-party impression baseline. If you do not pull it this week, the earliest data window will roll before you have anything to compare against.
The move this week (under an hour): open Search Console, export the new AI performance report for the last available window at the property and page level, and drop it in a dated tab in your reporting sheet. Separately, decide (do not toggle yet) whether the opt-out control is something you want to model against a page group before rolling it wider.
Search Engine Journal: SEO Pulse on Search Console AI reportsGoogle Search Central: gen AI performance reportsGoogle Search Console help
Reddit Max campaigns opened to all advertisers
Reddit announced on September 3, 2026 that Max campaigns, its automated buying product covering targeting, creative selection, placements, and budget allocation, are now available to all advertisers. The expansion adds API integration, an onboarding path through Smartly, and asset group capability. Reddit's beta claims are 17% lower CPA and more than 27% additional conversions on average, which is Reddit's own number and worth treating as directional.
Who this hits: Reddit advertisers running manual campaigns, and agencies whose reporting is built around manual placement logic. As Reddit tilts optimization weight toward Max, manual campaigns are competing for the same auction with less algorithmic support.
The move this week (under an hour): duplicate one existing manual campaign as a Max test with a small, non-critical budget, keep the manual original running, and set a two-week read window. Do not migrate the account. You are buying a comparison, not a conversion.
Shopify POS now shows an identified customer's online cart to store staff
As of September 3, 2026, authorized staff on Shopify POS v11.14 can look up an identified customer and see items from an eligible abandoned online checkout. The customer must be identified, have an email, have an eligible abandoned checkout, and the POS session must have location context. Staff see the cart contents, not browsing history. The permission is off by default and has to be enabled at Settings > Users > Roles > POS Role(s) > Customers, under View abandoned checkouts.
Who this hits: Shopify merchants running both online and physical retail, particularly in categories where in-store consultation closes the sale (apparel fit, beauty, home). Every walk-in with an open online cart is a recovery opportunity your associates currently cannot see.
The move this week (under an hour): enable the permission for one store's POS role, brief associates on the two-question script (confirm identity, offer help with the item in the cart), and pick one KPI to watch, either recovered cart value or attach rate on identified customers. Two weeks is enough to know whether to roll it wider.
B2B product pages are earning six times more AI citations than Reddit
A Ten Speed study covered by Search Engine Journal reports product pages earn roughly 24% of B2B AI citations against 4% for Reddit and YouTube, based on 170 vendor-comparison prompts monitored through Peec AI across ChatGPT, Perplexity, Claude, and Gemini. Two caveats matter: a prompt-count visual in the piece said 220 rather than 170 (Ten Speed confirmed 170 is correct and will fix the chart), and the number of distinct client brands behind the 7,387 citations was not disclosed. Treat the direction as more reliable than the exact percentages.
Who this hits: B2B and considered-purchase operators whose product pages are still thin marketing landers rather than structured specification pages. At the vendor-comparison stage of the AI buying funnel, the page an assistant can quote from is the page that gets cited.
The move this week (under an hour): pull your top five product or solution pages and check three things per page: is there a clear one-sentence definition of what the product is, are the specifications or capabilities in a structured list rather than prose, and is pricing or a pricing model present or explicitly explained. Fix the worst offender first.
Search Engine Journal: product pages get 24% of AI citations
Checked, and quiet
One signal in the input set did not clear the bar this week. The Ask Al video on Google Postmaster Tools could not be verified at its source URL, so we dropped it rather than paraphrase around it; the underlying advice to check Postmaster Tools before your next send remains sound regardless.
Signals are checked weekly, and quiet weeks are admitted rather than padded. If you want these same checks run against your own site and stack, the free read at app.signalandscience.com will do it in a few minutes.
Get the next Signal & Science brief when there is something worth acting on.