Google Ads started rolling out AI-generated dashboards to some advertiser accounts in early September 2026, letting marketers type plain-text prompts to get campaign performance charts and AI-written summaries of what changed. The move signals a shift toward natural-language reporting inside ad platforms at a time when attribution data itself is under scrutiny for reliability.
How the AI dashboards work
According to reporting by Search Engine Land on September 8, 2026, the dashboards accept plain-text prompts and return visualizations alongside narrative summaries that explain performance shifts. Google first announced the capability in August, and it is now appearing in accounts following that announcement. The same report noted that Google has added AI-powered insights elsewhere in Google Ads, including on the account homepage and through an in-product assistant called Ask Advisor.
The dashboard summaries aim to tell advertisers what changed and why, though the reporting does not specify how the system distinguishes between modeled estimates and directly observed conversions. For marketing teams managing large accounts, the feature could reduce time spent manually pulling and interpreting reports. It also raises a practical question: whether an AI-generated explanation of AI-optimized campaign performance creates a feedback loop that's hard to audit.
New tools for physical store sales
In a separate Search Engine Land report published the same day, Google Ads introduced two features for multi-location businesses. The first is Local Customer Optimization, a toggle inside Performance Max store-goals campaigns that prioritizes budget toward nearby in-market users across Maps, Waze, and local Search. The second is a new Data Manager option that lets advertisers connect CRM or Google Sheets data to Google Ads for measuring in-store sales.
Google said Local Customer Optimization is rolling out now, and the Data Manager option is expected in the coming weeks. For retail and franchise marketers, these updates tie digital ad spend more directly to foot traffic and point-of-sale data. The tools reflect Google's ongoing push to make Performance Max campaigns - which already span Search, Shopping, Display, YouTube, and Maps - useful for brick-and-mortar measurement, not just e-commerce.
Attribution data under the microscope
Around the same period, Search Engine Journal published an analysis by Bengu Sarica Dincer arguing that modeled and estimated data increasingly fills gaps left by signal loss in analytics platforms. The piece warned that treating modeled figures as equivalent to directly observed measurements can lead to misallocated budgets. Dincer recommended labeling modeled or estimated metrics as "directional rather than definitive" and comparing results across more than one attribution model.
The timing is relevant. As Google Ads introduces AI-generated summaries of campaign performance, the underlying data those summaries draw from may already include modeled conversions, view-through estimates, or other approximations. Marketers who treat AI-generated narratives as ground truth risk compounding the attribution problem Dincer described.
llms.txt files and ChatGPT Ads benchmarks
In separate reporting, Search Engine Journal covered a Common Crawl review of over 500,000 llms.txt files, which found that 68% originated from a plugin or template and 22.56% contained no links at all. The crawl also discovered that some files included crawler-access rules the llms.txt format itself cannot enforce, and that 136,578 robots.txt files were located at the llms.txt path. Among 32 files that appeared to deny Common Crawl's own bot, none of the 31 it could check actually blocked that crawler in robots.txt.
For marketers paying attention to how AI models access and train on their content, the takeaway is clear: having an llms.txt file does not by itself control crawler access. Enforcement still depends on the rules set in robots.txt.
Another Search Engine Journal report, by Brooke Osmundson, noted that OpenAI has not published performance benchmarks for ChatGPT Ads roughly six months after ads began testing. Advertiser-shared cost-per-click figures ranged from about $3 to over $22 depending on market and campaign type, and the platform still lacks an auction-insights-style reporting view. Without standardized benchmarks, marketing managers evaluating ChatGPT Ads as a channel have little to go on beyond their own test budgets.
Why this matters for marketers
The AI dashboard rollout changes how campaign performance gets communicated inside Google Ads, but it does not change the underlying data quality problem. If you adopt the dashboards, verify whether the summaries treat modeled conversions as observed ones. For multi-location brands, the Local Customer Optimization toggle and Data Manager connection offer a direct line from ad spend to in-store revenue - but the same caution applies to any modeled attribution those tools rely on. And if you are using llms.txt files to manage AI crawler access, check your robots.txt rules first. The file alone won't stop anything.
Your membership also unlocks:
