July 2026 — SEO, AEO & GEO news.
What changed, as it happens.
Tracked daily from primary sources. The items below are what we're watching for clients right now.
Google adds AI image generation inside AI Overviews
Google is adding AI image generation directly into AI Overviews responses in Search, so an answer can render a generated image alongside its text — rolling out over the coming weeks. It pushes the AI answer surface further toward being a rich, self-contained destination rather than a stepping stone to links.Search Engine Journal →
Mueller: big A/B-test differences can surface in search — but long tests aren’t a penalty
John Mueller said that when the two versions in an A/B test differ significantly, Google can reflect that difference in search results and may use one version for indexing. He separately clarified that running an A/B test for a long time does not, by itself, cause an SEO demotion — useful reassurance if you run extended experiments on key pages.Search Engine Journal →
Mueller on “first-link priority” and hiding internal links
Mueller addressed a technical-SEO scheme that obfuscates homepage links so that only one internal anchor text is counted toward a page (“first-link priority”). The takeaway: engineering around how Google reads internal anchors is fragile and easy to get wrong — clear, honest internal linking is the durable approach.Search Engine Journal →
Merchant Center adds a beta “AI Performance Insights” report for AI Mode and AI Overviews
Google added a beta report under Merchant Center’s Analytics → Products tab showing share-of-voice, competitor share, and query frequency and type for how a brand’s products surface in AI Mode and AI Overviews. It covers organic-AI traffic only, with no clicks metric yet, and is piloting in the US before expanding to Australia, Canada, India, and New Zealand — the product-feed-side counterpart to Search Console’s AI performance reporting.Search Engine Roundtable →
OpenAI clarifies its crawler user-agent versioning and a robots.txt marker
OpenAI updated its crawler help documentation to note that the OAI-SearchBot user-agent version number may change over time, and to describe an additional robots.txt-specific marker included in the user-agent string. There’s no new action for site owners, but it’s worth a note if you match crawler user-agents by exact string when managing AI-crawler access.Search Engine Roundtable →
Study: AI-visibility rankings are mostly noise until you sample dozens of times
A forthcoming preprint, corroborated by an unaffiliated University of St. Gallen paper, tested AI visibility and citation-share rankings from SearchGPT, Gemini, and Perplexity and found them wildly unstable run to run: across 30 platform-topic tests, rankings needed 33–94 sampled answers to stabilize, and 3 of 30 never settled within 125 questions. The practical takeaway — treat a single-prompt or small-sample AI-“share” check as directional noise, not a ranking, and query dozens of times before acting on it.Search Engine Journal →
Google says fixing a canonicalization issue can take up to two weeks
Google added a section to its “Fix canonicalization issues” help documentation noting that resolving a canonicalization problem can take up to two weeks, depending on how the page’s cluster is structured. Useful expectation-setting if you’ve consolidated duplicate or near-duplicate URLs and are waiting on Google to settle on the canonical.Search Engine Roundtable →
Search Console’s Generative AI controls expand to the US and beyond
Google is extending the Generative AI controls in Search Console — which let site owners set include, exclude, or inherit states for how their content is eligible to appear in AI Overviews, AI Mode, and Discover generative features — beyond the UK to sites in the US and other countries, after roughly a month of UK-only availability. The controls govern eligibility only and Google has said they carry no ranking impact.Search Engine Roundtable →
Google appears to be testing google.com/goto passthrough tracking on result links
Some Google search-result links are being spotted routing through a google.com/goto tracking URL instead of pointing straight at the destination page. First noticed around June 23 and since corroborated by a handful of practitioners, the behavior is unconfirmed by Google and not yet consistently reproducible.Search Engine Roundtable →
Google Business Profile appeals now take evidence uploads inline
Google redesigned its Business Profile suspension-appeal flow so businesses can attach supporting documents directly in the appeal form via a single multi-file upload step. It replaces the old process, which required submitting evidence through a separate form within a 60-minute window after filing the appeal.Search Engine Roundtable →
Search Console adds “Platform Properties” to track social and video content
Google launched a new Search Console property type that lets creators and publishers verify Instagram, TikTok, X, and YouTube accounts — even without a website — to see how that content performs in Search and Discover. It builds on a December 2025 experiment and is rolling out gradually over the coming weeks.Search Engine Roundtable →
Google adds Product.category and “Sale duration” to its Merchant listing schema docs
Google updated its Merchant listing structured-data documentation with a `Product.category` property and a new “Sale duration” section covering validFrom, validThrough, and priceValidUntil. The additions bring on-page schema markup into line with attributes already used in Merchant Center feeds — worth syncing if you hand-roll product schema.Search Engine Roundtable →
Mueller: a well-built site already serves AI agents — parallel Markdown just adds debt
John Mueller said a properly built website already works for AI agents, search engines, and human visitors alike, so maintaining a separate Markdown or llms.txt version mainly creates ongoing rework. It continues a consistent line from Mueller and Martin Splitt against parallel AI-oriented site versions.Search Engine Journal →
Study: 62% of AI brand recommendations don’t survive a single follow-up question
A Clovion AI study of 69,120 multi-turn conversations across ChatGPT, Gemini, and Claude found 62% of AI brand recommendations no longer held up after one qualifying follow-up from the user. It also found Claude and ChatGPT tended to underclaim product features while Gemini overclaimed them — a reminder that AI visibility rides on defensible, specific product facts.Search Engine Journal →
Mueller: Cloudflare’s Content-Signal directive has “no effects whatsoever”
John Mueller said Cloudflare’s Content-Signal robots.txt directive — meant to express AI training, search, and input preferences — has “no effects whatsoever for any crawler or LLM,” calling it unnecessary bloat, and reconfirmed Google doesn’t use llms.txt or llms-author.txt files. Another data point that AI-specific config files aren’t the lever; crawlable content is.Search Engine Roundtable →
Custom ecommerce engineering, AI apps, and SEO — typically 60–80% less than traditional agency pricing.
20+ years of BigCommerce engineering, now AI-augmented. Tell us your store, your stack, and your deadline — we quote fixed scope on the first call.