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 Business Profile products have been stuck in Pending since late July
Merchants are reporting that products in Google Business Profile sit in "Pending" status and never go live, with the behavior running for several days from around July 27. It affects newly added products, and in multiple reports it has also pulled previously approved products back into Pending. The reports come from independent merchants rather than one account or one vertical, and Google has neither acknowledged the problem nor offered a fix.Source: Search Engine Roundtable →
Our take: Wait it out. Approved products reverting across unrelated accounts is a queue problem on Google's side, not a policy flag on your listing, and deleting and resubmitting adds work without moving you up the queue. Check the status again next week.
Read our full take →Google dropped the 2007 rule about blocking internal search pages, then said keep blocking them
John Mueller confirmed on Search Off the Record that Google Search Essentials no longer lists blocking internal search-results pages as a requirement, retiring a rule that dated to the 2007 webmaster guidelines, which classified those pages as automatically generated content or "infinite spaces." "Nowadays, we don't have that listed in the search policies," Mueller said. He recommended blocking them anyway for two reasons unrelated to policy. The first is crawl budget and server load: uncached search pages force a database lookup on every fetch, and a flood of new URLs can push Google to reduce crawling across the whole site, which Mueller illustrated with "you can imagine what happens when we see, I don't know, 100 million pages that are new from Martin Splitt; we'll go off and try to crawl those." The second is spam, since indexable search pages let attackers surface queries for adult terms and phone numbers that make a site look hacked and can trip Search Console flags. On method he suggested a single broad robots.txt rule such as "/search/?" rather than enumerating patterns, noted that noindex still permits one crawl before blocking, and warned against using Search Console removal tools or 500 error codes for this. He carved out an exception for category pages powered by search functionality, which he said warrant indexing for their structural value.Source: PPC Land →
Our take: Nothing to do here, which is the point. The documentation caught up with reality: Google stopped enforcing this as a rule long before it stopped printing it, and every reason to block was always independent of the rule. Mueller's carve-out for search-powered category pages is where stores get hurt. Plenty of storefronts generate category listings through the site's own search engine, and a blanket /search/ disallow takes the catalog down with the junk.
Read our full take →From September 15, your CDN may start blocking AI crawlers whether your robots.txt does or not
IAB Australia published a Bots and Crawler Guidance and Decision Matrix that sorts every automated visitor into five job categories — Discovery & Search (Googlebot, bingbot), AI Training (GPTBot, ClaudeBot), Live AI Agents (ChatGPT-User, Claude-User), Operational & Advertising (AdsBot-Google, verification vendors), and Unknown/Unverified — and applies one of four verdicts: allow, allow with conditions, require licensing, or block. The deadline inside it matters more than the framework. Before 15 September 2026, new domains and free-tier zones sitting behind Cloudflare get Training and Agent bots blocked by default on any page that displays advertising — a decision made at the edge, not in your robots.txt. Cloudflare data cited in the guidance puts automated requests at 57.5% of web traffic as of June 2026, the first recorded crossover past human traffic. The guidance is also careful about something worth repeating: one operator often runs several bots for several jobs, so blocking a training crawler does not necessarily pull your content out of the AI answers built on the same index.Source: IAB Australia →
Our take: If your store runs on Cloudflare's free tier and shows ads, put September 15 on the calendar. Log in, open the bot settings, and make the AI-crawler decision yourself instead of inheriting the default. A block at the edge is invisible to every SEO tool that only reads robots.txt.
Read our full take →AI Mode shows about 95% fewer product listings than standard search — and barely the same ones
Productrise tracked more than 100,000 identical shopping queries and over two million product listings across 21 days in July 2026, in the US and UK. Standard search returned at least one product on roughly 88% of queries, averaging 22.5 listings per page; AI Mode cleared that bar on roughly 23%, averaging 4.3 — about 95% fewer listings overall. The number that should change how you read it is the overlap: only about 0.8% of products visible in standard search also appeared in AI Mode for the same query on the same day. These are not trimmed-down Shopping results, they are a largely different slice of the same catalog, which means Shopping visibility does not carry over to AI Mode on its own. The study notes its own limits — keyword queries rather than conversational prompts, and only the two markets where the Shopping Graph currently feeds AI Mode.Source: PPC.land →
Our take: The 0.8% overlap means ranking in Shopping today tells you almost nothing about whether AI Mode will show your product. Treat it as a separate surface with its own inputs: complete attributes, clean feed data, schema that agrees with the page.
Read our full take →AI Overviews now appear on somewhere between 43% and 48% of Google searches
Two trackers, two numbers: Similarweb puts AI Overview presence at roughly 43% of Google queries, Semrush at roughly 48%, with informational queries running higher still. The five-point gap is itself informative — nobody outside Google measures this exactly, so treat any single presence figure as an estimate rather than a metric. Both agree on where this is heading: close to half of searches now resolve with an AI answer sitting above the links.Source: Search Engine Roundtable →
Our take: Half of your customers' searches now start with an answer instead of ten links. The response isn't panic, it's checking whether the queries you sell against trigger AI Overviews (Search Console's generative-AI report shows this) and whether you or a competitor is the one being cited.
Read our full take →Mueller: there’s no precedence order when your page, your markup, and your feed disagree
Asked which value Google trusts when a product page reports one availability in its server HTML, another after JavaScript renders, and a third in the Merchant Center feed, John Mueller answered that there is “no publicly defined order of precedence for metadata reconciliation” — and that this holds across visible content, structured data, sitemaps, and feeds alike. His advice was blunt: “fix it, not analyze if it’ll work regardless.” For a catalog, that’s the whole point. Availability drifts out of sync easily, and there is no hidden tiebreaker to fall back on, so three-way consistency between feed, page, and markup isn’t a best practice — it’s the only defined behavior.Source: Search Engine Journal →
Our take: Run the three-way check on your top sellers this week: feed availability and price versus the page versus the markup. On BigCommerce the usual drift is a stale feed after a price change, and five minutes in Merchant Center diagnostics catches most of it.
Read our full take →Search Console’s Platform Properties are now live worldwide
Google confirmed that platform properties — performance data for the content you publish on Instagram, TikTok, X, and YouTube — are now globally available in Search Console, alongside a new help doc covering the Insights report and data exports. It tells you nothing new about your own site’s rankings, but for a brand whose products get found on social as much as in Search, it puts both sets of numbers behind one login.Source: Search Engine Roundtable →
Our take: If your brand publishes on YouTube or Instagram, connect the platform property once. You get Google search performance for that content next to your site's, which is how you prove whether social content earns search impressions or just likes.
Read our full take →Ahrefs: Google punishes bad content, not AI content — but the gaps are real
Ahrefs analyzed roughly 331,000 pages — drawn from a million top-10 results across 100,000 searches, plus 80,861 pages tracked in Search Console over a year — and found AI-written pages ranking fine: 9% of top-ranking pages are at least 80% AI-generated, and 5.3% of top-3 positions are fully AI-written. The reassuring headline hides two real gaps, though. Pages with very high AI content were indexed 40.35% of the time versus 49.28% for low-AI pages, and low-to-moderate AI pages pulled 2–3x the organic impressions of the heavily AI-written ones. Author Ryan Law’s reading is that heavier AI use correlates with worse content — repetition, no multimedia, flat academic prose, inaccuracies — rather than AI use being penalized as such. That matches what the indexing data has said all year: Google is filtering for generic, whoever wrote it.Source: Ahrefs →
Our take: The catch is the indexing gap: heavily AI-written pages get in the door less often. Use AI for the draft, then force in what AI leaves out — real numbers, product specifics, an actual position. That's the difference between the 9% that ranks and the rest.
Read our full take →Changing an unavailable_after date does nothing until Google re-crawls the page
Google’s Gary Illyes confirmed that pushing an unavailable_after expiry date forward is fine, but the new date only registers after Google fully re-fetches the page — “we rarely see HEAD requests so we’d still download the content.” For inventory that lives 24–72 hours, that re-crawl lag means Google can be holding a stale expiry the whole time the offer is live. Illyes also scoped the directive tightly: it governs serving eligibility only, not crawl budget and not 404 handling — the classifieds operator who raised it was still seeing more than half of Googlebot’s requests hit 404s, which no expiry tag fixes. Worth knowing before you reach for it on flash sales, event pages, or fast-turning catalog listings.Source: PPC.land →
Our take: If you run flash sales or fast-turning inventory, don't lean on expiry tags — Google can hold a stale date for the offer's whole life. Serve a real 404/410 when something is gone for good, or keep the URL stable and rotate the offer on it.
Read our full take →Search Console’s page-indexing report keeps dropping back to weekly data
Since June 11, Search Console’s page-indexing report has intermittently stopped showing daily data and fallen back to weekly grouping. Google has fixed it several times without shipping a permanent resolution. If your indexing chart looks flat or coarse this week, check the granularity before you read it as a real change in how much of the site is indexed.Source: Search Engine Roundtable →
Our take: Before you diagnose an indexing problem this month, check the chart's granularity — weekly grouping looks exactly like a stall. Cross-check against the sitemap-level report before you act on it.
Read our full take →Google hit a 20-year-old forum with a thin-content manual action — and SEOs suspect its AI chatbot
Google issued a partial manual action for “thin content with little or no added value” (message WNC-651700, dated around July 17) against the /threads/ section of Windowsforum.com — roughly 168,000 visible threads and 500,000+ posts built over 20 years — with no example URLs attached. SEO Gagan Ghotra suggested the real target is the forum’s AI chatbot staff account, which has posted more than 111,000 replies since March 2023; Google hasn’t confirmed that, and the site owner notes two more obviously thin sections (a syndicated headline feed and an archived knowledge base) were left untouched. The theory is speculation, but the underlying point isn’t: content an AI assistant publishes on your domain is indexable content Google can judge — and act on — like any other.Source: Search Engine Journal →
Our take: Whatever triggered it, the lesson stands: anything an automation publishes on your domain is your content in Google's eyes. If a chatbot or AI assistant writes pages on your store, audit its output the way you'd audit an employee's.
Read our full take →Search Console’s robots.txt recrawl request is erroring — Mueller sees it too
An SEO reported that Search Console’s “Request a recrawl” function for robots.txt was returning an “Unknown error” across several properties, and John Mueller confirmed on Bluesky that he saw the same failure on some of his own sites. It hasn’t been confirmed as a broad, platform-wide bug. If you’ve just edited a robots.txt file and the recrawl request fails, the error is reproducing on Google’s side too — log it and move on rather than tearing apart your own setup looking for the cause.Source: Search Engine Roundtable →
Our take: Nothing to fix on your side — but if you just edited robots.txt and the recrawl button fails, wait rather than re-editing. The file gets picked up on Google's normal cadence anyway, usually within a day.
Read our full take →Rank trackers lit up again over the July 24–26 weekend
Search Engine Roundtable reported elevated ranking volatility beginning late Thursday July 23 and intensifying through the weekend of July 24–26, describing it as another unconfirmed Google ranking update. The Google Search Status Dashboard showed no incidents and all services green at the time of the report, so Google hasn’t acknowledged anything. If a client’s rankings moved this week, log the movement and let it settle — rewriting pages against a signal Google hasn’t confirmed is how you end up chasing noise.Source: Search Engine Roundtable →
Our take: Log it, don't chase it. If rankings moved during an unconfirmed shuffle, don't rewrite pages against a ghost. Compare again after a week before touching anything that currently ranks.
Read our full take →A Googlebot section in robots.txt makes Google ignore your user-agent: * rules
John Mueller explained that robots.txt precedence is specific-beats-general: if your file contains a “user-agent: Googlebot” section, Google reads only that section and ignores the “user-agent: *” rules entirely — so disallows you wrote for all bots get silently skipped for Googlebot. The case that surfaced it was a store whose search-results pages kept getting indexed because the disallow lived in the wrong block. Fix it by listing the user-agents together above one shared set of rules, repeating every rule inside the Googlebot section, or dropping the bot-specific section altogether. And for search pages specifically, noindex does the job robots.txt can’t.Source: Search Engine Journal →
Our take: Pull up your robots.txt and look: if there's a user-agent: Googlebot section anywhere, Google ignores every rule in your user-agent: * block. Stores inherit this from old plugins and platform defaults more often than anyone expects.
Read our full take →Google writes fake and undisclosed incentivized reviews into the review-snippet rules
Google added a guideline to its review-snippet structured-data documentation: don’t include fake or undisclosed incentivized reviews on the page or in the markup. The examples given are reviews written without genuine experience of the product or service, and reviews exchanged for money, discounts, vouchers, or free products without disclosing the incentive. It reads as formalizing an existing expectation rather than a new rule — but it now sits in the docs that govern review-snippet eligibility, so if you run an incentivized-review program, the disclosure is what protects your rich results.Source: Search Engine Land →
Our take: If you incentivize reviews — discounts, points, free product — the disclosure is now part of rich-result eligibility, not just FTC hygiene. Build the disclosure into the review-request flow and the markup stays clean on its own.
Read our full take →Google now emails you when a Business Profile review gets rejected
Google now sends an email notification when a review someone submitted to your Business Profile is rejected for breaking Google Maps content policies, instead of letting it quietly disappear. It gives a business a clearer signal that a legitimate review was pulled — and a concrete starting point for an appeal — rather than wondering why a review never showed up.Source: Search Engine Roundtable →
Our take: Those emails are appeal fuel. When a genuine customer review gets pulled, you now hear about it — keep the notice and appeal instead of shrugging. Reviews are an AI-answer input now, not just stars on a listing.
Read our full take →Semrush: only 15.2% of ChatGPT topics have a clear brand owner
A Semrush analysis of 1,094 topic categories found that a single brand held dominant ownership — appearing in at least four of five related prompts with a five-point-or-greater lead — in just 15.2% of them. The practical takeaway for AI-search visibility: one favorable ChatGPT mention rarely means you own a topic, so measure presence across a spread of related prompts rather than a single lucky answer.Source: Search Engine Land →
Our take: One good ChatGPT mention isn't ownership. Measure across a spread of related prompts before claiming — or conceding — an AI surface. That's how we run visibility checks: real buyer questions, several platforms, a presence grid rather than one lucky answer.
Read our full take →
Custom ecommerce software, AI apps, and SEO — the work other agencies quote around, built in-house.
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.