SEO Recap covering July 15, 2026

7 min read·13 stories

Daily summary of what matters in SEO, GEO, AEO, and AI search generated with Claude Code (beware of hallucinations)

Generative Engine Optimization (GEO) & Answer Engine Optimization (AEO)

AI agents pull most B2B website content directly, but pricing questions often route them to third-party sources.
The gap stems from pricing being hidden, gated, or absent from crawlable pages.
The piece recommends publishing clear, retrievable pricing information so agents cite your site rather than competitors or aggregators.

Invoca's July 13 report puts the lead rate for ChatGPT-referred calls at 49%, about 10 points above the seven-channel average and 6 above Google Business Profiles at 43%.
Those leads convert at 40% on the call, essentially the all-channel average of 42%, so the edge shows at qualifying but not at closing.
Figures come from over 70 million calls across 10 industries; Invoca doesn't disclose ChatGPT call volume and notes overall AI-driven volume remains very low.
Gemini, Claude, and Perplexity weren't measurable in the dataset, and 64% of businesses never ask callers to buy or book.

GA4 added a native AI Assistant channel on May 13, but the same source (chatgpt.com) can split across AI Assistant, Referral, and Unassigned because channels are set by source and medium together.
The recognized platform list keeps shifting: Claude was quietly dropped and Perplexity still lands in Referral, so don't hard-code a platform list into reports.
The fix is a custom channel group matching on source only, which merges the fragments, applies retroactively, and lets you add platforms Google omits.
Even a perfect channel can't capture referrer-less app traffic (Direct) or AI Overviews and AI Mode, which Google counts as organic; Similarweb clickstream shows ChatGPT referrals converting around 7%.

High-CPC searches were far more likely to trigger AI text ads in Google's AI Mode.
Paid placements rarely boosted a brand's citations or organic rankings within AI responses.
The finding suggests advertisers cannot buy their way into being cited by AI Mode; earned visibility and paid ads remain separate levers.

A new report finds Google-owned business and product cards showing up more prominently in AI Mode.
Local and shopping searches see the strongest presence of these Google-owned modules.
Brands relying on local and product queries should track how much AI Mode real estate goes to Google's own surfaces versus third-party links.

Robby Stein said Personal Intelligence in AI Mode now connects to Google Calendar and can add invites or meetings directly.
Calendar is the first Personal Intelligence connection that can create entries, not just reference data like Gmail and Photos.
iPullRank's May report found that connecting Gmail changed which brands appeared in AI Mode for identical prompts, and Calendar adds timing as a new personalization variable.
The connection is live in the U.S. now, with more countries to follow; the same query can now produce different answers per user.

AI Overviews can now generate custom images from a text prompt directly in the search response, using Google's latest Nano Banana model.
The feature is rolling out over the coming weeks in English across regions that already support image creation in AI Mode.
Barry Schwartz warns it will likely reduce clicks to publisher links and to Google Images.

John Mueller said there is no penalty or demotion for serving varying content, though it can make debugging and monitoring harder.
If test variants are close enough, indexing is unaffected; if significantly different, one version may be indexed and show in results.
Google's guide still warns that unnecessarily long experiments serving one variant to most users may be read as an attempt to deceive search engines.
Best practices: use rel=canonical, 302 (not 301) redirects for random variant serving, avoid cloaking, and limit test duration.

John Mueller said that if A/B variants differ significantly, that difference could become visible in search results, and one version may be used for indexing.
He confirmed there is no known penalty or demotion for varying content, but constant changes make debugging and monitoring harder.
Google's 2022 A/B testing guide notes small changes like button color, size, or CTA text usually have little or no impact on snippets or rankings.
Anyone running tests who cares about SEO should follow Google's official A/B testing guide.

The EU decision could force changes to how Google surfaces its own comparison, travel, and shopping services on high-value queries.
Comparison sites, travel platforms, and shopping services stand to gain visibility if self-preferencing is curbed.
Practitioners in affected verticals should watch for shifts in SERP layout and referral traffic tied to any mandated changes.

The author argues AI has moved buying decisions beyond the single click, blurring the line between search channels.
SEO and PPC are framed as complementary parts of one strategy rather than competing budget choices.
Practitioners should coordinate organic and paid efforts around the full buying journey instead of pitting them against each other.

The author observes that discovery is often emotional and starts on platforms like TikTok, with Google used later to validate the choice.
This reframes Google's role from first touchpoint to verification step in the buying journey.
SEO should account for branded and validation-stage searches, not just top-of-funnel discovery queries.

The patent (Contextual estimation of link information gain) describes scoring a document by how much new information it adds beyond pages a user already saw, and has been cited 24 times with US protection extended to 2039.
Documents that add little novel information may be reranked, demoted, or dropped for a user's next set of results; Google also runs OriginalContentScore and ContentEffort systems.
Only 32% of Google searches end in a click and 66.61% of those go to the open web, still about 2.92 billion open-web clicks a day.
Kevin Indig's research is cited: first-party research is rare in AI citations but earns 3.3x more, and original data is the strongest predictor of page originality.

Published on July 16, 2026