The strongest AEO signal from the week of May 11-18, 2026 is that the shortcut era keeps losing ground. Google published an AI search guide and used it to say that generative visibility is still built on SEO fundamentals. Ahrefs then published data showing that one of the most repeated AEO shortcuts, "just add schema," did not produce a major citation lift across Google AI Overviews, AI Mode, or ChatGPT.

That combination matters. If the major platforms are making provenance more visible while the data keeps weakening tactic-of-the-week advice, AEO work is becoming more operational. The useful questions are less "what hack are we missing?" and more "can this page be crawled, grounded, attributed, and measured cleanly?"

TL;DR

  • Google's new AI search guide is the week's most important item. As of May 15, 2026, Google is explicitly saying AEO and GEO are still SEO, and it names llms.txt, chunking, and special markup as tactics you can ignore for Google Search.
  • New Ahrefs and Semrush articles push the same conclusion from different angles. Schema is not showing a clear citation payoff, while prompt tracking and AI referral measurement are becoming the more practical discipline.
  • OpenAI and Bing both added more visible attribution or trust cues. That does not solve the attribution problem, but it does make answer-surface visibility more important to audit.

Is Google shutting down the AEO shortcut economy?

Google is doing exactly that, at least for Google Search.

Google says AEO and GEO are still SEO

Source: Google Search Central

Google's new guide is the clearest official AEO document the company has published so far. The most important lines are simple. Google says AEO and GEO are "still SEO" from Google's perspective, and explicitly says you do not need llms.txt, special schema, or special markup just to appear in generative AI search. It also points site owners back to crawlability, snippet eligibility, unique content, and technical clarity.

For AEO, that matters because it narrows the field. A lot of vendor and consultant advice has treated generative visibility like a separate optimization stack. Google is now saying the opposite for AI Overviews and AI Mode. If you are trying to win citations in Google, the priority is still useful content, clean technical foundations, and pages that can be indexed and safely surfaced.

What should you do? Treat this as a filter for your backlog. If a proposed AEO task does not improve crawlability, extractability, entity clarity, or content usefulness, it probably belongs below the line. Our earlier guide on how answer engines discover, retrieve, and cite pages is still the right operating model.

Ahrefs says adding schema still is not a citation lever

Source: Ahrefs

Ahrefs published one of the better anti-hype studies we have seen in this space. It tracked 1,885 pages that added JSON-LD schema, matched them against 4,000 control pages, and measured citation changes across Google AI Overviews, AI Mode, and ChatGPT. The topline finding was blunt: adding schema produced no major uplift in citations on any platform.

That does not mean schema is useless. It still matters for rich results, entity understanding, and general technical hygiene. But the specific AEO pitch that schema is a near-term citation shortcut now looks weaker than ever. Coming in the same week as Google's new guide, the study reinforces a cleaner reading of the market: structured data can support good SEO, but it is not a magic bridge into answer-engine citations.

The practical move is to stop selling schema as a standalone AEO win. Use it where it supports search features, but do not confuse "helpful infrastructure" with "measurable citation boost." If you need the broader context for how pages actually get selected, our guide on how answer engines discover, retrieve, and cite pages is the better internal companion.

Are answer engines making provenance more visible?

Yes, and that is changing what counts as visibility.

ChatGPT is adding more attributed images inside free answers

Source: OpenAI Help Center

In OpenAI's May 12, 2026 release notes, the company says free ChatGPT users will now see more inline images from the web in answers, and that users can click those images to view them at full size and see source attribution. That is a product detail, but it is also an AEO signal. ChatGPT is expanding the answer surface beyond linked text and making attribution more visual.

For publishers and brands, that creates a new visibility layer to watch. A source can now shape the answer through an attributed image even when the text citation is not the main interaction point. That will matter most in product, travel, people, and location queries, where images clarify the answer faster than prose does.

The Monday-morning implication is simple: audit whether your high-value pages have original, indexable images with useful context. If ChatGPT is giving more space to visual attribution, weak image assets become an AEO problem.

Bing is treating trusted-source visibility as part of answer quality

Source: Bing Search Blog

Microsoft's May 13 Bing post is nominally about safety, but it reveals something broader about how answer engines are thinking. Bing describes AI-powered search as a system that interprets intent, generates summaries, and decides what content becomes visible first. The examples focus on Public Safety Announcements, authoritative resources, and SafeSearch controls, but the underlying point is bigger: visibility is now an answer-quality problem, not just a ranking problem.

That matters for AEO because it pushes trusted-source selection upstream. When a system decides that some situations require elevated authority, support resources, or explicit intervention, the content pool is no longer competing only on relevance. It is competing on whether the engine considers it safe enough to surface prominently.

Most commercial sites will not trigger a PSA box. But the principle still applies. If your category touches risk, health, finance, security, or sensitive life decisions, the threshold for citation-worthy evidence is probably rising, not falling.

Is measurement finally getting more practical?

It is getting more practical, but also less comfortable for teams that want one clean KPI.

Semrush says prompt tracking should follow buying moments, not vanity visibility

Source: Semrush

Semrush's prompt-tracking piece is useful because it frames the measurement problem correctly. The article argues that a prompt portfolio should be small, focused, and organized by business impact rather than vanity visibility. It breaks the portfolio into revenue, reputation, competitor, and gap prompts, and says tracking 25 well-chosen prompts beats tracking 500 random ones.

That is a better operating model than the generic "AI visibility score" dashboards now flooding the market. AEO teams do not need more synthetic coverage for low-intent questions. They need a stable set of prompts tied to conversion, brand narrative, and competitive substitution.

If you manage AEO like rank tracking, you will over-measure noise and under-measure decision moments. Build a prompt set that reflects real buyer journeys, then watch which brands get mentioned, which pages get cited, and which prompts you are absent from entirely.

Ahrefs says AI chatbot traffic is still tiny, but not trivial

Source: Ahrefs

Ahrefs' traffic piece adds an important reality check. Using referral data across 74,752 websites, Ahrefs says all AI chatbots combined sent 3.5 million visitors in March 2026, or 0.28% of total web traffic. It also says ChatGPT sent 2.7 million visitors that month, while Perplexity and Gemini each sent around 230,000.

The most useful takeaway is not the size of the channel. It is the distinction between AI crawler traffic, AI mentions, and real human click-through traffic from cited answers. Those are different phenomena, and a lot of reporting still mixes them together. Ahrefs also says its own AI search visitors represented a small share of visits but a much larger share of signups, which fits the broader idea that answer-engine visits can be lower-volume but higher-intent.

For AEO teams, that argues for two dashboards. One should track citations and mentions. The other should track the actual referral and conversion behavior that comes from those surfaces.

What gets the easy story wrong?

The easy story is that Google just killed AEO, schema no longer matters, and AI traffic is finally becoming a normal acquisition channel. None of those reads is precise enough.

Google only killed a certain kind of AEO story: the one built on hacks, renamed SEO basics, and thin technical rituals. Schema still has value for search features, entity clarity, and structured interpretation even if it is not delivering a measurable citation lift in the Ahrefs sample. AI traffic is measurable, but it is still too small to become your main reporting lens. And Bing's trust-heavy framing does not mean every niche is suddenly held to YMYL standards.

The better read is narrower. AEO is maturing into a discipline where source eligibility, entity clarity, and prompt coverage matter more than one-off optimization tricks.

What to do Monday morning

1. Cut every AEO task from your backlog that depends on llms.txt, chunking, or special AI-only formatting as the primary lever. 2. Review three revenue-critical pages for crawlability, snippet eligibility, dated claims, and direct answers under the main headings. 3. Split your monitoring into two layers: prompt coverage by buying stage, and source visibility by answer engine. 4. Audit image assets on pages you want cited by ChatGPT and Google, especially where visuals are part of the answer. 5. Keep using schema where it supports search features and entity clarity, but stop promising it as a short-term AI citation lever.

What to watch this week

  • Watch whether Google expands the AI optimization guide into more reporting guidance. The advice is clearer now than the measurement story.
  • Watch whether more answer engines make attribution more visual, not just more clickable.
  • Watch whether teams that track prompt portfolios, rather than broad AI visibility scores, start finding cleaner revenue signals.

Sources