Google's New AI Guide Just Debunked Your AEO Checklist

Real SEO and AEO fundamentals for an MSP marketing site in late 2026 are the same fundamentals that mattered before generative search existed: a crawlable, indexable site built around clear entities and genuine expertise. Google's own July 2026 guidance confirms this directly, and it also confirms that several tactics MSPs have been sold as AEO extras do not move the needle at all.

What Are MSPs Actually Getting Wrong About AEO?

Most MSP marketing sites treat AEO as a separate discipline from SEO, one that needs its own file, its own markup, and its own writing style. That framing sells well because it sounds like a new service line you can charge for. It also happens to be wrong, according to the company running the largest AI-powered search surface on the web.

What Does Google's New AI Guide Actually Say?

Google's guide on optimizing for generative AI features, last updated in July 2026, states plainly that to appear in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in regular Search with a snippet. There is no separate technical bar to clear. The guide even carries a section built specifically to correct the record.

The guide's mythbusting section lists the exact tactics that have been marketed as AEO essentials and states they are not required. That list includes llms.txt files, forced content chunking, AI-specific schema markup, rewriting pages for AI readability, and inauthentic mention campaigns meant to manufacture brand signals.

Google is explicit that none of these change whether a page shows up in an AI Overview or AI Mode result. The document reframes AEO as an extension of ordinary technical and content SEO rather than a parallel skill set, which matters for how you brief a team or a vendor:

• No llms.txt file is needed for Google's AI features, though Google says it is fine to keep one if another service asks for it.
• No special AI-facing schema is required beyond standard structured data already used for regular Search.
• No mandatory content chunking into short, isolated blocks is needed for extraction.
• No AI-specific rewriting of existing pages is necessary to become citable.

If your marketing site, or a client's, has spent budget on any of those four items expecting an AEO lift, that spend was not buying what it was sold as.

Does llms.txt Actually Help You Get Cited?

No, according to the largest independent test run on the file so far. SE Ranking's analysis of roughly 300,000 domains found llms.txt adoption sitting at 10.13 percent, and found no statistically significant relationship between having the file and how often a domain gets cited in AI answers.

The more telling detail is what happened when SE Ranking removed llms.txt as an input to its predictive model for AI citation frequency: model accuracy improved. The file was adding noise, not signal. Among the fifty most-cited domains in the study's AI answer sample, only one carried an llms.txt file. For an MSP deciding where to spend a marketing hour this quarter, that is a fairly direct answer.

Why Did Google Just Kill FAQ Rich Results?

Because the markup type became a spam vector, not because FAQ content stopped mattering. FAQ rich results stopped appearing in Google Search on May 7, 2026, with the Search Console reporting for them retired the following month and API support scheduled to end in August.

Per Search Engine Roundtable's reporting on the change, Google engineer Ryan Levering explained the reasoning directly, saying FAQ markup had gotten out of control and was making structured data look like a spam vector, in the same category as fake review-star abuse. Sites had been bolting generic FAQ blocks onto pages that were not genuinely FAQ content, purely to claim extra search real estate. Google had already restricted eligibility to authoritative government and health sites back in 2023, and this year it finished the retirement rather than continue policing it.

The lesson is not that structured data stopped mattering. Schema.org itself keeps expanding. Version 30.1 shipped on September 16, 2026. Its own usage statistics dataset is now published for anyone to study adoption patterns across the vocabulary. The lesson is that markup applied to game a result, rather than to describe content that is actually there, eventually gets pulled.

Is There a Shortcut Standard Coming for AI Crawlers?

Not yet, and that gap is itself informative. The IETF's AI Preferences working group was chartered to make it easier to express how AI models should use internet content, originally aiming to submit its draft standards to the IESG by August 2025. That date came and went. As of September 2026, the group's core vocabulary draft is still an active Internet-Draft that has not been submitted to the IESG and is explicitly marked as not reflecting working group consensus.

That means there is still no finished, adopted protocol that lets a site express AI-specific preferences beyond the voluntary conventions crawlers already choose to honor or ignore. Anyone selling you a compliance checklist against a standard that has not shipped is selling you paperwork, not results.

What This Means for Your MSP Marketing Site

Put the developments together and the fundamentals reassert themselves. Google's guide, the FAQ retirement, and the llms.txt data all point in the same direction: effort spent on AI-only workarounds is effort not spent on the things that were always the actual ranking and citation drivers.

For a trust-sensitive buyer like an MSP prospect, that last row carries extra weight. Buyers evaluating a managed services provider are already primed to distrust anything that looks manufactured, so a site built on genuine service descriptions, named team expertise, and verifiable client outcomes reads as credible to a human and remains eligible for AI-generated answers for the same reason. You are not optimizing for two audiences with different rules. You are optimizing for one audience judged by two overlapping systems.

In practice that means the highest-value fundamentals work left on most MSP sites is unglamorous. It is making sure every service page is actually indexed and crawlable rather than orphaned behind a broken internal link. It is naming the engineers and analysts behind a page instead of leaving copy anonymous, so entity signals tie real expertise to your domain. It is keeping structured data accurate and limited to content that genuinely matches the schema type, so nothing you publish ends up in the next abuse sweep the way FAQ markup just did. None of that is new advice. What is new is having Google's own documentation, a 300,000-domain study, and a retired rich result all confirming the same thing in the same year.

Turning Fundamentals Into a Sellable Service

This is where the fundamentals stop being defensive reading and start being a service line. Every MSP still has clients whose sites carry an abandoned FAQ schema block, a chunked-up blog nobody enjoys reading, or an llms.txt file some vendor upsold them on last quarter. Auditing against what Google actually says, rather than against AEO folklore, is a concrete, repeatable engagement you can package and resell.

That is the exact gap a white-labeled grader like ActiScore is built to score against, since it checks a site's real indexability, structured data hygiene, and entity signals rather than the fads that come and go each quarter. If you are not sure which of these fundamentals your own site or a given client's stack is actually missing, running it through the stack builder is a faster starting point than guessing from a vendor's pitch deck, and it will point you toward where a tool actually closes a gap versus where you already have it covered.

None of this requires betting on a standard that has not shipped or a file format Google itself says is optional. It requires doing the boring, verifiable work well, then being able to prove to a client that you did it with a report, not a promise. Pair that audit habit with the rest of the product catalog and you have a repeatable offer built on evidence instead of hype, one that holds up even after the next AEO trend gets debunked the way llms.txt and FAQ schema stuffing just were.

See the full stack to see how an audit line like this fits alongside the rest of what you can already resell under your own brand.