Google Just Published Their AI Search Playbook. We Already Do All of It.

· Chris Dolan

Google published their official AI Optimization Guide on May 15th. If you've been in SEO circles this week, you've seen the takes - add llms.txt, chunk your content, rewrite everything for AI.

Every one of those is wrong. Google said so themselves. And more importantly: the things they actually recommend are exactly how we built RankingFactory.

Let me walk through what they said, why it matters, and how the system we've built quietly checks every box - in ways most tools can't.

What Google Actually Said

The guide has three real pillars. Everything else is noise.

1. Create non-commodity content with a unique point of view. Not "7 tips for homebuyers." Something that could only come from someone who has done the thing, knows the subject, and has a credible reason to be the one writing it.

2. Build a clear technical structure. Crawlable pages, indexed with snippets enabled, semantic markup, fast load, no duplicate content.

3. Build real entity and business presence. Google Business Profile, structured data, consistent NAP - signals that confirm a business is what it says it is, everywhere they look.

That's it. Three things. The rest of the guide is Google telling you to stop doing the things that don't work.

What Doesn't Work (According to Google)

This section is the most important thing in the guide and the least discussed.

llms.txt files. Google explicitly says you don't need them. They may discover and index the file - it isn't treated specially. We have one, because AI crawlers from other providers (Perplexity, Claude, GPT) do respect it. But it won't move your AI Overview position on Google.

Chunking content. Google's systems understand full pages. Breaking your content into fragments doesn't help and adds maintenance overhead.

Rewriting for AI keywords. AI systems understand synonyms and intent. You don't need to engineer around exact-match queries. You need to write well.

Inauthentic mentions. Solicited reviews, fake forum posts, manufactured citations - Google's spam systems catch these. They actively hurt you.

Obsessing over structured data. Schema.org markup isn't required for AI features. It helps with rich results. It should be part of your technical baseline - not your entire strategy.

How RankingFactory Handles the Three Pillars

Pillar 1 - Non-Commodity Content

The blog pipeline in RankingFactory doesn't just generate articles. It generates articles with authors - and those authors matter.

Every post we publish is attributed to a named person with a job title, a bio, a photo, and optionally a real LinkedIn profile URL. That attribution is wired into the page's JSON-LD as a Person schema object, the same schema Google's systems use to cross-reference credentials. The author page at /blog/author/{slug} surfaces their full profile - verifiable, indexable, linkable.

We recently added two tools in the admin panel that take this further. If a client wants to use their real identity - their actual expertise, their real LinkedIn - they can paste their profile text directly and the system extracts their name, role, bio, and LinkedIn URL automatically. If they're running an agency persona, the AI Generate Personality tool builds a credible specialist profile from scratch. Either way, every post is signed by someone. Not by "the team."

That's E-E-A-T. Experience. Expertise. Authoritativeness. Trustworthiness. Google's guide doesn't use the phrase once - but the entire pillar is about it.

On the content itself: the pipeline generates posts from topics you control, in a publication cadence you set, with review gates before anything goes live if you want them. The auto-publish warning in the admin panel exists for a reason - mass publishing without human review is the fastest way to produce commodity content at scale. We flag it. We don't hide it.

Pillar 2 - Technical Structure

This one is invisible when it works and catastrophic when it doesn't.

Every blog post in RankingFactory renders:

  • JSON-LD with BlogPosting schema - title, description, author (full Person object), publisher (Organization), publish date, modified date, and a speakable selector pointing at the actual visible excerpt on the page
  • Open Graph and Twitter Card meta tags - injected dynamically when you navigate to a post, not just set once in the HTML template
  • Canonical tag - set to the post's own URL to prevent duplicate indexation
  • Sitemap - auto-generated and served at /sitemap.xml, updated when posts are published

The speakable property is worth calling out specifically. It tells Google which part of the page to read aloud in voice search and which part to feature in AI Overviews. Most SEO tools don't implement it at all. We implement it on every post, pointed at the excerpt paragraph that appears directly below the post headline - exactly where Google expects to find the page summary.

Pillar 3 - Entity and Business Presence

This is where RankingFactory goes beyond the guide.

Google's advice is to use GBP and Merchant Center. That's the floor. What we've built is an entire entity signal network using Google's own properties as the distribution layer.

DriveStack generates Docs, Sheets, Slides, and Calendar events pointing at your entity - each one referencing the same NAP data, the same business URLs, the same branded anchor text. CloudStack creates HTML pages across multiple cloud providers reinforcing the same entity footprint. GBPStack automates posts, Q&As, and service attributes to the Business Profile directly.

SmartStack - the adaptive engine that runs nightly - measures the strength of that entity footprint with an Entity Score, then decides what's missing, what's stale, and what to repair. Drive permissions break. Blogger hits rate limits. Tokens expire. The repair engines catch all of it automatically, before it shows up as a gap in your signal profile.

The result is a business that looks coherent to Google from every angle they check: the website, the GBP, the structured data, the indexed documents, the linked properties. Incoherence is what tanks AI visibility. Coherence is what builds it.

The Real Lesson From This Guide

Google didn't publish a list of new tricks. They published a defence of the fundamentals.

The signal that matters most - the one that cuts through both traditional search and AI Overviews - is whether a real entity with real expertise has published something genuinely useful, on a technically sound page, backed by a coherent business presence across the web.

That's been the answer for years. AI search didn't change it. It made it harder to fake.

We built RankingFactory for that world. Every pipeline, every signal engine, every repair job, and every blog post - all of it is pointed at the same outcome: a business that Google can find, understand, trust, and cite.

Want to see how your current entity strength stacks up? Run a free AI visibility audit - it checks your schema, your GBP, your indexation, and your structured data against the signals that AI search surfaces actually look for.