Get my diagnosis
GEO checklist

The GEO Checklist: What to Check Before You Publish (or Fix What's Already Live)

Eleven checks, three groups, no fluff. Run it before a post goes out, or run it against everything you've already published — it works either way.

The short answer

This is a short, honest checklist covering whether AI bots can reach your content, whether it's structured in a way that gets quoted, and whether it's kept current — plus the one item almost nobody checks: whether the content links back to a product. Pass every item and you've removed the obvious blockers. That's what this list can promise, and no more.

Most GEO advice is written as a wall of theory. This is written as a list you can actually run down, item by item, against a post you're about to publish or one that's already been live for two years. Where a topic has a full guide behind it, it's linked inline — this page is the map, not the whole territory.

Can AI even reach it

Nothing below this matters if the bots can't get past the front door.

  1. robots.txt doesn't block the AI bots. GPTBot, ClaudeBot, PerplexityBot and the rest each need an explicit allow — not just the absence of an obvious block. This is often broken by an inherited theme or a CDN rule nobody remembers setting.

    → Run the free test
  2. llms.txt exists at your domain root and matches reality. A short, honest map of what you sell and your main categories — not a copy of your homepage marketing copy that the site itself doesn't back up.

    → Full guide → Generator
  3. Nothing carries a stray noindex or leftover Disallow. A common casualty of a staging-to-production migration: a rule meant to hide a test environment quietly ships to the live site and nobody notices for months.

  4. Your sitemap is current and submitted. An assistant that leans on Bing or Google's index can't cite a page that index has never seen, no matter how good the content on it is.

Is it formatted to be quoted

Access solved, the next question is whether an assistant can lift a clean answer out of the page.

  1. Every page answers its own question in the first few lines. Before the backstory, before the SEO throat-clearing — the direct answer, high enough that a model can quote it without reading the rest.

    → Full guide
  2. Structured data matches what's actually on the page. FAQPage, Article, Product — whichever applies. Schema that describes content the page doesn't have is worse than no schema at all.

    → Full guide
  3. Headings read as questions or direct claims, not labels. "Why this happens" beats "Overview." And dense paragraphs give way to lists and tables wherever a list or table would say it faster.

  4. You could explain, in one sentence, what GEO is actually asking of your content. If that's fuzzy, everything above it on this list will be too — this is the framing that makes the rest make sense.

    → What GEO means → AI search optimization

Does it lead anywhere commercially

The part most GEO checklists skip entirely, because most of them are written for blogs with nothing to sell.

  1. Old posts get revisited on a schedule, not left to rot. A page dated three years ago with three-year-old prices or a discontinued product mentioned by name undermines every other item on this list — stale content reads as unreliable, to a person and to a model.

    → Full guide
  2. You know whether assistants already mention you. Not a guess — an actual answer to "does ChatGPT, Perplexity or Gemini bring up my store when someone asks a relevant question." Most people running a blog have never checked.

    → Full guide
  3. The one almost nobody checks

    Your blog posts link to an actual product or category page — not just to each other. A post can pass every item above, get cited by an assistant, bring a reader to your site, and still make you no money if it dead-ends instead of pointing at something you sell. This is the gap we see most often, and it's the one this checklist won't let you skip past.

    → Get my diagnosis
Let's be honest about this

This is a self-check list, not a certification. Passing every item doesn't guarantee any specific AI assistant will cite you for any specific question — nobody can honestly promise that, and anyone who does is selling something. What passing every item does is remove the obvious blockers: the things that make citation impossible outright, as opposed to merely unlikely. That's a real and useful thing to fix. It just isn't a guarantee.

Frequently asked questions

How often should I run this checklist?

For a store that publishes regularly, once a month is enough to catch drift — a plugin update that quietly changed robots.txt, a new post that shipped without structured data. Also run it whenever you migrate platforms, redesign a template, or publish a batch of posts at once, since those are the moments things silently break.

Do I need to fix everything before publishing?

No. The access items — robots.txt and llms.txt — are worth getting right before you publish anything, because nothing else on this list matters if the bots can't get in. The formatting and maintenance items can be fixed in batches afterward, and this checklist works just as well run retroactively across posts you already have live.

Can a tool check all of this on its own?

Some of it, not most of it. Bot access you can check instantly yourself with our free citability test, and an llms.txt file you can put together with the generator. But whether your formatting actually reads as citable, whether your posts link to real product pages, and whether assistants already mention you — that takes a person looking at your content and your catalog side by side. That's exactly what the free diagnosis does.

What if I fail most of the access section?

Fix that first, before touching anything else on this list. If GPTBot, ClaudeBot or PerplexityBot are blocked, the best structured data and the most citable formatting in the world go nowhere — the bots never get far enough to read them.

Does passing every item guarantee an AI assistant will cite me?

No, and treat anyone who tells you otherwise with suspicion. This checklist removes the obvious blockers — the things that make citation impossible outright. It doesn't control whether a specific assistant chooses your page over a competitor's for a specific query. Nobody can promise that.

Is this checklist specific to ecommerce?

Most of it is general GEO practice that applies to any site. The item that isn't — whether posts link to a product or category page — is the one written for stores specifically, and it's deliberately the most emphasized item here, because it's the one almost no generic GEO guide mentions at all.

The checklist tells you what's wrong. The diagnosis tells you where

A free diagnosis of your blog runs this same judgment against your actual posts and your actual catalog — including how many posts link to a product page today.

Get my diagnosis

no card · the diagnosis is free · we email you once

Explore the full cluster