Free checklist
GEO readiness checklist
Sixteen checks to do before buying any GEO tool. None needs a paid tool, and each says how to know it's done.
1. Crawl access
AI crawlers are allowed in robots.txt
Open yoursite.com/robots.txt and look for rules naming GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot or Google-Extended. A Disallow: / under any of them keeps that engine out.
Done when: Each bot you want is either not mentioned or has Allow: /.
Your CDN or firewall isn't blocking bots
Bot protection at the CDN can block AI crawlers even when robots.txt allows them. Check your CDN's bot settings for AI crawler rules.
Done when: Fetching a page with curl -A "GPTBot" returns the page, not a challenge.
Key content is in the HTML
Many crawlers don't run JavaScript. If your pricing, features or answers only appear after scripts run, they may not be read.
Done when: View source (not the inspector) on your key pages and find the text there.
Your sitemap is current and listed
A sitemap.xml that lists every page you want found, linked from robots.txt with a Sitemap: line.
Done when: Every important page is in it, and none of its URLs return errors or redirects.
2. llms.txt
Publish /llms.txt
A plain-text Markdown file at the root of your site, proposed at llmstxt.org: your name, a one-line summary, then links to your most useful pages with a sentence on each.
Done when: yoursite.com/llms.txt loads as text and every link in it works.
Point it at the pages that answer questions
Pricing, docs, comparisons and FAQs are what answer engines quote. Lead with those, not the blog archive.
Done when: Someone reading only llms.txt could say what you sell, for whom, and what it costs.
3. Structured data
Organization and WebSite markup on the home page
schema.org Organization (name, url, logo, sameAs links to your profiles) tells engines which entity your site is.
Done when: Google's Rich Results Test or the Schema.org validator reads it without errors.
The right type on each kind of page
Product or SoftwareApplication on product pages, Article on posts, FAQPage where you answer questions, ItemList on ranked lists.
Done when: A sample of each template validates.
Markup matches what's visible
Prices, names and dates in the markup must match the page. Mismatches can make engines distrust both.
Done when: Spot-check three pages by eye.
4. Pages worth citing
Answer first
Put the direct answer in the first paragraph under a heading that matches the question, then the detail.
Done when: Your top five pages answer their main question in the first two sentences.
Specific, checkable facts
Numbers, plan names, supported platforms, dates. Engines lift facts, not adjectives.
Done when: Each key page has facts a reader could verify.
Dates and freshness
Show when a page was last updated, and update pages whose facts have changed.
Done when: No key page states a price, feature or year that is no longer true.
5. Citation tracking
Write your prompt list
Ten to thirty questions your buyers would ask an AI: "best [category] for [use]", "[you] vs [competitor]", "how do I [job you solve]".
Done when: The list is written down, and each prompt is one a real buyer would type.
Take a manual baseline
Ask each prompt in ChatGPT, Perplexity, Gemini and Google (for AI Overviews), and note whether you are mentioned, cited, or absent, and who is cited instead.
Done when: A spreadsheet with one row per prompt per engine, dated.
Watch AI traffic in your analytics
Visits referred from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com show answers sending people to you.
Done when: You can see last month's AI-referred visits by source.
Decide what a tool must add
Once the manual baseline is too slow to repeat, you know what you need: more engines, more prompts, history, API access, or help making changes. Take that list to the ranking.
Done when: You can name the two criteria that matter most to you.
Get the next version of this list
We update the checklist as engines change what they read. Leave your email to hear when.