ChatShortlist
Score my site

Free · 25 checks

Your website already has an AI score. You just haven't seen it.

ChatGPT, Claude, Gemini and Perplexity read your site before they recommend anyone. We fetch it exactly the way they do and grade what comes back — out of 100, with the evidence for every point.

No account, no card — see your score first. A free account opens the findings behind it and keeps your reports from expiring.

  • 25 checks
  • 5 categories
  • Results in seconds

Want it watched continuously instead? Join the monitoring waitlist →

Why a score

Search collapsed into a single answer

A results page gave ten businesses a chance. An assistant names a handful and moves on. Whether you are one of them is decided by what a crawler could read — usually months before anyone asks the question.

  • 1

    answer, not ten links

    An assistant returns a single response. There is no page two to be found on.

  • 3–5

    sources named in it

    Everyone outside that shortlist is invisible, however good their website is.

  • 0

    lines of JavaScript run

    Most AI crawlers read your raw HTML and nothing else. Whatever renders afterwards does not exist to them.

The rubric

Twenty-five checks, five categories, 100 points

Weighted by what actually keeps a business out of an AI answer. Every check reports what we found, so you can verify the number rather than trust it.

  • AI crawler access

    25 pts

    Whether GPTBot, ClaudeBot, PerplexityBot, Google-Extended and the rest can fetch your pages at all. Most sites that score badly here have no idea they are blocking anything.

    • robots.txt resolved per crawler, the way each one resolves it
    • Bot protection quietly refusing non-browser traffic
    • noindex, nosnippet and noai directives in the page and the headers

  • Structured data

    25 pts

    The machine-readable facts a model reads instead of inferring from your prose. Missing or contradictory facts make an assistant hedge — and a hedging assistant names someone else.

    • schema.org JSON-LD present, valid, and attached to this URL
    • An Organization or LocalBusiness entity with a real name behind it
    • FAQPage, Service, Product or Article markup it can quote

  • Answer-ready content

    20 pts

    Language models lift passages, not pages. This measures whether yours survive being pulled out of context and dropped into an answer.

    • Enough substantive text in the HTML to be worth citing
    • Headings phrased the way customers actually ask
    • Self-contained paragraphs rather than walls of copy

  • Identity & citability

    15 pts

    The metadata that lets an assistant say who you are with confidence, and link back to the right page when it does.

    • Title, meta description, canonical URL, declared language
    • Open Graph fields and an XML sitemap
    • An llms.txt summary written for machines

  • Delivery & rendering

    15 pts

    Whether your content survives the trip to a crawler that does not run JavaScript and will not wait around.

    • Content present in the raw HTML, not assembled client-side
    • Time to first byte, HTML weight, redirect chains
    • HTTPS end to end

How it works

We read your site as a machine, not a browser

  1. Paste a URL

    No account, no card, no crawl queue. Type an address and the scan starts, and your score comes back either way. An account is what opens the findings behind it, keeps reports from expiring, and raises the limit past the few scans an anonymous visitor gets.

  2. We fetch it the way a bot does

    Your page, robots.txt, sitemap and llms.txt, requested with a declared crawler user agent and no JavaScript execution — exactly like GPTBot and ClaudeBot. What we see is what they see.

  3. Read the receipts

    Twenty-five checks across five categories. The score and where you are losing points are free; a free account opens what we actually found on each check and what to do about it — no verdict you have to take on faith.

Coming soon

One scan is a snapshot. Sites drift.

A framework upgrade moves rendering client-side. Someone tightens robots.txt to stop scrapers and blocks four assistants by accident. Monitoring watches for it and tells you the day it happens.

  • Every page, not one URL

    We crawl the whole site and score each page, so a template-level mistake shows up once instead of eighty times.

  • Alerts when a deploy breaks you

    A robots.txt edit, a framework upgrade that moves rendering client-side, JSON-LD that stops validating. You hear about it that day, not next quarter.

  • Score history

    Watch the number move as you fix things, with a record of what changed and when it changed.

  • Competitor comparison

    Your score next to the businesses assistants currently name in your category.

Join the waitlist

We'll email you when monitoring opens. Scanning a URL stays free either way.

One email when monitoring opens, then occasional notes on how AI search is changing. No spam, unsubscribe in one click. See our privacy policy.

FAQ

Questions we get asked

What is an AI search readiness score?

An AI search readiness score measures how well a website can be crawled, parsed and quoted by AI assistants such as ChatGPT, Claude, Gemini, Perplexity and Copilot. ChatShortlist grades a page out of 100 across five categories: AI crawler access, structured data, answer-ready content, identity and citability, and delivery and rendering. It measures technical readiness, not whether an assistant will choose to recommend you.

How is this different from a normal SEO audit?

A traditional SEO audit optimises for a ranked list of ten links. This scan optimises for a single generated answer that names three to five businesses and cites a handful of sources. Crawlability and speed overlap. What differs is what matters most here: per-crawler robots.txt rules for AI agents, structured data an assistant can turn into facts, whether your content exists without JavaScript, and whether individual passages can be lifted and attributed.

Which AI crawlers do you check?

GPTBot, OAI-SearchBot and ChatGPT-User for OpenAI; ClaudeBot and Claude-SearchBot for Anthropic; PerplexityBot for Perplexity; Google-Extended for Gemini; Bingbot for Copilot; and Applebot-Extended for Apple. Each is resolved against your robots.txt separately, because several of them only honour directives addressed to them by name — a wildcard rule that looks permissive can still be blocking three of them.

Does the scan change anything on my site?

No. The scan is read-only. It makes up to four ordinary GET requests — your page, robots.txt, llms.txt and sitemap.xml — using a declared user agent, ChatShortlistBot. It does not log in, submit forms, crawl beyond the URL you gave us, or execute JavaScript.

How accurate is the score?

Every check shows the raw observation next to the points it earned, so you can verify each one yourself rather than taking the number on faith — that is what a free account opens. The limits are worth stating too: we score the single URL you submit rather than your whole site, we see your server as it responded at that moment, and we measure technical readiness rather than the reputation and third-party citations that also decide whether an assistant recommends you.

Can you guarantee that AI will recommend my business?

No, and you should be wary of anyone who does. Assistant output is probabilistic and shifts with every model update. What a score tells you is whether you are even in the running: whether an assistant can reach your site, understand what you do, and quote you accurately. That is a necessary condition, not a sufficient one.

What does the monitoring waitlist get me?

Continuous monitoring: your score re-run on a schedule across every page rather than one URL, alerts when a deploy breaks something — a robots.txt change, structured data that stops validating, a page that starts requiring JavaScript — score history, and comparison against the competitors an assistant currently names in your category. Scanning a single URL stays free.

Is the scan free, and do I need an account?

Scanning is free and needs no account: paste a URL and you get your score out of 100, your grade, and your points in each of the five categories. A free account opens the rest — what we actually found on each of the twenty-five checks and how to fix it, ranked by points. It also keeps reports indefinitely rather than seven days, raises the hourly scan limit, and adds history and share links. Anything scanned before signing up opens automatically once you do; you do not have to run it again. There is no card involved either way.

Free · 25 checks

See your score

Paste a URL and create a free account. We'll tell you what an assistant can read, what it can't, and which fix is worth the most points.