AI Search Readiness Audit

What's inside

When a buyer asks ChatGPT, Claude, or Perplexity for a business like yours, 3 things decide whether you appear: whether AI crawlers can read your site, whether the engines know who you are, and whether you get named in your own category. This audit measures all 3 against your live domain and ranks what to fix first.

What you get

A report scored 0-100 across 3 weighted legs, emailed to you:

• Technical readiness — 50% of the score. 8 scored checks, 100 points: AI crawler access in robots.txt (30 pts, tested per bot for GPTBot, ClaudeBot, PerplexityBot and Google-Extended), sitemap.xml validity and URL count (15 pts), JSON-LD schema types (15 pts), llms.txt (10 pts), title tag (10 pts), meta description (10 pts), H1 structure (5 pts), homepage weight (5 pts). Every check prints its points and the raw value behind it.
• Entity recognition — 25%. We query live AI models about your business by name and domain, then score how accurate and complete the answers are. The actual answers are quoted in the report.
• Citation presence — 25%. We ask the category questions your buyers ask, and record whether you were named, ranked, or absent.
• Prioritized fixes. Each gap becomes a specific fix ordered by impact — not "improve your SEO" but "ClaudeBot is blocked at line 4 of your robots.txt, so 3 of the 4 major assistants cannot read your site at all."

Two real results, not mockups

A live run on 2026-08-03 scored a restaurant site 70.0/100 on technical readiness, 16.7/100 on entity recognition, and 40.0/100 on citations — 49.2/100 overall, "PARTIALLY READY". All 4 AI crawlers were allowed and robots.txt was clean, yet the site still landed under 50, because the engines could not reliably say what the business was or name it in its category. The half you can check yourself was fine; the other half was the problem.

A second run the same day scored anthropic.com 75.0/100 on technical readiness: 30/30 on crawler access, 15/15 on a sitemap carrying 508 URLs, 10/10 on a 126-character meta description, 5/5 on a 176 KB homepage — but 0/15 on JSON-LD schema and 0/10 on llms.txt. Even a well-built site leaves 25 points of machine-readability on the table.

How it works

Enter your domain at checkout. That 1 field is the only input we need, and we never guess it — if it is missing or unclear, we email you before running anything.

Turnaround

Within 72 hours — a deadline, not a hope. Most audits finish in under 5 minutes; the 72-hour window exists so a provider outage or a site that is briefly down never becomes a missed promise.

What this is not

Not a consulting call, not an implementation service, not a ranking promise. We do not edit your site and we do not submit you anywhere. This is a measurement of where you stand on 3 axes and what to fix first.

Honest degradation: if a leg cannot be assessed — your site is unreachable, a model provider is down, or your robots.txt blocks our fetch too — the report marks that leg "not assessable" and reweights the remaining legs. In the sample above, an earlier run of anthropic.com scored 75.0/100 on 1 assessed leg and said so on its face rather than inventing the other 2. We never fill gaps with filler.

Refund

If we cannot complete your audit, you get a refund instead of a padded report. Reply to the delivery email — it reaches a real person, usually the same day.

SteelWorks Intelligence LLC, Acworth GA — https://steelworksintelligence.com