CED Systems · AI-Readiness Audit
How we score: the audit methodology, in full
Your AI-readiness score (0–100) measures one thing: how easily AI assistants can find your business, understand what you do, and recommend you to a customer. Not how good your work is. Not how pretty your website is. Whether the machines doing the recommending can read you at all.
This page is the whole rubric — the same five categories, the same checks, and the same point values we apply to every site, including our own. No secret sauce, and no private "better" score.
Why it matters now
Customers are skipping the scroll. Instead of reading ten search results, they ask ChatGPT, Grok, Muse, or Google's AI: "Who's the best contractor near me?" Each AI names only a short list — typically four to six businesses. We ran that test across four AI models plus Google's own results: forty different plumbers were named across everything we checked, and not a single one appeared everywhere. The shortlist is small, and it's different every time.
Google goes further: its "Have AI check pricing" feature has Google's AI call local businesses on a customer's behalf, ask about pricing and availability, and hand the customer a comparison. If the call hits a dead line or voicemail, the AI reports back empty for you — while competitors who answered show real answers.
If your business isn't machine-readable, you're not on any version of that list. You never even know you lost the job.
The five categories
1. Crawler treatment — 20 pts
Are the AI bots welcomed in, or blocked by accident?
- Your robots.txt file: does it exist, and does it let named AI crawlers (like GPTBot, ClaudeBot, and Google's AI crawlers) in — or shut them out by default?
- Your sitemap: is there one, is it reachable, and does it tell the bots where your pages are?
2. Machine-readable facts — 25 pts
Can software read your business facts without guessing?
- Structured data (schema / JSON-LD) on your site: business name, address, phone, hours, and services in the format machines parse directly.
- Facts stated once, consistently — not buried in images, PDFs, or scattered across pages that disagree with each other.
3. Agent ID files — 20 pts
Do you have the files that introduce your business to AI assistants in their own language?
- /llms.txt — a plain-language file that tells AI assistants who you are and what you offer.
- /agent.json — a machine-readable ID card: name, contact paths, and what an agent can do with you.
- We verify the files actually contain what they claim — an error page served under the right name doesn't count.
4. Contact actions — 20 pts
Can an assistant hand a ready customer straight to you?
- Tap-to-call phone links a phone (or an AI caller) can actually dial.
- A working email link and a working contact or quote form.
- No dead ends: the contact paths on the page have to work, not just be drawn on it.
5. Sharing readiness — 15 pts
Does your link look real when an AI cites it or a customer shares it?
- Page titles and descriptions that say what the business is and where it is.
- Preview tags (Open Graph) so shared and cited links render with a proper title, description, and image.
- Headings that structure the page so a machine can outline what you do.
How to read your score
90–100: excellent — AI assistants can find, read, and recommend you. 70–89: good, with a few specific gaps. 40–69: needs work — you're partially visible and losing the rest. Below 40: effectively invisible to AI assistants.
Every point missing is a specific, fixable gap. Most fixes are adding files and markup your site should have had all along — which is why the audit names each gap, what it costs you, and what the fix is.
How scoring is done — and our promises
- Same checks for everyone. The rubric on this page is the audit. Our own site, cedservice.com, scored 41/100 under it before we fixed it (98/100 on re-audit).
- Free rechecks, any time. Fix things yourself and ask — we'll re-run your score and update it, free.
- Same score everywhere. The score we show you is the score we publish. If your site earns a high score, your report says so.
- Re-audits over time. Scores are re-run so movement is tracked — the before/after is the proof, not our word.
Full report, every gap in plain English. Details on the audit page.