AI visibility readiness checklist for service businesses

Use this worksheet to review how customers find and evaluate your service in search and AI answers. It is for a business owner and the people responsible for their website, content and enquiries. It does not assess your team’s use of AI software.

The four layers are Discoverable, Understandable, Citable and Routable. They organize the review; they are not an engine’s ranking formula. There is no total score, pass mark or automatic recommendation to buy work.

Choose the question before checking the page

Pick one service you want to sell, its intended buyer and a relevant buying question. Review the service page and one supporting page that should help that buyer. Start a separate worksheet for a different service or market.

If the symptom is an incorrect answer, a missing recommendation or unsuitable enquiries, use the service-business diagnosis to choose the first investigation. A single omission does not establish a fault in your website.

How to record the evidence

For each check, record the URL, system or source, date and observation in your notes. Use the check ID to connect the note with the worksheet. Mark one state:

• Confirmed (C): Recorded evidence supports the check for the stated URL, system and date.
• Gap (G): Recorded evidence shows a specific failure.
• Unknown (?): Not checked, inaccessible or insufficient evidence.

Keep an unchecked item as Unknown. A gap needs an observed failure, not an assumption. Technical checks may need your current web or SEO supplier’s help.

The PDF has room to record one priority issue and its next action. Keep supporting evidence in separate notes, keyed to the check IDs. The twelve questions below match the PDF; the explanations below add detail for completing them.

1. Discoverable

D1. Do recorded checks show the service page is accessible to intended search crawlers?

Ask the web team for the URL response, relevant robots rules and verified retrieval or host/CDN evidence, naming the crawler and date. Robots permission alone does not confirm retrieval; a browser visit does not establish crawler access. If evidence is missing, choose Unknown. The ChatGPT access guide separates search crawling, training preferences and user-triggered visits.

D2. Do inspection records show the intended indexing and canonical status?

For Google, inspect the service URL in Search Console and record the reported status, selected canonical and crawl date. Distinguish stored inspection data from a current access test. Indexing does not guarantee selection in an AI answer. Use the indexing investigation to interpret a missing-page result. For Google AI features, also record snippet eligibility and the effective Search generative AI inclusion setting; the Google AI guide explains these separate checks.

D3. Can a buyer reach the service page through a relevant link on the site?

Record an actual route from navigation or a relevant supporting page. The service page does not need to link to itself. If the route is missing, identify the page where the link would help a buyer.

2. Understandable

U1. Does the opening explain the service, intended buyer, market and next step?

Quote the relevant wording. A market may be a type of customer or a delivery area; do not invent local coverage. Note which part a prospective customer would still have to guess.

U2. Does each reviewed page have a distinct job in the buying decision?

Write one sentence describing each page’s job. A cost guide and a service page can serve different tasks. Similar wording alone is not a reason to merge pages or redirect them.

U3. Do the business facts agree with the authoritative sources you checked?

Compare service names, business identity, coverage and relevant qualifications with the records your team maintains. List exactly which sources were checked. If an external profile is wrong, name its owner; do not claim a complete web-wide audit.

3. Citable

C1. Can a buyer find clear answers about fit, scope and process?

Record the sections that answer who the work suits, what is included and what happens next. A reader should not have to infer the service from a list of broad benefits.

C2. Are cost factors and delivery dependencies explained?

Record the explanation or its contextual link. State what affects scope, cost and timing where a fixed figure would mislead. This check does not require inventing a price or a completion promise.

C3. Does each reviewed result claim identify its source, period and limits?

Identify the claim and its evidence. Separate an observation from a causal explanation and a client’s outcome from a forecast. If there are no result claims to review, record that scope limit as Unknown; do not create a claim just to complete a check.

4. Routable

R1. Does each reviewed page offer an appropriate next step for its reader?

Follow the actual link. A guide may lead to a service explanation; a service page may lead directly to an enquiry, with proof already alongside it. There is no compulsory sequence of pages.

R2. Does a mobile enquiry test confirm both success and delivery?

Coordinate one clearly marked test with the recipient. Record the confirmation and the received message. A success screen alone does not confirm delivery. Do not submit repeated unsolicited tests.

R3. Can the team record enquiry fit and source, including unknown attribution?

Agree what a qualified enquiry means for this service: need, customer fit and delivery market. Check that the team can record it consistently. Use available referral or landing-page evidence; a change in timing alone does not establish attribution.

Record an AI answer without turning it into a ranking

If your starting point is an AI answer, retain the exact question and full visible answer, cited URLs, platform, visible model or mode, date, service and market. Note whether the business was named, whether the conversation was fresh, and relevant location or memory settings. Mark unavailable context Unknown. Keep repetitions and unsuccessful attempts in the record as well.

Separate the observation from the cause you suspect. “Our business was not named in this answer” is an observation. It does not tell you why, or whether another buyer would receive the same answer. The diagnostic guide shows how to choose the next check.

Choose one action the evidence supports

Record the check ID, what you observed, the evidence source and date, the person who owns the next action and when to review it. Correct a verified failure, investigate an unknown, or leave a supported check alone. Do not turn twelve checks into twelve projects.

A clear wrong business fact or access failure can go to its current owner. A page-fit problem may need a content revision. If evidence is missing, request the check before commissioning a rebuild or more pages.

Bring your priority finding to a free Clarity Call. For an initial written reading of the site and the first useful investigation, get your free Outside View. If the issue is already clear, the AI SEO plan explains implementation. The worksheet does not predict rankings, citations or leads.

Guidance behind the technical checks

• Google: AI optimization guidance. Check indexing, snippet eligibility and the effective Search generative AI inclusion setting; no special AI markup is required.
• OpenAI: crawler controls. OAI-SearchBot, GPTBot and ChatGPT-User serve different purposes. Browser access is not proof of crawler retrieval.
• OpenAI: ChatGPT Search. Location and relevant saved memory can affect search context.
• Google: Generative AI performance report. Check the report if available for your property. An unavailable report does not establish zero visibility.

Explanatory guidance reviewed 5 October 2026. The twelve worksheet questions and PDF are unchanged. Download the matching one-page worksheet.