How to Resell AI Visibility Audits to Your Clients

What an AI visibility audit measures and why clients ask

Clients now come to their marketing or SEO agency with a direct complaint: a buyer asked ChatGPT who to hire in their city, and the assistant named local competitors while leaving the client out. Search Console tables and ad dashboards do not answer that complaint, because they measure search result pages rather than generated answers.

An AI visibility audit, also called a GEO audit, records what AI assistants answer when a buyer asks for a provider in a trade and location, and checks whether the client's site and public listings give those assistants verifiable facts. Agencies can package this diagnostic and ongoing monitoring under their own brand through AIGROW for agencies.

A fair audit never asks an assistant whether it knows the client by name. Prompted with a company name, a language model almost always returns an agreeable summary. Instead, the test leaves the business name out and asks natural buyer questions for that category and area.

On AIGROW's GEO audit page and free visibility scan, the scanner reads the client's site, turns it into five unbranded buyer questions for that trade and area, and puts them to four AI assistants side by side: ChatGPT, Claude, Gemini and Grok. Across those twenty answers, the reading records:

• AI visibility score and recommendation rate: a score from 0 to 100 combining how often out of twenty answers the business is named across ChatGPT, Claude, Gemini and Grok, how specifically it is described, how many of the five buyer questions it wins, and whether its homepage carries machine-readable basics.
• Every answer word for word, competing businesses and sources: the full text returned by each assistant, which rivals are named first, and which directories, review pages or articles are linked.
• Site and crawler readability: whether robots.txt blocks GPTBot, ClaudeBot, PerplexityBot or Google-Extended, whether llms.txt is present, and whether the homepage states the business name, phone, address, opening hours, description and identity Schema.org markup in plain HTML.
• Ordered list of fixes: the specific observations behind missed answers, ordered by what moves the score fastest.

Packaging a one-off audit and a monthly retainer

Separate the initial diagnostic from ongoing monitoring into a one-off baseline audit followed by a monthly retainer.

The one-off baseline audit

Scope the one-off audit as a fixed onboarding deliverable:

• Approved prompt panel: unbranded buyer questions in the client's language and area, separate from brand lookups.
• Baseline readings: saved answers from ChatGPT, Claude, Gemini and Grok showing where the client is named, which rivals appear and which external sites are cited.
• Technical and crawler check: robots.txt rules for nine AI user agents checked with the AI crawler checker, indexability and snippet eligibility under Google's AI features documentation, and crawler rules under OpenAI's crawler documentation, which separates OAI-SearchBot for search results, ChatGPT-User for user questions and GPTBot for model training, and notes that robots.txt updates can take roughly 24 hours to take effect.
• Public facts review: service pages, Schema.org markup under Google's LocalBusiness structured data documentation (name and address required; telephone, openingHoursSpecification, priceRange, geo and url recommended), and up to six directory profiles compared with the NAP consistency checker.
• Ordered action plan: a prioritized backlog of site edits, schema additions, review steps and directory fixes.

The monthly monitoring retainer

Once the baseline audit is delivered and initial fixes start, move the client onto a monthly retainer. Directory updates, schema and revised pages take time to be recrawled, and assistant answers shift as models and web sources update. Using the scope on /agencies, a monthly retainer covers:

• Weekly readings: checking up to 30 buyer questions per client site every week across up to 6 markets, each in its own language, against up to 5 local competitors.
• Questions won and lost: comparing the month's latest reading against the previous month for newly won questions, newly lost questions and prompts where a rival was named without the client.
• Sites to reach: third-party domains that assistants read on answers that left the client out that month, exportable as CSV.
• Accuracy and AI visits: checking whether any assistant stated an outdated phone, address, hour, city, service or price claim, alongside AI visits from the client's GA4 export.
• Fix tracking: recording which technical or listing fixes remain open, in progress or completed.

Share our AI search visibility metrics guide with your team for metric definitions.

Explaining results to clients without overpromising

Do not present an AI visibility check as a static rank table. Generative assistants sample responses and refresh web lookups, so answers change from one run to the next even when nothing on the client's site has changed.

Treat a single run as one sample rather than a fixed position. If an assistant names the client in one weekly run and omits them the next week on the same prompt, neither run stands alone. Reading results across repeated weekly runs over a month shows the monthly average and the low-to-high range.

Separate normal variation from a real trend. In AIGROW's monthly client reports, small moves within a five-point volatility band count as steady because assistant answers move a few points on their own between readings. Focusing client reviews on sustained shifts across repeated runs prevents wasted meetings over sampling noise.

Compare only runs measured under the same conditions. If a client changes tracked questions or switches between quick probes and full web-search answers, the monthly report leaves incompatible readings out of the comparison. Tie your agreement to inspectable deliverables: the baseline audit, updated pages and schema, corrected listings and weekly readings.

Running the service under your agency brand

AIGROW for agencies provides the monitoring seats, branded reporting, website widget and wholesale fulfilment on a self-serve basis.

Client seats, branded PDFs and private links

On /agencies, choose your client seats and pay by card with no contract and no sales call, with a per-seat volume rate applied automatically at higher seat counts. Each seat monitors one client site: up to 30 questions checked weekly, up to 6 markets in their own languages and up to 5 competitors. You can change your seat count or cancel in any month; reducing seats pauses the newest client sites above the new number rather than deleting them.

In report settings, add your agency name, logo and colour once. Every monthly PDF and private client link then carries your agency brand without AIGROW's name, header or menu, and AIGROW never contacts your clients. Send clients the printable PDF or a private link on a plain noindex, nofollow page.

The embeddable audit widget

Paste the embed snippet from /agencies or the agency kit into any HTML block on your site. The widget runs without an iframe. Anyone can embed the free widget with AIGROW branding; an agency with client seats can create a one-year signed embed key so the widget and its PDF show the agency's name alongside a small Powered by AIGROW line.

With optional lead capture enabled on the agency kit, when a visitor requests an email copy of a scan, AIGROW emails the reading to the visitor and delivers their email, site and reading to your verified agency account.

Wholesale done-for-you services

When an audit uncovers fixes your team prefers not to fulfil in-house, /agencies offers four one-off wholesale services delivered under your agency's name and logo:

• AI source listings (retail scope): up to 20 listings on sites AI assistants read for the client's trade and country, with an approved site list, logins handed over, and readings before and at day 30.
• Google Business Profile setup (retail scope): completing and correcting the client's Google profile, with matching details on Bing Places, Apple Business Connect and Facebook.
• Structured data and answer pages (retail scope): Schema.org markup on up to 20 pages of the client's site and 5 pages answering customer questions.
• Google reviews kit (retail scope): a review link, QR code, printable card, request messages, replies to the last 20 reviews and a day 30 check.

At checkout enter your client's business name and website. Within 2 to 3 working days of your intake reply, you receive a proposal to approve before anything changes.

Deliver AI visibility audits and monthly reports under your own brand See agency seats and wholesale services

Mistakes agencies should avoid

Avoid these errors when pitching and delivering AI visibility work:

• Promising a position in ChatGPT: nobody can buy or force an AI answer, and no agency has a special arrangement with OpenAI, Anthropic, Google or xAI. Sell the diagnostic, factual fixes and ongoing measurement.
• Selling a single screenshot as proof: a prompt typed into a personal browser reflects account history, location and single-run chance. Use repeatable runs with saved questions and dates, as covered in our guide to evaluating AI visibility tools.
• Stuffing the prompt set with branded queries: testing questions that already contain the client's business name inflates the score and hides whether unbranded buyers find the client.
• Ignoring the client's own site and listings: pitching third-party citations while the client's site blocks crawlers, lacks clear service descriptions or shows conflicting contact details wastes budget. See our GEO vs SEO guide for how both fit into one backlog.
• Violating directory or review rules: never invent storefront addresses or add marketing taglines, store codes or phone numbers to a business name against Google's business representation guidelines, and never discourage negative reviews, selectively solicit positive ones or offer incentives against Google's maps user-contributed content policy.

Agency launch checklist

Follow this checklist to launch your AI visibility offer:

1. Set up agency branding: choose client seats on AIGROW for agencies and add your agency name, logo and colour in report settings.
2. Embed the prospect widget: generate your branded embed key on /agencies or the agency kit and paste the checker onto your site.
3. Run baseline audits: use the free GEO audit or your first client seats to record unbranded buyer questions, assistant answers, named rivals and technical gaps.
4. Freeze questions and competitors: set up to 30 unbranded buyer questions per client across up to 6 markets and languages, and add up to 5 local competitors.
5. Execute initial fixes: resolve crawler blocks, unclear service pages, missing schema and inconsistent listings in-house or through wholesale services on /agencies.
6. Send monthly branded reports: review the monthly score range, questions won and lost, outreach domains, accuracy checks and GA4 AI visits, then send the branded PDF or private link.