Agency client audits
Who it is for
An agency or freelancer who audits several client sites for AI search readiness and wants one repeatable routine.
The job
Find what stops each client’s site from being read and quoted by answer engines. Fix the biggest causes first. Then show the client whether AI names them.
Tools and skills
analyze_website_aeoon the aeo-research server. No sign-in. It crawls 10 pages and returns at most the top 8 fixes. It is a shortlist, not the full audit.- The hosted audit at advancelabs.dev/tools. Free, no account. It runs the 44-check audit and returns a weighted, per-rule report.
- The
aeoskill. It reads an audit, groups the failures, and hands each group to the skill that owns it. - The AEO Toolkit dashboard, for tracking client questions over time. The pricing page lists what each plan includes.
Steps
- Connect aeo-research:
claude mcp add --transport http --scope user aeo-research https://aeo.advancelabs.dev/api/mcp/research/mcp. - For each client, ask your agent to run
analyze_website_aeoon the client’s home page. Note the score, the AI-bot access result and the top fixes. - Run the hosted audit on the same URL when you need the full 44-check report. Keep the per-rule result as your baseline.
- Install the plugin so the agent has the skills:
/plugin install aeo-toolkit --marketplace Advance-Labs/aeo-toolkit. Ask it to “do AEO on” the client site. Theaeoskill orders the work and routes each failing rule to its owning skill. - Ship the fixes. Re-run
analyze_website_aeoand the hosted audit after each batch. - Add the client’s buyer questions to a project in the dashboard. It asks each question of Perplexity, GPT with live web search and Gemini on a schedule, and records whether the client is named.
- On the Agency plan, the pricing page lists white-label client reports, shareable as a link or a PDF, and prospect workspaces for one-off snapshots when you pitch a new client. Check the page for current limits before you promise a client anything.
What you get
- A score out of 100 with a grade, and a prioritised fix list per site.
- A list of failing rule IDs, each with one owning skill. Rule definitions are in the audit rules.
- Over time, whether the client was named for each tracked question, and which competitors were named beside it.
A finished report ranks the fixes and shows each page and rule. This one is the team’s own site, before and after.

Limits
- The 8-fix list from
analyze_website_aeois a first look. Use the full audit before you quote a score to a client. - An audit score lists known causes of being skipped or misread. It does not measure whether anyone cites the site. Tracking does that.
- Tracking answers are API answers, not screenshots of the consumer apps.
AEO Toolkit is an open-source instrument by Advance Labs.