A good SEO and AEO report separates four things: visibility (impressions, rankings and appearances in AI answers), traffic (clicks and visits from search and AI tools), outcomes (enquiries, sales or donations) and the work done. Every metric should name its source and what it can't prove. AI visibility should come from a fixed set of real buyer questions, run on named engines at least three times each per period, with the raw answers and cited URLs kept. Reports should end with decisions, not just charts.
Why do most SEO reports fail?
Most reports measure what is easy to collect, not what helps a decision. Common problems: a third-party authority score presented as progress; rankings for terms nobody searches; traffic totals with no link to enquiries; AI "visibility scores" from an undisclosed method; and pages of charts with no recommendation. A report should let a non-specialist answer three questions: are more of the right people finding us, are they doing what we want, and what are we changing next?
Which metrics belong in an SEO and AEO report?
| Metric | Source | What it shows | What it can't prove |
|---|---|---|---|
| Impressions | Search Console; Bing Webmaster Tools | How often your pages appeared in results | That anyone saw or clicked them |
| Clicks and CTR | Search Console; Bing Webmaster Tools | Visits earned from search, and how compelling listings are | The quality of those visits |
| Average position | Search Console | Typical placement across queries and locations | A single "rank" everyone sees |
| Generative AI appearances | Search Console's Generative AI performance report, where available | Appearances in Google's generative features | Visibility in other AI tools |
| AI mention rate | Your dated prompt log | Share of tracked answers that name you | Behaviour of every user or prompt |
| AI citation rate | Your dated prompt log | Share of tracked answers that link to your site | Clicks from those answers |
| AI share of voice | Your dated prompt log | Your presence vs named competitors on the same prompts | Market share |
| AI referral visits | GA4 referrals (chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com) | Visits that came from AI tools | Visits from Google's AI features, which count as organic |
| Organic sessions and engagement | GA4 | Visits from unpaid search and what people did | Why they came |
| Enquiries, sales, donations | Your CRM, form records or checkout | Business outcomes from search | Exactly which touchpoint caused them |
| Indexing health | Search Console page indexing | Whether important pages can appear at all | Whether they will rank |
| Core Web Vitals | Search Console; PageSpeed Insights | Real-user speed and stability | Rankings on their own |
Bing Webmaster Tools increasingly reports how Microsoft's AI experiences use your pages, alongside its search data; include it where it is available to your site, because Bing's index also feeds Copilot.
How should AI visibility be measured?
AI answers vary from run to run, by location and by whether the user is signed in. A trustworthy method fixes as much of that as it can and states the rest:
- Agree the prompt set. 20–50 real questions buyers ask, including unbranded ("best charity website agency UK"), comparison and problem questions. Keep it stable so periods are comparable; change it only at quarterly reviews.
- Name the engines. For example Google AI Mode or AI Overviews, ChatGPT search, Perplexity and Microsoft Copilot. Record the model or mode where shown.
- Fix the conditions. Country, language, signed-out where possible, a clean session, and the date and time.
- Repeat each prompt. At least three runs per engine per period, so one unusual answer doesn't swing the result.
- Record the evidence. Whether you were mentioned, cited or both; the cited URLs; competitors named; and any inaccurate statements about you. Keep screenshots or exports.
- Report rates with the sample. "Cited in 14 of 45 answers (31%) across 15 prompts × 3 runs on Perplexity" is useful. "AI visibility score: 72" is not.
Our public prompt panel shows this method applied to our own site, including answers that did not cite us and the stale figures we corrected as a result.
A sample monthly SEO and AEO report
| Section | What it contains |
|---|---|
| 1. Summary | Three to five sentences: what moved, why we think it moved, what we will do next. Written for someone who reads nothing else. |
| 2. Outcomes | Enquiries, sales or donations from organic and AI sources, against the same period last year, with the definition of a qualified enquiry. |
| 3. Search visibility | Search Console and Bing impressions, clicks and CTR for priority pages and query groups; notable gains and losses. |
| 4. AI visibility | Mention, citation and share-of-voice rates by engine, the prompt set and run counts, cited URLs, competitors, inaccuracies found. |
| 5. Technical health | Indexing issues, Core Web Vitals, crawl errors and anything blocked; what was fixed. |
| 6. Work completed | Pages published or improved, fixes shipped, links or mentions earned, with links to each. |
| 7. Next month | The prioritised plan and what it depends on (approvals, developer time, content from you). |
| Appendix | Raw prompt log, query exports and definitions, so anything in the report can be checked. |
How do you connect visibility to enquiries?
Start with records you control: every enquiry form, call or checkout should capture the landing page and source where consent allows, and someone should mark which enquiries were genuine and relevant. Compare outcomes with the same period last year to allow for seasonality. Treat AI visibility as an influence you observe, not a revenue line you can calculate: if AI answers increasingly cite your pricing page and enquiries mentioning "I saw you recommended" rise, that is evidence worth reporting, with its limits stated. With low enquiry volumes, say so rather than presenting a percentage change from a handful of leads.
How often should you report?
Monthly suits most businesses: long enough for search data to settle, short enough to act. Review strategy, the prompt set and priority pages every quarter. During a site launch, migration or campaign, add a weekly check of indexing, errors and key queries. Avoid daily ranking alerts; they encourage reacting to noise.
How do you judge an agency's SEO and AEO reporting?
If you're comparing agencies, the reporting tells you a lot about the work. Ask each one:
- Can I see a real (anonymised) monthly report before signing?
- Which prompts, engines and run counts do you use for AI visibility, and can I see the raw answers?
- What do you count as a citation versus a mention?
- How do you define and track a qualified enquiry?
- Which metrics do you not report, and why?
- What happened on an account where results went down, and how did you report it?
Red flags: guaranteed rankings or AI citations; a proprietary "visibility score" with no method; authority scores presented as results; no access to your own Search Console or analytics; reports with no recommendations. For the wider buying decision, see how to choose an agency and our honest comparison of UK GEO and AEO agencies.
Which tools do you need?
The core is free: Google Search Console, Bing Webmaster Tools, Google Analytics 4 and a dashboard such as Looker Studio. A spreadsheet is enough for a prompt log at small scale; dedicated AI-visibility trackers help when the prompt set grows. Our AI visibility checker helps prepare and organise manual checks; it earns us no commission. For more, see the best SEO tools.
How we report: our AEO and SEO retainers include monthly reporting built on this structure, with access to the underlying data.