Every vendor selling AI visibility leads with a dashboard. Dashboards are how this industry offloads work onto the buyer: you get 4,000 rows of green and red, and the job of deciding what any of it means becomes yours. The buyer of a GEO/AEO program — a founder, a CMO, an agency owner with a client to answer to — reads one page. If the report is longer than one page, it is not a report, it is an assignment. This post is the template we use, with a filled-in sample you can copy.
Why one page
Three constraints, all about the reader:
- The owner reads one page. The person who paid for the program is not opening a tool. They are reading a page in an email, before or after a meeting, and forming one judgment: is this working, and is the money well spent? One page is the format that survives that moment.
- A page forces the call. The moment you have unlimited dashboard space, you dump the finding and let the client interpret it. A page with four fixed sections forces you to convert findings into decisions: what changed, what it means, what was done, what happens next.
- One page is diffable. Next month’s page sits next to last month’s page. That is the entire accountability model: a number that moves, or a number that gets explained.
The one-page report is also what the free quick check returns — same shape, one month’s worth of depth added.
The four sections
1. Score + month-over-month delta. One headline number: citation share — across the mapped money queries and the four AI surfaces (ChatGPT, Perplexity, Gemini, Claude), how often the brand is cited, mentioned, or absent, versus the competitors the answers name instead. Then the delta, with the reason. If the number moved, the reason. If it did not move, the reason for that. The raw captures behind every score are retained and available on request — the score is reproducible, not vibes.
2. What changed. Every query where the answer changed: the query, the surface, the old and new position, and a short verbatim excerpt of the new answer. Excerpts, not screenshots — excerpts are copyable, searchable, and can’t be cropped to flatter.
3. Top 3 fixes, ranked. The three highest-leverage fixes for the month, ranked by revenue impact, not effort. Each one: the fix, the query(s) it targets, and how it will be verified against the next run. Three, not thirteen — a list of fifteen is a list of zero, because the reader cannot see the priority.
4. What we did. What was actually implemented last month, what shipped, and what did not and why. This is the section that separates a service from a tool — a tool has no “what we did” (the full tools-vs-service cost breakdown is here).
The filled-in sample
The table below is the shape of section 1’s scorecard. This is SAMPLE data — illustrative rows, not live results for any real brand — using the same shape as the sample scorecard on our homepage. The queries are anonymized in category form.
AI visibility — SAMPLE scorecard (illustrative data, not live results)
| Money query | ChatGPT | Perplexity | Gemini | Claude | Δ vs last month |
|---|---|---|---|---|---|
| “best [category] for [use case]” | Cited #2 | Absent | Cited #4 | Mentioned, no link | ▲ improved |
| “[your product] vs [competitor]” | Absent | Cited #1 | Cited #2 | Cited #3 | — flat |
| “[competitor] alternatives” | Cited #3 | Cited #2 | Absent | Absent | ▼ declined — fix queued |
| “[product] pricing” | Cited #1 | Absent | Cited #2 | Cited #1 | ▲ improved |
| “is [category] worth it for [use case]” | Mentioned, no link | Absent | Absent | Mentioned, no link | — flat |
| “[category] security / compliance” | Absent | Cited #3 | Absent | Absent | — flat |
What changed (SAMPLE)
- “best [category] for [use case]” (Gemini, Claude): newly cited/mentioned this month. Excerpt: "…for teams in [use case], [Brand] is frequently positioned alongside [Competitor] for its [differentiator]…" — first citation on this query in our tracked window.
- "[competitor] alternatives" (Gemini, Claude): declined. The competitor published a pricing page this month; both answers now cite it for tier detail. Fix queued: pricing transparency pass on our own pricing page.
- "[your product] vs [competitor]" (ChatGPT): still absent. Perplexity/Gemini/Claude stable. No change in the answer’s source set — this is a content-depth gap, tracked since month 2.
Top 3 fixes (SAMPLE, ranked by revenue impact)
- Pricing transparency pass (targets “[competitor] alternatives”, “best [category]…”): publish tier-level detail; verify against the month-3 run.
- Comparison page depth (targets “[your product] vs [competitor]”): add the three missing trade-off sections the competitor’s page covers; verify on month-3 run.
- Security/compliance page (targets “is [category] worth it…”, “[category] security / compliance”): concrete compliance claims with names attached; verify on month-3 run.
What we did (SAMPLE)
- Implemented: llms.txt v2 fix + markdown alternates on 12 pages; structured data added to pricing and 3 comparison pages.
- Shipped: docs restructure for 20 top-first-question pages (5 of 20 complete; remainder in flight).
- Did not happen: third-party review-platform gap (vendor profile update blocked on customer consent) — carried to next month, reason noted.
Copy the shape, not the content. Swap in your query map, your surfaces, your reasons. The sections are the contract; the rows are yours.
What never goes in it
- Traffic promises. No “projected AI traffic,” no “on track to 10x.” The published data says AI referral volume is small even while it grows — promising it is how a report loses credibility. The number is citation share; traffic is reported as an observed row, never as a target.
- Vanity screenshots. A screenshot of a dashboard, a “we got cited!” chat with no context, a cropped answer — none of it verifies anything. Verbatim excerpts tied to a query and a date, always.
- Unattributed numbers. Every external statistic in a report names its source and date. Every internal number reproduces from the query map and the raw captures. If a number cannot be traced to a capture or a named source, it does not go in the report.
- Lists without priorities. “15 things we noticed” is the dashboard dump again. Three fixes, ranked, or the section fails.
This is the same honest-data rule the whole site runs on: source and date on every number, sample data labeled as sample, and no promises the data doesn’t support. The report is where that policy becomes a deliverable instead of a slogan — the mechanics are documented in the methodology.
Next step: if you would rather we run it for you — the measurement, the diffs, the fixes, and the page, monthly — the free quick check is the same format with this month’s snapshot. And if you sell this format to your own clients rather than buying it, the white-label ops post covers how the pipeline runs under your brand.