What the template is, next to the checklist and the report
Three documents run a defensible audit, and conflating them is why most downloadable audit templates disappoint. The template is the workbook: the structure you fill while doing the work. The checklist grades that workbook after the fact; ours is the 47-point scoring model, and it assumes a workbook exists to grade. The report is the client-facing compression of the workbook; the report template covers that layer. Agencies searching for a GEO audit template usually want all three and download whichever one they find first, which is how audits end up with a beautiful report format wrapped around an evidence record that never existed.
Most templates currently circulating are also tool assets. The strongest vendor example, Ahrefs’ audit report template, is built around its own Brand Radar data, which is labeled clearly enough but means the method travels only where the subscription does. A workbook should survive a tool change, because your evidence discipline is the part of the audit the client is actually paying for.
The eight sections and what each one forces
The worksheet has 34 fields across eight sections. Each row carries the field, what to record, the evidence artifact to save, and a status. The sections exist in dependency order: a later section filled before an earlier one is the template telling you the audit skipped something.
| Section | What it forces | Fields |
|---|---|---|
| Engagement scope | One business decision under test, platforms with account state, geography, written exclusions, recorded consent | 6 |
| Approved facts | A client-approved ledger of names, offers, locations, and sourced claims to grade answers against | 5 |
| Technical eligibility | Public responses, crawler rules, index state, rendered text, structured data agreement, CDN interference, each with dated evidence | 6 |
| Prompt panel | Decision-derived prompts, branded and unbranded strata, a freeze date, a run plan | 4 |
| Observation log | Complete run records including absences, mention and citation classified separately, accuracy graded against the ledger | 4 |
| Source map | Cited URLs inventoried, ownership classified, competitors observed on the identical panel | 3 |
| Findings | Observation-grounded findings, prioritized actions with owners, method limits in the main document | 3 |
| Handoff | A locked rerun specification, an evidence archive, client acceptance | 3 |
Two sections do most of the differentiating work. The approved facts section exists because representation review, the part clients care about most, is impossible without client-approved facts to grade against; an auditor without a ledger is grading AI answers against their own guesses. The handoff section exists because the second audit is where sloppy first audits die: a rerun with a drifted panel or changed account state produces a trend chart that measures the auditor, not the client.
AI search visibility audit templateAll eight sections with every field, what to record, the evidence artifact to save, and a status column. Duplicate the file per client and per audit wave; the structure is the reusable part.
CSV worksheet, 34 fields Grade the filled workbook against all 47 checksThe interactive checker runs the scoring model this template is built to satisfy: confirm what your workbook already records and see every unconfirmed check ranked by weight, with the evidence that would close it.
Score your protocolFilling the technical section without inventing requirements
The technical eligibility fields deserve a note, because this is where audit templates most often import myths. The documented reality is narrower than most worksheets assume: Google states that appearing in AI Overviews and AI Mode requires being indexed and snippet-eligible under the normal Search technical requirements, with no special files or markup, and ChatGPT search eligibility runs through an ordinary crawler allowance documented in OpenAI’s crawler docs. So the template’s technical fields check access, indexability, rendered text, and controls, all things the platforms document, and none of it asks you to score a page against an invented AI readiness standard. The full crawler decision logic lives in OAI-SearchBot vs GPTBot, and the myth-versus-requirement sorting for Google’s side is in the AI Overview technical requirements guide.
Every technical field requires a date on its evidence. “Robots is fine” is not a finding; a robots.txt capture with a body, a status code, and a timestamp is, because the next site deploy can silently change the answer.
Reusing the template across clients
The reusable parts are the structure, the field definitions, and the evidence discipline. The non-reusable parts are everything the fields contain: the business decision, the fact ledger, the prompt panel, and every observation. A template reused correctly produces audits that look identical in shape and completely different in content.
Three practices keep multi-client reuse honest. Duplicate the file per client and never edit a delivered copy, because a delivered workbook is evidence. Version the template itself separately from any engagement, so a field improvement does not retroactively change what an old audit claims to have checked. And keep the prompt panel out of the template file: panels are derived per client from their buyer decisions, and the derivation method in the prompt set guide is the part that transfers, never the prompts themselves.
The template deliberately promises nothing about outcomes: filling it proves what was checked and observed, not that citations or rankings will follow, and that boundary belongs in the scope section in writing. Price the work the workbook actually demonstrates with the audit pricing model, and when the audit closes, the handoff section’s rerun specification is what turns a one-off project into a measurable baseline for the next wave.