AI visibility audit worksheet

By Abiot Y. Derbie - Founder, RankEcho - Updated 2026-09-02 · Sources checked 2026-09-01 · Independent reviewer: unassigned · No independent external review is claimed.

audit_version,example_status,observation_id,cell_id,panel_version,window,prompt_id,repeat_index,prompt_text,surface,locale,account_state,planned_cell,availability,run_status,collected_at_utc,access_evidence_artifact,answer_artifact,visible_source_urls,brand_mentioned,owned_link,other_citation,recommendation,absence,named_competitors,exclusion_rule,competing_explanations,owner,next_decision,notes
1.0,synthetic,SYNTHETIC-OBS-001,SYNTHETIC-CELL-001,SYNTHETIC-PANEL-1,before,SYNTHETIC-PROMPT-001,1,"Which inventory platform fits a two-person operations team?",declared-example-surface,en-US,signed-out,yes,available,observed,2026-08-15T09:00:00Z,synthetic-access-before-001.json,synthetic-answer-before-001.txt,https://example.org/roundup,no,no,yes,no,no,"Example Rival A|Example Rival B","Exclude only unavailable or failed attempts from observed-outcome rates.","Crawler access|prompt fit|entity clarity|ordinary answer variability",unassigned,"Verify the selected hypothesis","Illustrative observed row only; replace every value and retain the raw answer outside this CSV."
1.0,synthetic,SYNTHETIC-OBS-002,SYNTHETIC-CELL-001,SYNTHETIC-PANEL-1,after,SYNTHETIC-PROMPT-001,1,"Which inventory platform fits a two-person operations team?",declared-example-surface,en-US,signed-out,yes,unavailable,not-run,2026-08-25T09:05:00Z,synthetic-access-after-001.json,,,,,,,,,"Unavailable surface; exclude this attempt from observed-outcome rates.","Provider availability|account state|collection failure",unassigned,"Re-run only under the registered rule","Illustrative unavailable cell; outcome fields are blank because no inspectable answer exists."

Scope box: what does this page cover?

One row per planned prompt-surface-repeat-window cell, joined across windows by a stable cell ID.

The worksheet structures observations; it does not certify completeness or explain a provider's selection.

What are the instructions?

Download or copy version 1.0, replace every synthetic value, and register the panel before collecting the after window. Use one row for every planned cell in each window, including unavailable and failed attempts; reuse the same cell ID for the same prompt, surface, conditions, and repeat index. Keep access records and raw answers in a stable evidence store referenced by the worksheet, and leave answer-outcome fields blank when no inspectable answer exists.

Free audits check Perplexity and Gemini once per prompt. Paid and trialing accounts add ChatGPT, Claude, and Google AI Overviews when those adapters are configured, for up to 5 engines. These product defaults do not determine the scope of a separately run external protocol. Google AI Overviews is reported only when an overview is shown and retrievable; otherwise that cell is marked skipped/unavailable rather than counted as a citation miss.

Which fields are included?

The preview groups fields by the decision they support.

What does the worked example show?

The two included rows are illustrative and synthetic. They reuse one stable cell ID across before and after windows and each points to its access-evidence record. The first also points to an observed answer with a third-party URL and named example rivals; the second records an unavailable cell, so its answer artifact, visible URLs, and outcome fields are blank. It remains in the planned-cell count and outside available-case and common-cell outcome denominators.

Replace the example domain, prompt, surface, timestamps, and outcomes. Do not present the sample rows as customer evidence or a benchmark.

How do you copy or download the worksheet?

Use the CSV link for an editable local copy. A renderer may also expose the versioned copy text from the asset specification; both actions return the same deterministic payload.

What deployment validation and proof checks follow completion?

Check that each planned cell has one availability/run state, a retrievable access-evidence pointer, and a UTC timestamp; require retrievable raw-answer pointers for observed cells. Verify that URLs follow a frozen ownership rule and that unavailable or failed rows have blank answer-outcome fields and stay outside available-case and common-cell outcome rates. Preserve the worksheet version and stable cell-key rule used for the baseline.

The worksheet feeds Find and the fixed-prompt retest. It cannot show that a policy, page, or source caused an outcome; link the accepted deployment and report later observations with uncertainty.

What is the version history?

Version 1.0 — 2026-09-02: initial pilot CSV with stable cell IDs, window and repeat fields, availability, access-evidence and answer-artifact pointers, non-overlapping outcomes, exclusions, alternatives, owner, next decision, and a visibly synthetic example.

Where does this asset sit in the workflow?

Use it in Find to freeze the observation record, hand its IDs and evidence to a bounded Fix ticket, and reuse the same panel version in Prove.

Who owns the page, review, tests, and corrections?

Author and accountable publisher: Abiot Y. Derbie, RankEcho founder. Published 2026-09-02; updated 2026-09-02. Independent reviewer: unassigned. No independent external review is claimed.

Tested scope: CSV columns, deterministic payload parity, synthetic row labels, and links; users remain responsible for validating their collected evidence. Material provider and standards statements use the dated source records below. Examples are synthetic, not customer results. Send corrections through /contact; material corrections should update the visible date and version history.

Sources reviewed

Provider eligibility and measurement claims below were checked against primary documentation. These records do not establish a universal selection formula, causation, or a guaranteed ranking, impression, recommendation, or citation.

Frequently asked questions