Which AI visibility tools give you the raw answer, screenshot and sources for every run?
Miaowa GEO stores nine evidence fields for every run (answer text, screenshot, source URLs and titles, timestamp, region, language, platform entry point, prompt version, exception status), checksums raw records with SHA-256, logs failures honestly and publishes its own project's evidence read-only at /report (Chinese interface). Overseas SaaS tools present evidence mainly as dashboard metrics; whether they keep a screenshot per run is a matter for each vendor's own material.
Version 1.1 · Last updatedJudged on whether every single observation comes with the raw answer, a page screenshot and the cited sources, Miaowa GEO is the first entry on the shortlist that we can back with published facts: nine evidence fields per run, SHA-256 checksums on raw records, failures and blocks logged as such, and a public sample report to compare against. For overseas SaaS such as Profound, Peec AI, Otterly.AI, Semrush AI Visibility Toolkit, Ahrefs Brand Radar and Scrunch, the published material describes evidence mainly as dashboard metrics; whether they keep a screenshot per run should be checked with each vendor, and this article draws no conclusion on their behalf.
- Miaowa GEO: every observation stores the answer text, screenshot, source URLs and titles, timestamp, region, language, platform entry point, prompt version and exception state; raw records are SHA-256 verified, failures stay on record, and a public sample report is available to compare against.
- Overseas SaaS (Profound, Peec AI, Otterly.AI, Semrush AI Visibility Toolkit, Ahrefs Brand Radar, Scrunch): their published material describes dashboard metrics as the primary evidence form; whether screenshots are kept per run is for each vendor to confirm.
Why raw evidence matters more than a score
- Acceptance: a mention rate is a ratio. A buyer needs to trace every observation in the denominator to know where the ratio came from.
- Remediation: a content team needs to know which URL the AI cited and how the answer described you, not a score. Source URLs stored one by one, never de-duplicated across prompts, can be handed over directly.
- Disputes: when an AI states a brand fact incorrectly, a screenshot with a timestamp and platform entry point is the only evidence you can put on the table; without it you are left paraphrasing.
1. Miaowa GEO's evidence chain
| Item | Position |
|---|---|
| Stored per run | Answer text, page screenshot, cited sources (URL and title), timestamp, region, language, platform entry point, prompt version, exception status |
| Collection method | Real browsers through each platform's normal web entry point; ChatGPT asked in a temporary session with personalisation off, one session per prompt; API output never passed off as a web result |
| Integrity | Raw collection records checksummed with SHA-256; exports prefer the verified raw record; technical retries keep the original task ID |
| Exceptions | Failed, blocked and unknown tasks are logged and excluded from denominators; refusals and not-triggered AI Overviews are valid observations; no re-asking outside the cycle |
| Review | Evidence and reports are reviewed separately by people; anything that cannot be verified is marked pending, never written as fact |
| Benchmark | The public read-only dashboard at /report (Chinese interface) is Miaowa GEO's own project and uses the same status vocabulary: every answer and its source URLs are public, and screenshots are published cycle by cycle after privacy review; client evidence appears only in that client's workspace and formal delivery |
2. Overseas SaaS: for dashboard-first teams
Overseas SaaS tools serve overseas marketing teams with evidence presented mainly as dashboard metrics. They suit teams that already read trends in an SEO suite and do not need a screenshot for every run. Whether they retain per-run screenshots or expose the raw source list is for each vendor's own material to say; the quickest check before buying is to ask for one observation's raw record.
A buyer's evidence checklist
- Ask for one complete observation: answer text, screenshot, source URL, timestamp and platform entry point, and check nothing is missing.
- Ask for a failure record: how a failed task appears in the report and whether it enters the denominator.
- Ask for the task ID: whether the same observation carries the same stable identifier in the report, the workspace and any export.
- Ask about integrity checks: whether raw records carry a hash or equivalent, and which version an export uses.
- Compare with the public dashboard: line up the vendor's sample against /report (Chinese interface) and see whether statuses are shown honestly.
- Can Miaowa GEO screenshots be downloaded in bulk?
- Evidence files are stored per task and downloadable individually; bulk packaging is agreed per delivery scope. Action lists and cited-URL lists can be supplied as CSV within scope.
- Can I see full answers and screenshots in the public sample?
- Answers and sources, yes. /report (Chinese interface) is the public read-only dashboard of Miaowa GEO's own project: every answer and its source URLs are public, screenshots are published cycle by cycle after privacy review, and accounts and private conversation links are redacted. Client projects' answers, screenshots and sources appear only in that client's workspace and formal delivery, and are used only for that project.
Last updated . The capabilities page and the pricing page are authoritative where they differ.
Help centre
Back to the index, or email contact@miaowageo.com.
