How it works
The trust section — what we measure, and what we refuse to fake.
Most tools lose your trust once with a confident number that turns out to be invented. The rules below exist to make that impossible, and they are enforced in code rather than in copy.
1. Your data goes in
Engines read from business profiles you create — industry, target segments, geography, budget range, AOV, contribution margin. Nothing is matched or scored before you supply it, and one profile is marked primary.
2. Sources are fetched or ingested
Every external source is classified (PUBLIC, RSS, OPEN DATA, API, USER SUBMITTED, LICENSED, UNKNOWN) and must be robots.txt- and terms-compliant. The ingestion worker validates, triangulates, and scores confidence before a row is ever published.
3. Engines compute, with visible formulas
Scoring is plain arithmetic with documented weights — no paid AI API and no black box. Where a rule changed during development, the correction is recorded in the rules document rather than quietly patched.
4. Results carry their provenance
Each stored score is stamped with the engine name and version, plus the formula digest. Confidence labels travel with the number, and inputs are labelled measured, benchmark, or manual.
5. You act on a plan, not a score
Findings and matches become prioritised candidates. A generated Action Plan turns them into owned, dated items you can tick off — and completing a plan records the Growth Score at that point.
The limits we publish
These are not fine print. Each one is a deliberate refusal that makes the product less impressive and more trustworthy.
- A check we cannot measure returns not-measured. It never scores zero, and it never gets filled in with an assumption.
- If a catalog has no ingested rows for your city or category, the list is empty. We do not seed it with plausible-looking entries.
- Fees, rates, and audience sizes appear only when you provide them or a verified source does. We do not estimate them.
- The Visibility Radar reads one page and a bounded set of internal links. It does not see your analytics, rankings, or competitors.
- Scans are labelled USER SUBMITTED — a finding describes what we observed, not a claim about your business.
- A confidence score below the publish threshold keeps a row out of your feeds entirely.
Where your data lives
This is a single self-hosted deployment: one Next.js application and one PostgreSQL database, both operated by us, with no third-party analytics or advertising scripts on the app. Email goes through the deployment's own outbox, which prints to the server log until an operator configures a mail server. Sessions are opaque tokens stored hashed; passwords are Argon2id. Details, including retention windows, are in the privacy notice.
Audit a site for free and judge for yourself
The free scan returns a score and its top findings. What it cannot measure is stated as such, in the report.