Market Score v1.2 · pending legal and beta review

How a score is earned.

A plain-English summary of the research score method. Evidence comes first; if there is not enough, we show no number.

Evidence before a number

Member-initiated research applies this method when enough fresh evidence is available. Provider or evidence gaps keep the score hidden. Weekly scheduling is not active. The 72 in the demo workspace is fictional.

What it measures

How discoverable a business is in public sources. It does not measure quality, revenue, or how consumers rank local businesses.

50% AI visibility

Fixed category-and-location questions are asked to OpenAI, Gemini and Perplexity, three times each. We count how often the confirmed business is named, matched by name plus location or website, never name alone. Each engine counts equally.

30% profile match

Six exact comparisons between facts you save and your matched Google listing: name, phone, website, full address, the exact displayed Google category, and weekly hours. A confirmed Google listing match is required. Phone, address, category and hours must all be saved for profile scoring. Google details are compared transiently; only a yes, no or unknown result is kept, never Google text. A provider failure counts as unknown, not wrong.

20% website review (v1.2)

Ten equal checks on the page fetched from your saved website, only with your permission: business name, contact details, products or services, location or service area, page title, search description, main heading, mobile viewport tag, secure connection, and a clear contact action. Each is worth 10 points of the website value.

Title (at least 10 characters), meta description (at least 40) and main heading (at least five) are simple heuristics, not SEO requirements. This is not a speed, full-site or accessibility audit. Unknown is not zero.

Earlier v1.1 reports used four literal checks below and keep their original scores; each report shows its method version.

v1.1 website checks (historical)

Four literal checks on your own saved website, fetched directly only when you tick the permission box: business name, phone, the short product or service phrase you saved, and service area. Each match keeps a short quote, the page address and the date. Phone and the phrase must be saved for website scoring. We never fetch competitor sites directly for scoring and run no fresh competitor crawl; a fetch failure counts as unknown.

Score Enhancers

Profile posts, website FAQ answers and review responses drafted from your facts. Generating, approving or copying them does not change the score and no gain is guaranteed. Once you publish, a fresh research run rechecks the measured signals.

When no score is shown

The score stays hidden unless at least 80% of the weight is measured, two engines completed all repetitions for every question, and your identity is confirmed. Missing data never counts as zero and is never filled with sample results.

AI checks expire after 14 days; website and profile checks after 7. A one-check free scan normally shows its real results and "insufficient data for a score."

Ranges, not certainty

We would show mention counts like 1/3, run dates, the exact question, and a 95% range. Repeated model calls are a small experiment, not a survey of customers.

History

Each weekly score keeps its inputs, links, dates and method version. Weeks are compared only when methods and coverage match. No invented past trend.

Before launch

Weights will be reviewed with real owners in beta, and any version change will be published here. v1.2 changed only the website checks; weights, coverage thresholds and freshness windows are unchanged.