# TrustScope
> TrustScope reads the public Google reviews behind a venue's star rating and reports a trust score: whether those reviews carry the signals of real, engaged reviewers, what people actually talk about, and how the reviews spread over time. A separate, older surface tracks Google review deletions city by city. Currently covering Germany, expanding further.

## About
- Publisher: Edem Devlet (@DevletEdem on X)
- Contact: support@trustscope.app (also https://x.com/EdemDevletEN)
- Venue owners: a venue's stored analysis can be removed on request under Art. 21 GDPR - email support@trustscope.app with the subject "GDPR: Objection (Art. 21)" and the venue's Google Maps link. A suppressed venue's page stays reachable but is served `noindex`. See /en/contacts.
- Review authors: review text, author names and profile pictures are deleted as soon as an analysis has been computed. Only Google's own review identifiers and aggregate numbers are retained - no text, no names, no pictures.
- Data source: the public reviews on a venue's own Google Maps listing. Nothing other than those reviews determines the score.
- Last updated: 2026-09-01
- Coverage: Germany. A country is offered only once it holds at least 10,000 analysed venues; today that is Germany alone. Roughly 600,000 German venues have been analysed, across every city with enough of them to rank.
- Languages: English and German (Englisch und Deutsch). Every page exists under /de/ (default) and /en/; the bare root / redirects to /de.
- Browser extension: "Worst-Case Rating" for Chrome (https://chromewebstore.google.com/detail/worst-case-rating/kljgcgijgpadigheeipmebfolfophfdg) and Firefox (https://addons.mozilla.org/en-US/firefox/addon/worst-case-rating/) - shows the trust score and the worst-case adjusted rating directly on the Google Maps page you have open. No data is collected.

## The two products on this site
TrustScope is primarily a **review-analysis** product. A second, older surface - the deleted-reviews map - is still published but no longer the main thing the site is about. They answer different questions and share no data:

1. **Review analysis (the main product).** Lives at `/`, `/categories`, `/search` and `/venue/{slug}`. Given a venue, it scores the credibility of its Google reviews and breaks down what those reviews say. This is what "TrustScope" and "trust score" refer to.
2. **Removed Reviews Map (secondary).** Lives at `/map` and `/statistics`. Given a city, it maps venues whose Google listing shows a review-removal notice and recomputes a worst-case rating as if every removed review had been 1 star. Nothing on the home page, the category pages or a venue page relates to it.

## Trust Score - how it is computed
- Sample: a venue's 500 most recent Google reviews. A venue is analysed only once it has at least 50; below that no analysis is produced at all.
- Each review scores 10-100 points: a base of 10, plus text depth (up to 55, logarithmic, full credit at ~10 words), reviewer history (up to 15, logarithmic, full credit at ~50 reviews written by that author), an attached photo (10), and a real uploaded profile picture rather than Google's default avatar (10).
- The venue's score is the average of its reviews' scores, minus a penalty of up to 10 points for the share of reviews matching an unusual pattern. That penalty is logarithmic and caps at a 10% share.
- Bands (same thresholds as the browser extension): >= 70 "Credible reviews" (green), >= 55 "Mostly credible reviews" (light green), >= 40 "Weak signals" (orange), below 40 "Something looks off" (red), no analysis = grey.
- What it means: the trust score measures credibility *signals*, not truthfulness. A low score means the reviews are thin, anonymous or photoless. It is not evidence that any individual review is fake, and it is information rather than a decision - nothing is granted, refused or decided on the basis of it.

## Other things a venue page reports
- **Unusual patterns**: thin 5-star reviews clustered on a single day, counted across all days and shown as a share of the sample. A statistical pattern only, never a claim about a specific review or author.
- **Review-quality signals**: four percentages of the sample - detailed (>= 10 words), with photos, real profile picture, established reviewers (author has >= 5 reviews). These are the raw ingredients behind the trust score.
- **Aspect radar**: five topics chosen by the kind of place - e.g. restaurants/cafes/bakeries/bars get Food, Service, Atmosphere, Speed, Cleanliness; hotels get Rooms, Service, Cleanliness, Location, Value; clinics/doctors/dentists get Care quality, Staff, Waiting time, Cleanliness, Communication; pharmacies, gyms and hairdressers/spas each have their own. Each axis is a *sentiment ratio*, not an average star rating: 50 is balanced, above 50 means more positive than negative mentions.
- **Rating distribution**: how many of the collected reviews gave each star rating. Reflects the most recent 500, so it can differ from the lifetime average Google displays.
- **Review volume over time**: reviews per day over the last 30 days, or per month across the collected history. Descriptive only - nothing on it is flagged.

## Known limitations
- Recency bias: the sample is the newest 500 reviews, not a random sample of all of them.
- Sentiment is read in English via Google's own translation of each review; an idiom may not survive it. The trust score, the quality signals, the star distribution and the unusual-pattern rule are language-independent.
- Mention is not attribution: within one sentence ("great service, awful food") both sentiment words count for both topics.
- Below 50 reviews, no analysis exists.
- Everything here describes patterns. It surfaces reviews worth a closer look; it does not determine that any specific review is fake.

## Pages
Every path below exists under both /de/ and /en/.
- /de , /en : Home - search a venue, browse categories and recently analysed venues
- /de/search , /en/search : Venue search (noindex - a query, not a document)
- /de/venue/{slug} , /en/venue/{slug} : One venue - trust score, quality signals, aspect radar, rating distribution, contact details, hours. Server-rendered per request; there are roughly 1.2M of these URLs and they are listed in the sitemap, not in an on-site index
- /de/categories , /en/categories : Every kind of venue covered, with the specific Google types filed under each
- /de/categories/{category} , /en/categories/{category} : One category ranked by trust score (e.g. /en/categories/restaurant)
- /de/categories/{category}/{city} , /en/categories/{category}/{city} : That ranking narrowed to one city - "Best restaurants in Bremen" (e.g. /en/categories/restaurant/bremen)
- /de/what-is-trustscope , /en/what-is-trustscope : What TrustScope is and the problem it solves
- /de/how-trustscope-analyzes , /en/how-trustscope-analyzes : Full methodology - every formula and threshold above, in detail
- /de/about , /en/about : Who is behind the project and why it exists
- /de/contacts , /en/contacts : Contact and GDPR requests, including the Art. 21 objection route for venue owners
- /de/partnerships , /en/partnerships : Integrations, data exchange and collaboration
- /de/extension , /en/extension : The Worst-Case Rating browser extension
- /de/extension/faq , /en/extension/faq : FAQ about the extension and the analysis it shows
- /de/map , /en/map : Removed Reviews Map - the interactive deletion map
- /de/statistics , /en/statistics : Germany-wide deletion-rate rankings, plus the index of every city page
- /de/{city} , /en/{city} : Per-city deletion statistics (e.g. /de/munich, /en/berlin)
- /de/{city}/{type} , /en/{city}/{type} : Per-city, per-type deletion statistics - types are restaurants, cafes, clinics, hotels (e.g. /de/munich/restaurants)
- /de/map/methodology , /en/map/methodology : How deleted Google reviews are collected and how the worst-case formula works
- /de/map/faq , /en/map/faq : FAQ about deleted Google reviews and the map
- /sitemap.xml : Sitemap index. Children are gzipped (sitemap-pages-N.xml.gz, sitemap-venues-N.xml.gz)

## Categories
The 13 top-level categories every venue is filed under: Restaurants, Cafés, Bars, Bakeries, Hotels, Clinics, Doctors, Dentists, Pharmacies, Gyms, Spas, Hairdressers, Other. Each maps to many specific Google types ("Ramen restaurant", "Sushi restaurant", …), which the category page lists.

## Removed Reviews Map - how it works
Google allows businesses to request removal of reviews under local defamation law. This surface makes that process transparent: for every venue whose Google listing carries a removal notice, it shows how many reviews were removed and what the rating would be in the worst case.
- Formula: `adj_score = (visible_reviews × visible_rating + deleted_estimate × 1) / (visible_reviews + deleted_estimate)`, where `deleted_estimate` is the midpoint of the range Google discloses - for "201 to 250 reviews removed", 225.5.
- It is a deliberate worst case, not a measurement: it assumes every deleted review was 1 star. The true rating lies between it and what Google currently shows.
- Data source: the deletion notices Google itself publishes on venue pages in Google Maps. Only metadata is stored - current rating, visible count, disclosed deletion range. Review texts are not collected.
- A deletion notice is not evidence the venue did anything wrong.
- Venues are colour-coded by deletion severity: slate-grey = no deletions, then a light-to-dark blue gradient where darker means a larger share removed.
- The statistics view ranks by highest deletion rate, worst worst-case rating, and cleanest venues.
- Map fields: `total_reviews` (visible count), `deleted_reviews_min`/`max` (from the notice), `percentage_deleted`, `real_score` (the worst-case rating), `review_notice` (Google's own notice text), `venueType`, `city`.