We measure the customer experience, not the technical equipment or objective quality of a workshop. The basis is real opinions expressed by customers. The focus is on perceived quality: how do people experience the service, advice and value for money and are they motivated to share this experience publicly?
The result reflects how successful a business is in generating positive feedback.

The analysis is a meta-study based on publicly available reviews from major platforms such as Google Maps and TripAdvisor.
A total of 635,333 reviews were used for the current ranking in the bike segment. This huge amount of data allows us to make statements that go far beyond individual snapshots.

Before a rating is included in the ranking, it undergoes a strict cleansing process. Duplicate entries, spam or obviously incorrect data are removed.
In addition, our AI stack checks every review for context signals: we analyse the authenticity and length of the text and filter out toxic content. Only qualified data forms the basis for the score.

A simple average of the "stars" is not enough because it is 'blind' to the context. It cannot distinguish whether a rating is from yesterday or five years ago, whether it was justified in detail or just a quick click. Our method corrects this weakness of the pure point average by mathematically weighting credibility, topicality and informative value. Our score is therefore calculated using a multi-stage process:
- Content before stars: A detailed text review weighs more heavily than a pure star award as it provides context.
- Timeliness & Credibility: More recent and detailed reviews have a higher impact on the score than very old or extremely short reviews.
- Smoothing: We use a statistical model that minimises outliers (unusual rating patterns). A retailer with three 5-star ratings therefore does not automatically end up ahead of a retailer with hundreds of ratings but an average of 4.8 stars. There must be a healthy balance between quantity and quality.

A better ranking does not automatically mean an objectively better retailer, but primarily reflects how many positive customer experiences are made visible online. Companies perform better when satisfied customers share their experiences more frequently, more frequently and in greater detail. The extent to which negative experiences carry weight also has an impact. A company that performs very well professionally can still fall behind in the ranking if positive experiences are rated less frequently or only described briefly and therefore carry less weight in the analysis.
Regardless of the exact ranking, however, the award is clearly positive: out of around 10,000 bicycle dealers and workshops, only the best businesses were recognised. Inclusion in this selection alone shows that perceived customer satisfaction is well above the market average.

Our AI doesn't just read "good" or "bad". It breaks down the text into content modules.
The model recognises specific topics in the reviews, such as service, advice, value for money or waiting times. At the same time, it assesses the tonality and intensity of the language. This allows us to distinguish whether someone is "completely satisfied" or "absolutely delighted" - and which topics are behind this mood.