Reading reviews and trusting star ratings

What does an AggregateRating actually tell you and what does it not? How do you read reviews critically, and which patterns indicate a reliable review trail?

Reading reviews: what do they and do not tell you?

This piece belongs to our pillar Neighbourhoods for discreet appointments and covers how you can interpret reviews on the platform — what the star rating means, how you distinguish approved from rejected reviews, and which patterns indicate a reliable review base. Reviews are one of the most widely used filters in our catalogue, so it is worth reading them correctly.

The basics: how we handle reviews

On the platform, reviews work via a three-step process:

1. Submitted by a registered user. We do not accept anonymous reviews. Every review is linked to an account that has itself been verified via e-mail confirmation. 2. Moderation. Every review passes a manual assessment by our admin team. Reviews that are not about the appointment, contain personal data of third parties, are spam, or are offensive in tone — are rejected. 3. Approved or rejected. Approved reviews count towards the star rating (AggregateRating in schema terms). Rejected reviews and reviews in moderation do not count.

This means that the star rating you see on a profile reflects only what has passed through our moderation. That is a deliberate choice — we do not want to be a platform where angry clients or jealous competitors can undermine a provider with spam reviews. It does introduce a gap: a provider with a high star rating may in practice have had just as many clients who had a neutral experience — those people simply would not have written a review.

What an AggregateRating does and does not tell you

In schema.org terms, the AggregateRating is the average star score across all approved reviews. What this means in concrete terms:

A provider with 25 approved reviews and an average star rating of 4.6 carries a more reliable signal than a provider with 3 approved reviews and a 4.9 — the samples are not comparable.
How long the profile has been active matters. A provider who has received 25 reviews in 6 weeks is a different signal from a provider who has received 25 reviews over 12 months.
A 5.0 star rating across a handful of reviews is less reliable in our data than 4.5 across a larger series. The reality of every appointment is nuanced; perfect every time is statistically unlikely.

We therefore give clients this practical heuristic: look not only at the star rating but also at the number of reviews. Three to five reviews is "new"; ten to twenty is "established"; more than twenty is "track record".

How to read an individual review critically

Not every approved review is equally informative. A few patterns we have seen as signals for more versus less reliable reviews:

More reliable reviews:

Specific details that only someone who was there would know (the general neighbourhood, the hotel brand without the exact room number).
A mix of positive and critical observations. An honest review typically contains both.
Concrete time indicators ("last week", "during a week's visit in October").

Less reliable reviews:

Generic superlatives without substantiation ("perfect", "fantastic", "the best") without any specific experience.
Reviews published within a few hours of the appointment. A short timeline is technically possible, but in our data we see that considered reviews are typically written a day or a few days later.
Patterns where multiple reviews from different accounts have identical sentence structures or word choices. Our moderation generally catches this, but in the early stages of a profile it may be visible.

What we changed in 2025

In the first six months after the catalogue went live, we made two adjustments to the review system based on what we observed in practice:

First adjustment. Initially, clients could submit a review as soon as they logged in, regardless of whether they had had an appointment. In our data we saw a pattern in which some profiles received an unnaturally high number of reviews in a short period — suggesting that a provider or her network was asking friends to write reviews. Since March 2025 we have required a minimum interval of 24 hours between registering an account and submitting a first review. That has noticeably reduced the pattern.

Second adjustment. We had no indicator for the number of reviews in moderation. This created a blind spot: a profile with "0 approved reviews and 8 in moderation" looked the same to a first-time reader as "0 approved reviews, none in moderation". Since September 2025 we display the number of in-moderation reviews (without revealing their content) on the profile. Clients can thereby gauge whether there is recent activity around a profile.

What we cannot do

An honest caveat: we do not detect every form of organised review fraud. In a catalogue of our size it would be virtually impossible to intercept every suspicious interaction between accounts before a review is published. We rely on a combination of manual moderation and pattern detection — and we know that we are not 100% effective in this.

For clients this means: trust reviews, but also trust your own judgement. A provider with good reviews and a first-contact experience that does not feel right remains a provider where you have every right to cancel the appointment. Reviews are a data point, not a guarantee.

GuidesReading reviews and trusting star ratings