What 2,604 Reviews Look Like When You Read the Bad Ones

An average rating is the least useful number attached to any studio. 4.9 out of 5 tells you almost nothing on its own: it is the same figure whether it comes from 2,000 projects or twenty, and it flattens every interesting disagreement into a decimal point.

The distribution underneath it is where the information is. Here is ours, and what we think it actually says.

The shape

Across 2,604 reviews: 2,486 at five stars, 103 at four, 10 at three, 1 at two and 4 at one. That is 95.5 per cent at five stars, and 118 reviews — about one in twenty-two — where the client had something less than complete satisfaction to record.

The first thing worth saying is that those 118 should exist. A studio with several hundred completed projects and a spotless record has either been extraordinarily lucky or is operating a filter. Design is subjective and clients are human; some proportion of any real practice ends in partial disappointment. A record that shows none of it is not evidence of quality.

What the four-star reviews are actually about

The 103 four-star reviews are the most useful group, because four stars is almost never about the work being wrong. It is about the experience having friction — a round that took longer than expected, a brief that turned out to mean something different to each side, a client who wanted more concepts than the tier they bought included.

That last one is the most common and the most preventable, and it is a pricing-design problem rather than a craft problem. When someone buys a package with one concept and expects three, the work can be excellent and the review will still land at four. We have since made the concept count the first thing stated on each tier for exactly this reason.

What the low ones teach

The 15 reviews at three stars or below are a small enough group to think about individually rather than statistically, and the honest summary is that most of them come down to expectation rather than execution.

The recurring pattern is a brief that was thinner than it looked. A client who says "modern and clean" has not told you anything — every studio believes its work is modern and clean — and when the first concepts arrive, the gap between what they pictured and what was described becomes visible for the first time. That is a failure of the intake process, and it belongs to us rather than to the client.

It is why the brief form now asks what the business actually does and who it sells to, rather than asking you to pick adjectives. A studio that lets you commission work off three vague words is setting up a disappointment it will later describe as a difficult client.

How to read anyone's reviews, not just ours

Four things are worth doing when you are evaluating any studio.

Read the three-star reviews first. One-star reviews are often about something that went catastrophically wrong once — a missed deadline, a payment dispute — and tell you little about typical work. Five-star reviews are frequently written in thirty seconds. Three-star reviews are written by people who were engaged enough to be specific and unsatisfied enough to be honest.

Check whether the response exists and what it says. A studio that replies to criticism by explaining what it changed is telling you something real. One that replies by arguing with the client is telling you something equally real.

Look at review velocity, not just volume. 2,000 reviews accumulated over years means something different from 2,000 in six months, and both mean something different from a studio whose recent reviews have thinned out.

Discount anything unverifiable. Testimonials on a studio's own website — including ours — are selected by the studio. They are not worthless, but they are marketing. Reviews attached to a completed transaction on a platform that verifies it are a different category of evidence, and that difference is why we have never marked our rating up as review schema on this site. Google treats reviews an organisation hosts about itself as ineligible for star display, and we agree with the reasoning.

The number we would actually judge a studio on

If we could only see one figure about a competitor, it would not be the rating. It would be the proportion of clients who came back for a second project.

A logo can be sold once to someone who never finds out whether it worked. Guidelines, packaging, a social kit and an animated version get commissioned by people who have lived with the mark for a while and decided to build on it. That is the closest thing to a verdict this industry produces, and it is the reason our portfolio carries as many multi-piece brand systems as standalone marks.