Our scoring

Sources

We blend two perspectives:

• Press
• Players

On each game page both points of view are displayed along with the number of reviews that support each one; if one of them is missing then we use the other one by itself and never make an extrapolation. We avoid extrapolating missing data to ensure accuracy and fairness, so scores never include assumptions.

Composition

1. Perspective

We calculate a weighted average of press and player scores, weighting each score by how well reviewed that group is. Each group reaches nearly its full weight after about 20 press reviews or 150 player votes, so 100,000 player votes do not outweigh 100 press reviews. A high press score and a lower player score, or the reverse, will result in a score between the two.

2. Confidence

A small number of votes matters less than a large number; scores from small groups are subtly moved toward the average to prevent outliers. We apply a Bayesian adjustment, where games with fewer reviews have their scores blended with the overall average, making the result more reliable until enough data is available. That is why you may see games with 100% player approval, but a lower score. The opposite is also possible.

3. Rarity

Players usually award higher scores than the press, so we adjust by treating a rare player score the same way we treat a rare press score. To do this, we calculate how common each score is among all reviewed games, separately for players and for the press. A high score that is given out frequently by players is considered less rare, while a similarly high press score is rare and carries more weight. If a game has a 90% approval rating from players, but many games reach that score, it will not boost the overall score as much as a 90% press score, which happens less often. This adjustment helps align the scale between players and press and makes the scoring fairer, so the final score takes into account how uncommon each result is, not just its absolute value.

4. Evolution

Player ratings have generally increased over time, so to ensure fairness, we compare each game to others released during the same period.

Limitations

Since the players in each game decided to buy it, the player base for each game is unique. Different groups gave the same approval to the games. Although we have tried several corrections, none of them was fair to all the games. At present, we have not found a better solution.

For some games, there are no player reviews. Since some platforms have only a small number or none at all, games which are sold exclusively on those platforms receive their ratings from the press only.

When the press and the players are in disagreement, one figure isn’t capable of showing that; the score comes out in the middle, and the game page gives both points of view so that you can see the difference.