Gameplay insights · July 31, 2026 snapshot

What we learned from 1,500 game starts

Undercover Word records a small set of anonymous product events so we can understand whether the game works for real groups. This is a transparent look at the first useful snapshot—and at what the numbers cannot tell us.

Data snapshot

1,517

game-start events

439

game-completion events

5

average players per event

3:37

average completed-game duration

The typical recorded group had five players

Across start and completion events, the average player count was five. That supports the practical advice in our hosting guides: a five-player group is large enough to produce several different clues but still small enough for every clue to remain memorable during discussion.

An average is not a recommendation by itself. It does not reveal the distribution of group sizes, and repeated games from the same group can appear more than once.

A recorded completed game averaged 3 minutes 37 seconds

The 217-second average reflects completed events with a duration. It suggests the core loop is genuinely quick: word reveal, one clue per active player, discussion, and a vote can fit into a short break or serve as one round in a longer game night.

This figure can be influenced by very short or long sessions. It is not a median, and it should not be interpreted as a guarantee for every group.

Start and completion events are not a conversion funnel

The snapshot contains 1,517 starts and 439 completions, but dividing those values would not produce a reliable abandonment rate. Tracking changed as the product evolved; browsers can close before sending a completion event; repeat rounds create additional events; and network or consent settings can prevent measurement. We use the counts to identify broad product patterns, not to claim a unique-user completion percentage.

Methodology and privacy

  • A start event contains event type, player count, and game mode.
  • A completion event may additionally contain duration and the round outcome.
  • The analytics event does not contain player names, room codes, assigned words, email addresses, or account IDs.
  • Events are anonymous product records, not verified unique people or controlled experiments.
  • The snapshot may include automated, duplicate, interrupted, or otherwise imperfect events.

For details about data handling, read the privacy policy and editorial standards.