Story and pacing
Story Choice Stats: Do Your Choices Really Matter?
Count what players chose and which endings they reached, then compare with the story as written. A choice nearly all answer alike is hardly a choice.
Pacing From Player Data: Find the Boring Bits
Compare how players really spend their time with what you planned, and look for long quiet stretches people leave after. Quiet alone is not a problem.
Story choice stats
Your AI counts what players really chose in a branching story and which endings they reached, and compares that with the story as you wrote it. It shows choices that are hardly choices because nearly everyone picks the same thing, endings nobody has seen, and picks that are not in your story at all.
Session pacing
Your AI shows the rhythm of a play session: how long people play, what they spend the time on, where nothing happens for a long while, and what they were doing when they stopped. It also measures how long it takes to reach a moment that matters, such as the first fun bit, and compares all of it with your plan.
How long does a chapter take a pair compared with a team of four?
A chapter planned for one length of time takes different groups very different times, and groups that change size halfway muddy any average. This entry shows how to compare group sizes fairly, which middle value to use, and when a size has too little data to speak for.
Do groups that start a chapter late at night finish it less often?
Finish rates by time of day are easy to get wrong, because the server clock is not the player's clock, and midnight breaks the buckets. This entry shows how to turn start times into local parts of the day correctly, including the groups across midnight and time zones.
A chapter planned for 20 minutes takes some players five and others an hour
The average time of a chapter looks fine while the experience is very uneven: some fly through, others get lost. The spread of times is the signal. This entry shows how to read it, and what it usually points to.
A chapter takes forty minutes, but how much of that is actual play?
Online chapters fill with waiting, watching and people stepping away. The length on the clock then overstates the play. This entry shows how to split the time into kinds, treat long pauses as someone being away, and report the share of real play.
Which story moments make the team talk, without reading what they write?
Chat is one of the few signs that a scene moved people, and you can use it without ever reading a message. This entry shows how to compare the number of messages after a scene with the number before it, why quiet scenes are not failures, and how to keep the data anonymous.
One player keeps everyone in a cutscene they all wanted to skip
If skipping a shared cutscene needs everyone to agree, a single person can make the whole team sit through it. The plain skip rate does not show this. This entry shows how to measure how many people wanted out versus how often the group actually got out.
When a timer picks for a silent group, the default option looks too popular
If a group does not answer in time and the game picks for them, that pick ends up in the statistics as if people had chosen it. This entry shows how big the distortion is and how to report popularity from real choices only, keeping the count of timed-out decisions visible.
Is the "decider" role in a rotating story really shared fairly?
When the right to decide moves around the group, equal numbers of turns are not fair if people were not all there for every decision. This entry shows how to work out how many turns each person should have had, and how far from that they ended up.