Story and pacing
Players never take the second research branch: is it a bad branch or never seen?
A branch picked by only a few players is not necessarily a bad one. Maybe most of them could not even afford it when they chose. This entry shows how to count only the players who really faced the decision, and what to do with the undecided.
Which step in the tech tree takes far longer than the others to reach?
A research tier that takes much longer than its neighbours is where the game slows down or players give up. This entry shows how to measure the time between tiers on one consistent group of players so the numbers do not shift between steps.
Which answers in a countdown decision are picked only in the last seconds?
When a crisis gives you a few seconds, a lot of people just grab something. An option that is mostly chosen at the very end may be confusing rather than popular. This entry shows how to tell hurried picks from considered ones, and what to ignore.
Which units or buildings does nobody use, even though they were available?
A unit that few people build may simply be unlocked too late for most players to ever see it. This entry shows how to count only people who had it available for long enough, so you can tell a dead option from a hidden one.
Which of the ways to win is never used, and which one can hardly be reached?
If a game offers several ways to win, one often turns out to be unreachable in practice. Looking at who won with it is not enough, because nobody wins with a path that nobody gets close to. This entry shows how to count progress as well as wins.
Which attack wave suddenly kills far more bases than the one before?
Counting deaths per wave makes late waves look easy simply because few players get there. This entry shows how to measure the risk among those who actually faced the wave, and how to tell a real jump in danger from a change in who is left.
One teammate is connected but the story went on without them: is it a stuck state?
After a reconnect a player can be in the game and yet not part of the scene the others are in, seeing nothing happen. From the server's list they look fine. This entry shows how to recognise such stuck players from progress times and how to avoid raising false alarms.
Do teammates really pick the same story choices, or does it only look like it?
When each player in a team makes their own story choice, a high share of matching picks can simply mean one option is popular. This entry shows how to measure whether friends really influence each other, and what the number means when everyone picks the same thing.
Which lines of dialogue do players click through without reading?
A line that most players dismiss in under a second is one they did not read, however well written. Comparing time on screen with the time reading would need shows which lines to cut, shorten or move.
How much of a chapter do the players spend far away from each other?
A story scene that needs the whole team is wasted if everyone is wandering in a different corner. This entry shows how to turn regular position snapshots into the share of time a group is together, how long the longest stretch apart lasts, and who counts as apart in a group of three.
How much of an hour with the game is spent in cutscenes nobody can skip?
Cutscene time looks small per chapter and large per evening. Averaging the rates of long and short sessions also gives a wrong picture. This entry shows how to measure the forced watching per hour of play in a way that one short session cannot distort.
Is the timer on group story votes too short, too long, or fine?
A countdown on a team decision either rushes people into a random pick or lets the group wait for nothing. The average time to decide cannot tell you which, because the rushed picks are exactly the ones that were cut off. This entry shows what to look at instead.