The pitch for live tracking is that you capture the game. You do not. One person with a phone captures the four things they decided to capture, and misses everything else completely, and that is fine as long as you chose the four things on purpose.
I built a courtside tracker because the alternative was a legal pad and a bad memory. It works, it runs offline, and it has changed what I can say at half time. It has also taught me exactly where its edges are, and the edges are more useful to know than the features.
What it genuinely captures
Things that happen and then vanish. The box score keeps shots, rebounds, turnovers. It does not keep how many times we let their five catch it inside the elbows, or how many possessions ended with the ball never touching the paint. Those are the numbers that change a half-time message, and they exist only if somebody counted them live.
The count you can hand to a player. "You're settling" is an opinion. "Nine of your last eleven were off one dribble or none" is a fact, and the second one moves a seventeen year old in a way the first never has.
Sequence. A tracker knows the order things happened in. Their run started after the second sub, not after the timeout. That is a rotation finding, and it is invisible in a final box score.
Evidence for the thing you already suspect. Most live tracking confirms rather than discovers, and that is a feature. A coach who walks in with a read and walks out with a count has something to say on Monday that survives an argument.
What it does not capture, and cannot
Anything you did not decide to track before tip. This is the whole game. You cannot retroactively count what you were not counting. Every tracker is a bet placed forty minutes before you know whether it paid.
Anything that needs two eyes. One person cannot watch the ball and the weak side at the same time. If your tracking plan needs off-ball detail, you need a second person or a camera, and pretending otherwise produces confident, wrong data.
Why. A tracker will tell you they scored eleven off the ball screen. It will not tell you whether that is a coverage failure, a personnel mismatch, or one guard having a bad night. That read is yours and it always will be.
Context the box score already has. If it is in the official box, do not track it. You will do it worse, slower, and with your eyes down. The whole value of a courtside tool is that it captures what the box misses.
The rule I actually use
Track four things, maximum, and pick them from the scout. Not from what would be interesting. From the two or three things the pre-game read says will decide the game, plus one thing that would tell you the read was wrong.
That last one matters more than the other three. A tracker that only counts evidence for your own plan is a machine for confirming yourself. One of your four counts should be the number that, if it comes back high, means you were wrong and you need to change something. Build the falsification in before tip, because you will not have the honesty to add it at half time.
The honest limitation of the whole category
Live tracking is the least reliable data collection there is. You are watching a game, thinking about the next substitution, and pressing buttons. The count drifts. It always drifts.
So do not treat a live count as a measurement. Treat it as a strong prior. It is precise enough to point you at the right two minutes of film, and it is not precise enough to publish. The workflow that works is track live, act at half time, verify off the tape on Monday. Anyone selling you live tracking as a substitute for film has never had to defend a number to a head coach.
The tool earns its place because at half time you have three minutes and no laptop. It does not earn a place in your season data. Those are two different jobs and the second one is not the phone's.