What replay analysis actually tells you, and what it doesn't
"Replay analysis" covers three fairly different things, and people mostly find that out after paying for the wrong one. Worth separating them, because each answers a different question and none of them answers all of it.
1. Box score stats
Goals, assists, saves, shots, score. Free, instant, and mostly a record of what happened rather than an explanation of why.
The trap is that these numbers measure activity. You can post a perfectly respectable line while losing, because nothing in a box score records the possession you gave away by challenging a ball you were never going to win. If your stats look fine and you keep losing, this is why.
Good for: confirming a pattern you already suspect. Bad for: finding one you do not.
2. Telemetry
The per-frame data pulled out of an actual replay file: boost over time, time spent in each third, supersonic percentage, distance from the ball, positioning heatmaps, percentile comparisons against your rank.
This is genuinely precise and it is the only layer that can tell you something like "your average boost is 27 and everyone at your rank sits at 41". When a leak is measurable, telemetry finds it faster than anything else.
Two real limits. First, it needs the .replay file, which consoles cannot export, so most of the playerbase cannot use it at all. Second, it tells you what without telling you why. A heatmap showing you are too far forward does not know whether that is because you challenge too early or because you are compensating for a passive teammate. Those need opposite fixes.
Good for: exact, measurable habits. Bad for: judgment and intent, and unavailable on console.
3. Watching the footage
A human, or an AI, actually looking at the game. This is the only layer that can say "at 0:45 you went for a 50 your teammate already had, and the back post was open both times you conceded".
It is less exact. It cannot give you percentiles and it should not pretend to. What it can do is name the decision, which is the thing you can actually change on your next kickoff. A number tells you that you are low on boost. Footage tells you that you keep leaving boost on the table because you take the same greedy line back every single time.
Good for: decisions, patterns, and the reasons behind them. Bad for: precision and any claim involving a percentile.
What none of them can do
Worth saying plainly, because plenty of marketing implies otherwise.
- Give you an exact MMR from one match. One game is a tiny sample. Any rank read from a single match is a band, not a number.
- Tell you what your teammates should have done in a way that helps you. It might be true. It is not actionable, and looking for it is the most common way people waste a review.
- Fix anything. Analysis produces a list. Ranking up comes from holding one change for twenty games, which is boring and is the actual work.
How to pick
If you are on PC, hardstuck, and want to know exactly how your numbers compare to your rank, telemetry is the sharpest tool available and there are free ones.
If you are on console, telemetry is not an option, so footage is the honest answer rather than a consolation prize. It also happens to be the layer that catches the decision-making leaks that decide most games below Champion.
And if your stats look fine but you keep losing, that gap is the whole reason these three layers exist. The numbers are describing a game you played well on paper. Something in the footage is describing the one you actually lost.