Why we price every leak in bb/100, not a population percentage
Most trackers show you a leak as a gap: your stat versus a target range, measured in percentage points. Fold to 3-bet is 68% against a "should be" of 60%, so you're told you're 8 points too tight. That number is easy to compute and easy to display. It is also close to useless on its own, because it doesn't tell you the thing you actually want to know: how much is this costing you?
The same-sized gap isn't the same-sized leak
Two leaks can have identical percentage-point gaps and be worth wildly different amounts, for one simple reason: how often the spot actually comes up, and how expensive it is on each branch when it does.
Take a real pair from a PokerForge database: a 32.9-percentage-point gap in "UTG open vs UTG/MP 3-bet" against a 17.3-point gap in "SB facing UTG/MP open." The first one looks almost twice as bad. Priced properly — using the player's own actual profit when they continue versus when they fold, in that exact matchup — the second leak turns out to be worth roughly 13.5 times more, because it comes up nearly forty times more often. Ranking by raw deviation alone would tell you to go fix the wrong thing first.
bb/100 is the unit that already accounts for frequency
bb/100 folds sample size and stakes into the same number by construction — it's a rate, not a snapshot. A leak priced in bb/100 from your own continue-vs-fold results already answers "how much does this cost me, given how often it actually happens," which a percentage-point gap never does on its own. Two different-looking gaps become directly comparable the moment you price them the same way.
What this changes in practice
PokerForge's Leaks by Matchup engine prices every preflop leak this way: from your own real results, in the exact opponent-position matchup it happens in, then ranks the list by what continuing instead of folding (or vice versa) is actually worth. The biggest-looking gap on the page isn't necessarily first — the leak that's actually bleeding the most bb/100 is.
None of this replaces raw stats entirely — VPIP, PFR and the rest are still there, and still useful for getting a read on a range. But when the question is "which of these should I fix first," a number priced in bb/100 from your own results is a much better answer than a percentage gap ever was.