Experiments

One randomised comparison per leakage category. A negative gap is left on screen rather than hidden — an intervention that does not work is a finding, not a bug.

Payment Failure

Payment Failure

p = 0.741·n = 45
Treatment
n=37
32.4%
+₹1,48,350+32.4 pts incrementalcollecting
Holdout
n=8
0.0%
Treatment
₹1,48,350
12/37 recovered
Holdout
₹0
0/8 recovered
95% CI on gap
-2.4 to 48.5 pts
still includes zero
Checkout Abandonment

Checkout Abandonment

p = 1.000·n = 22
Treatment
n=20
35.0%
+₹8,067+35.0 pts incrementalcollecting
Holdout
n=2
0.0%
Treatment
₹8,067
7/20 recovered
Holdout
₹0
0/2 recovered
95% CI on gap
-14.7 to 56.2 pts
still includes zero
B2b Receivable

B2b Receivable

p = 0.955·n = 23
Treatment
n=19
31.6%
+₹5,441+31.6 pts incrementalcollecting
Holdout
n=4
0.0%
Treatment
₹5,441
6/19 recovered
Holdout
₹0
0/4 recovered
95% CI on gap
-8.1 to 52.2 pts
still includes zero
Subscription Failure

Subscription Failure

p = 1.000·n = 23
Treatment
n=16
18.8%
−₹2,590-9.8 pts incrementalcollecting
Holdout
n=7
28.6%
Treatment
₹4,211
3/16 recovered
Holdout
₹1,621
2/7 recovered
95% CI on gap
-47.4 to 21.8 pts
still includes zero