Skip to content
Back to the site

Method demonstration

Proof, on data where the answer is known.

Hand a churn model real data and it can score well without being right — real data has no answer key. So we test ours on a simulated game where we planted the truth, and run the method blind to it.

What it recovers, every time:

  • the real driver of churn — accepted;
  • a convincing decoy — rejected;
  • the sampling trap that flatters an ordinary model — refused.

What it recovered, blind to the truth

Accepts the real driver, rejects the decoy
Illustrative synthetic data — demonstration only
0.0 0.5 1.0 1.5 2.0 2.5 the real driver accepted a decoy rejected accept threshold effect size

The real driver clears the accept threshold; a convincing decoy with no real effect does not.

The recovered exit shape
Illustrative synthetic data — demonstration only
0 1 2 3 how abruptly players leave → recovered planted

Read blind, the recovered shape of exit tracks the shape we planted.

And it holds under the tools a quant team expects by name — time-varying Cox, competing-risks, and a survival forest scored on held-out players.

Book a call