KOLLIEDVenture Builder

Experiments

Experiment design with pre-committed kill-criteria, result vs target judged by the engine, and learning routed to build / Go-Kill.

T2T2
Running
6
Pass rate
44%
Avg clock
18d
Kill-criteria set
83%

Design the experiment

T2

Set hypothesis, success metric, cost, clock and the pre-committed kill-criteria.

Dogfood-demo: Clearly-labelled dogfood demo: the 6 companies (MyStdy · Servicely · DifferenZ · Exponic · Digna · Kollied) are companies Espen holds a stake in, used ONLY as demo data to exercise the engines — never as proof that venture building works. Every demo row is flagged dogfood and held out of every client-facing aggregate (DEL 98-3).

Experiment design — pre-committed kill criterion

Hypothesis · success metric · cost · clock · kill criterion set BEFORE the result (not post-rationalised).

Onboarding < 1 day → activation
target: ≥ 40% activation in 30 days · kill criterion (pre-committed): < 25% activation in 30 days → kill
Paid pilot on a cold list
target: ≥ 3 paid pilots in 45 days · kill criterion (pre-committed): 0 paid pilots in 45 days → kill (pre-committed)
Price elasticity, high segment
target: ≥ 20% acceptance at +25% price · kill criterion (pre-committed): < 10% acceptance → drop segment

Results — judged via the engine

Result vs target → verdict + route. Only a miss that meets the kill criterion flips the hypothesis.

Onboarding < 1 day → activationmet → build
result: 44% activation · result cleared the pre-committed kill-criteria → met; route to build
Paid pilot on a cold listmissed → kill · hypothesis invalidated
result: 0 paid pilots · result met the PRE-COMMITTED kill-criteria → missed; route to kill (flips the linked hypothesis to invalidated)
Price elasticity, high segmentinconclusive → iterate
result: no reading · no decisive result observed — inconclusive; iterate the experiment

Learning → routing

Capture the learning and route to build / Go-Kill.

Fast onboarding drives activation — build on it.
Cold list does not convert — the kill criterion triggered; hypothesis invalidated.
No decisive reading yet — iterate.
Live commit · demo store