Reference table
Six rules for choosing controls in a churn study
Six rules for choosing the comparison group in a churn or win-loss study, each with why it matters and what it looks like in product research. They cover the source population, matching on tenure and company size, interviewing in the same weeks, using the same questions in both arms, fixing the sampling rule in advance and analysing broken matches separately, as Doll and Hill did in 1950.
Where this appears
What's in this reference table
| Control selection rule | Why it matters | In product research |
|---|---|---|
| Same source population | Controls must come from the population that would have become cases | Same plan tier, acquisition channel and region, not whoever answers |
| Matched on known confounders | Removes variables you already know drive the outcome | Match on tenure, company size, contract value, onboarding cohort |
| Same time window | Removes period effects, releases, pricing changes | Interview stayers in the same weeks as leavers, not months later |
| Identical instrument | Differences must come from respondents, not from the protocol | The same questions, asked the same way, in both arms |
| Fixed selection procedure | Removes the researcher discretion that creates bias | Define the sampling rule before you look at the list |
| Documented relaxations | Broken matching shouldn't silently contaminate the pooled result | Analyse relaxed-match cases separately, as Doll and Hill did |
Use this image
Free to use in your docs, decks and posts under CC BY 4.0, as long as you credit Koji with a link to this page. To embed it, paste this into your page:
html
<a href="https://www.koji.so/visuals/case-control-churn-study-control-selection-rules"><img src="https://www.koji.so/visuals/case-control-churn-study-control-selection-rules/case-control-churn-study-control-selection-rules.png" alt="Table of six rules for choosing a control group in churn research, from same source population and matching on tenure to an identical interview instrument." width="1200" height="902" /></a>
<p>Source: <a href="https://www.koji.so/visuals/case-control-churn-study-control-selection-rules">Six rules for choosing controls in a churn study</a>, Koji</p>
