AI-assisted preparation guide. All organisations, figures and incidents in the exercises are fictional. These are original practice tasks, not leaked employer questions, hiring-policy statements or accounts of candidate outcomes. No employer affiliation or human specialist review is claimed.
Define retention before calculating it
For this fictional UK subscription-service interview exercise, month-one retention means accounts from a starting cohort still active at the chosen checkpoint. The definition is only for this task. State the cohort-entry date, account unit and observation window before discussing a chart. Different definitions can produce different figures from the same underlying activity.
Calculate within each cohort
January begins with 100 accounts and has 60 active at its checkpoint. February begins with 20 and has 15 active at its comparable checkpoint. January retains more accounts in absolute terms, but its rate is 60% against February’s 75%. Those percentages do not explain why behaviour differs or prove that a product change caused it.
Investigate comparability and uncertainty
Ask whether the cohorts reached the same maturity and came from similar acquisition routes. Check exclusions and missing account states. A smaller cohort can change noticeably when a few accounts behave differently. Recommend a follow-up based on the decision being made instead of announcing a successful intervention from two percentages alone.
Practice question and worked response
Which cohort retained users better: January with 60 remaining out of 100, or February with 15 out of 20?
A possible response to adapt
On the exercise definition, February has the higher rate at 75%, while January has more retained accounts and a 60% rate. I would show both the numerator and denominator, check comparable observation windows and investigate acquisition or account differences. I would not credit a product change without evidence that distinguishes it from other explanations.
This is an illustrative response, not a personal achievement to claim. For an experience question, replace it with something you actually did.
Build your evidence sheet
| Evidence | What to prepare |
|---|---|
| Definition | Entry event, unit and retention checkpoint. |
| Calculation | Retained accounts divided by the starting cohort. |
| Caveat | Maturity, acquisition differences and limited evidence of cause. |
Check your answer
Explain both counts and rates without switching denominators. Make the time window explicit. If you suggest further analysis, say which uncertainty it addresses rather than promising that more data automatically proves causation.
Rehearse the next attempt
Record a first answer in your own words. Use our free mock interview scorecard to note one strength and one specific correction. Use the STAR worksheet when the question asks about your previous experience.
For optional typed preparation, give Cluegent a redacted practice brief and your draft. Ask it to challenge your assumptions rather than invent achievements. Explore Cluegent for Windows or macOS and check the current trial and paid terms. The worksheets are free; app access is separate. Follow assessment rules about outside assistance. Obtain required consent before recording or transcription; never upload private customer, employee or applicant information.
Where Cluegent helps
Cluegent supports permitted live workflows with transcript context, typed prompts, screenshot-aware answers, resume context, custom response behavior, quick action buttons, and a private desktop overlay. It is most useful when you already understand the subject and need help staying structured under pressure.
Frequently asked questions
Is a higher retention percentage proof of a better feature?
No. Check comparable definitions and other differences before making a causal claim.
Can I use SQL to explain the calculation?
Yes. Define cohort membership and checkpoints first so the query answers the intended question.