Acquisition or retention: comparing the next euro, not the averages.
Evidence status
What was reproduced, and what it does not prove.
The educational CSV and the Python run verify the six incremental contributions, net values and incremental-share bounds. They verify neither a caused retention, nor an additional acquisition, nor a market return.
- Technical specification of the acquisition–retention frontier, specification in French (opens in a new tab)
v1, 31 August 2026 · CC BY 4.0 · specification in French · Method document written by marketing-mix.net. The figures are illustrative: they are not market measurements.
Key distinctions
Three conditions before calculating.
Same economic unit
Acquisition and retention are converted into discounted incremental contribution after cost to serve, discounts, cannibalisation and risk. Customer lifetime revenue is not compared with an annual spend without aligning horizons and margins.
Genuinely avoidable churn
Not every customer retained after an action was retained thanks to it. Estimate a counterfactual through a test, a staggered rollout or a credible model. Targeting only at-risk customers can also confuse selection with effect.
Stock and capacity constraints
Effective retention does not indefinitely replace the entry of new customers; massive acquisition can saturate onboarding and support. Growth minimums and service maximums frame the frontier.
Method
Building a portfolio of comparable tranches
- 01Break down the cohorts
Segment by origin, tenure, margin, churn probability and capacity; compute the reference contribution and the horizon during which the intervention can act.
- 02Estimate the marginal responses
For each acquisition or retention tranche, estimate incremental customers, additional retention, contribution and interval; correct for saturation and overlaps.
- 03Add constraints and risk
Define onboarding capacity, contact volume, commercial pressure, segment mix and learning needs; penalise the tranches whose evidence is weak or highly correlated.
- 04Fund the frontier
Rank the tranches by prudent contribution per euro and by capacity; fund up to the threshold, then test sensitivity to a rise in churn, in CAC or in cost to serve.
The acquisition–retention frontier
The tranches are positioned by prudent contribution and capacity consumed in order to build an efficient portfolio.
€200k moved to retention adds €96k of contribution
The plan devotes €700k to acquisition and €300k to retention. The last acquisition tranche of €100k creates 260 customers at €310 of discounted contribution, that is €80.6k. Two retention tranches of €100k retain 420 and 300 customers respectively at €280 of contribution.
With incremental effects of 65% and 55%, the retention contributions are €76.4k and €46.2k before cost, that is −€23.6k and −€53.8k net: they do not justify the move. A retargeting overhaul reduces each tranche to €45k without changing the effect, giving €31.4k and €1.2k net. Compared with two saturated acquisition tranches at −€32k each, the reallocation improves the plan by €96.6k.
Only one retention tranche is funded in full; the second becomes a pilot. The arbitration rests on net marginal returns, not on the general idea that retaining costs less.
Acceptance conditions
What must be true to act.
- 01
Acquisition and retention share horizon, margin and incremental unit.
- 02
Avoidable churn is separated from observed retention.
- 03
Capacity, minimum growth and customer pressure frame the allocation.
- 04
Tranches are funded according to their prudent bound, with a learning budget.
Limits
What this analysis does not prove.
- Customer value is sensitive to margin and duration assumptions.
- Relational effects spread between cohorts and complicate the counterfactual.
- A quarterly optimisation can underinvest in structural capabilities.
References
Works cited.
- Gupta, S. & Lehmann, D. R. (2005), Managing Customers as Investments.
- Reinartz, W. & Kumar, V. (2000), “On the Profitability of Long-Life Customers”, Journal of Marketing.
- Fader, P. S. & Hardie, B. G. S. (2009), “Probability Models for Customer-Base Analysis”, Journal of Interactive Marketing.