Retention investment decisions

Choose a retention investment by comparing the extra contribution it could create with its full cost, the strength of the evidence and the team's ability to deliver it. A large pool of inactive customers is not automatically the best opportunity. First identify the customer problem each proposal would change.

Put competing proposals on one decision sheet

A shortlist might include repairing a recurring delivery failure, improving a reorder route, testing a reminder and expanding acquisition. Describe each option in the same terms: eligible customers, action, usual alternative, outcome, time horizon, cost, owner and earliest credible review date. Include continuing current practice as an option.

Keep obligations separate from discretionary investment. The investment question concerns additional work, such as preventing a problem from recurring.

Compare added contribution, not attributed revenue

Compare each option on the added contribution it could create relative to usual practice, not on attributed revenue or historical customer value. Use one consistent basis for discounts, refunds, variable costs, tax and time horizon across the shortlist. The usual-practice outcome is the difficult benchmark; where it cannot be observed, present a range of assumptions rather than a claimed return.

Keep the comparison at portfolio level: the purpose is to rank proposals and decide how much to commit, not to calculate a precise repeat-order contribution. Flag where costs, timing or data coverage are uncertain enough to change the ranking.

Check the cost and market views

Set a consistent cost basis before comparing proposals. A recorded product cost is not necessarily an all-in cost, so include fulfilment, payment, contact and staff costs where they differ by option. Keep any excluded taxes or charges visible in the comparison.

For any market view, check that customer charges and the corresponding costs are treated on the same basis. A market profitability ranking can show where economics differ, but it is not evidence that a proposed retention action will change purchasing. Use it to frame the comparison, not to justify the investment.

Check cost coverage before ranking proposals. If cost data is missing for products or variants in scope, the portfolio comparison may be incomplete, and the ranking should carry that uncertainty.

Match evidence to the commitment

Separate an observed problem from the expected effect of a fix. Complaints followed by inactivity can identify a reason to investigate, but that association does not show how many orders a repair would restore. An early estimate may justify a limited, reversible change. A costly rollout needs a stronger case and a way to check whether the underlying issue improves.

For a marketing action, a comparison group receiving usual treatment can help estimate the added effect. For a change applied to a warehouse or service team, compare suitable rollout groups and periods where feasible. Check previous trends and other changes before attributing a later difference to the repair. If a credible comparison is unavailable, report the operational result and retention trend separately.

A consumable buyer may have a near-term reorder opportunity; a durable-product buyer may not. Use a horizon that fits each proposal, then make differences in timing and uncertainty visible when allocating cash.

Interpret association before approving investment

Treat an observed relationship as a candidate problem or segment, not as proof that a proposed action will change customer behaviour. Use it to guide investigation, then require evidence that addresses the proposed action and outcome before funding a costly rollout.

When the evidence is only associational, limit the commitment: prefer a bounded, reversible test, define the comparison in advance, and record what result would justify extending it. If the pattern cannot be linked to a specific action the team can deliver, defer the investment or fund diagnosis first.

Decide, stage and revisit

Approve a proposal with a spending limit, accountable owner, implementation check and review rule. The decision may be to fund, run a bounded pilot, repair the data or process first, or defer. Record the assumption that would change it. If two proposals need the same staff or customer group, account for that constraint rather than treating both forecasts as independently achievable.

At review, check whether the change was delivered. Then compare the intended customer and commercial outcomes with the agreed alternative, including customers who did not buy. Reconcile costs before extending the work. An uncertain result may support another targeted test.

The supporting decisions need separate workings: comparing acquisition expansion with retention improvement, estimating the cost of one unresolved churn driver, deciding whether a segment recovery action is economical, and reviewing results after an operational change.

Retention Investment Decision Process