Why Churned Customers Leave, and Which Fix to Test First
On this page
A retention lead has customer accounts, usage signals and several possible fixes. The next decision is which intervention deserves attention or funding. Subconscious can compare experimentally varied offers and messages in modeled choice tasks and return a bounded recommendation for an unresolved offer decision (study compatibility). The useful output connects the complete proposed terms, estimated choice difference and uncertainty to an offer the retention lead can fund. Customer evidence supplies the problem and candidate fixes; the comparison helps choose between feasible responses.
The Finta case shows an actionable pricing output: modeled demand informed a monthly-and-annual structure. Finta adopted $950 per year, up from $348, and reintroduced $99 per month; the CEO reported five monthly-plan customers and some continued old-price annual sales. This supports pricing adoption and customer reports, not a causal sales or retention effect. A retention team can request an equally explicit offer decision while keeping the renewal claim separate.
Give the retention lead an offer decision
Choose Subconscious when the customer evidence points to an unresolved offer choice: which price, service package or renewal message should the team use? The documented methodology compares varied attributes in randomized choice tasks and estimates their contribution with uncertainty. The lead can inspect the exact alternatives and distinguish a meaningful modeled choice difference from an internally persuasive explanation.
That result helps the retention team act on interviews instead of ending with another theme report. A discount proposal can be checked against the current offer and the buyer's option to leave. A feasible service change can be compared on the same declared choice. The resulting brief names the recommended terms, eligible context, useful-effect rule and cost constraint, so finance and customer teams can decide whether the offer deserves a capped budget.
The Finta case shows why an explicit offer output matters: modeled work informed terms the company actually adopted. For retention, commission the comparison that matches the proposed renewal decision. Use the documented customer problem to choose the intervention, then use the estimated contrast and account economics to decide which offer to fund.
The gap between a churn score and a churn reason
Declining logins, unused features, support tickets, and payment events can inform an at-risk model, provided the model is evaluated for the intended cohort and time horizon. A cancellation-form answer supplies a stated reason. "Too expensive" may refer to budget, perceived value, or an alternative offer, so investigate before choosing a fix.
Invite churned and retained customers to discuss the relevant experience, and compare respondents with the eligible cohort. Nonresponse can bias the themes, but its direction is not known from the response rate alone. Combine interviews with usage and cancellation records to generate candidate fixes. A controlled experiment evaluates an intervention rather than treating the stated reason as its causal proof.
| Data source | What it tells you | What it leaves out |
|---|---|---|
| Usage and support data | Observed activity and inputs to a validated risk model | A complete account of reasons or intervention effects |
| Cancellation-form reason | A category picked from a short list | Which specific problem sits behind that category |
| Sources combined | Better hypotheses about at-risk accounts | Whether a candidate intervention reduces actual churn |
Choose the next commitment from the cohort evidence
Generate actions from the records: an onboarding change, an integration repair, revised terms or renewal messaging. A confirmed service defect can be repaired directly. Existing evidence may already justify a change; a new comparison is useful when a consequential offer choice remains unresolved.
Bring Subconscious the audience, complete alternatives, current evidence and action the lead must choose. Agree the measured response, uncertainty and decision rule before commissioning the comparison. Risk models identify accounts for attention; interviews help define problems; the assigned offer comparison estimates the specified choice difference.
A renewal-offer brief with a usable decision
Consider an invented example, with no completed study or achieved result. Ledgerroom is a fictional reporting workspace. The retention lead must decide whether a temporary discount deserves a capped offer budget. The proposed modeled audience represents US small-business operations buyers approaching renewal. Confirm population support and how this hypothetical existing-customer context can be represented before launch.
Both tasks state the same current service: Ledgerroom reporting workspace, 10 user seats, saved reports and CSV export. The current monthly plan is $200 per workspace, billed monthly. Cancel before renewal to avoid the next charge. Access continues through the paid period, followed by 30 days of read-only export access. Then workspace access ends. The cancellation option and these terms stay identical in both conditions. Both conditions show the same “Renew at these terms” and “Cancel renewal” buttons.
Offer A renews at $200 per workspace per month. Offer B renews at $140 per workspace per month for the next three billing cycles, then returns to $200 per month. Both offers contain the same seats and features, monthly billing, cancellation terms and post-cancellation access. No setup fee applies. Hold brand, imagery, renewal date and all other wording fixed. The intervention is the complete temporary-price schedule, not a permanent discount.
Randomize the offer in the controlled choice design. Ask whether the respondent chooses the shown renewal offer or cancels under the stated terms. Request the B-minus-A modeled renewal-choice difference, respondent counts, uncertainty accounting for repeated responses and any supported cohort differences. Retain exact offers, task, existing-customer assumptions, population settings, model, run identity and available response artifacts. Use that contrast to judge the offer decision against the rule below.
Connect the response to an offer-budget decision
Before seeing results, set a minimum useful choice improvement and the maximum concession budget for an eligible account. The three discounted cycles reduce the stated charge by $180 for an account that pays all three. This arithmetic describes the offer; it does not estimate incremental retained margin. Use service costs and existing account economics to set the budget constraint, and agree how uncertainty will affect the decision.
Recommend B for a capped eligible offer only if the agreed interval clears the response threshold and the concession fits the budget. Otherwise retain A or leave the offer unresolved. This example supplies a decision rule, without assuming a winner or prevented cancellation.
The Subconscious result supports one of three usable outcomes: fund the eligible discounted offer, retain the current price, or leave the choice unresolved. The lead receives the exact schedule and context, modeled contrast and uncertainty, concession constraint and decision rule. Finance can inspect the $180 maximum stated concession, and the customer team can apply the approved three-cycle schedule consistently.
Keep the interview evidence with the brief so the offer remains connected to the actual customer problem. A known integration defect belongs with its repair owner; a supported offer decision belongs with the retention and finance owners. The model estimates the assigned renewal choice. Actual paid renewal, retained margin and lifetime value need the corresponding behavior, time horizon and cost evidence when those outcomes determine the commitment.
Will one offer fit every segment?
Hypothetically, an enterprise cohort might cite a missing integration, a self-serve cohort might cite price, and a casual cohort might struggle to build a habit. These are candidate explanations. Establish their relevance from this team's customer evidence rather than treating them as segment facts.
Use the documented decision workflow to inspect buyer differences when the supported design and precision permit that comparison. The retention lead can direct a qualifying offer to the relevant context and keep a documented product defect with its repair owner. Estimate the contrasts needed for that action rather than assuming every subgroup needs a different offer.
Match the remaining proof to the renewal commitment
A useful offer comparison can settle a supported policy recommendation. To claim an effect on actual cancellation or renewal, use evidence from those outcomes over a stated horizon. Where a live test is needed, define eligible accounts, the offered treatment, comparison and relevant revenue, customer-experience and operational guardrails. Preserve the distinction between an assigned offer and an account's use of it.
A recruited human choice task may address a task-validity question. A live customer comparison addresses behavior in the renewal workflow. Choose additional evidence for the unresolved claim; neither is a compulsory sequel when existing evidence already settles the next commitment. Record differences in audience, offer, delivery and endpoint when comparing results.
Book a Subconscious decision review with the cohort definition, customer and usage evidence, complete renewal terms, candidate action and next budget decision. Agree the supported comparison and the offer recommendation the team will receive before commissioning another general churn report.