Ecommerce & Retail · Growth problem
Low repeat purchase and retention
Retention problems are frequently obscured by CRM reporting that credits email and SMS with revenue that would have happened regardless, leaving leadership with an inflated sense of how well repeat purchase is actually performing. A proper view requires cohort-level repeat rate and time-to-second-order data, read against the natural purchase frequency of the category, before any conclusion about retention health is reliable.
Symptoms
What this usually looks like
- Repeat purchase rate is flat or declining across recent customer cohorts
- Time-to-second-order is lengthening
- CRM-attributed revenue looks strong but overall repeat rate isn't improving
- A large share of revenue depends on continually acquiring new customers
- Customer lifetime value estimates vary significantly depending on who in the business is asked
Diagnostic questions
What we would test first
- Build repeat purchase rate by monthly acquisition cohort over the last 12–18 months
- Calculate time-to-second-order distribution and compare against category norms
- Run or review a CRM holdout/suppression test to estimate incremental revenue
- Segment repeat rate by acquisition channel, isolating discount/promo-acquired customers
- Establish natural category purchase frequency using top-decile loyal customer data
- Recalculate 12-month LTV by cohort using an agreed, conservative methodology
Root causes
Why it happens
- 01
Category purchase frequency mismatched with retention expectations
Some categories are naturally infrequent purchases — durable goods, occasion-led categories — and applying a replenishment-style retention benchmark to them will always look disappointing. The right benchmark is the category's natural repeat cycle, not a generic retention target.
- 02
CRM attribution inflating perceived retention performance
Post-purchase and browse-abandonment flows are commonly credited with revenue from customers who intended to purchase anyway, which can make retention performance look considerably healthier in CRM reporting than the underlying repeat-purchase cohort data shows.
- 03
Post-purchase experience not building towards a second order
Delivery experience, product quality perception, and how (or whether) the business follows up after the first order all influence whether a customer considers a second purchase — gaps here often show up as a retention problem long before they're recognised as a post-purchase experience problem.
- 04
First-order customer quality varying by acquisition channel
Customers acquired through deep-discount channels or one-off promotional codes frequently have structurally lower repeat rates than those acquired through organic or referral channels, meaning a retention problem can actually be a downstream effect of acquisition channel mix.
- 05
Lack of segmentation in retention activity
Treating all customers with the same generic post-purchase journey, regardless of category purchased, order value, or first-order experience, tends to under-perform against a segmented approach built around actual repeat-purchase behaviour.
Evidence
The numbers we would look at
These are the metrics that make the constraint visible, and the cuts that stop them being reassuring by accident.
| Metric | What it tells you | How we read it |
|---|---|---|
| Repeat purchase rate by cohort | The core measure of retention health, tracked by acquisition month or quarter to reveal trend rather than a single snapshot.Compare across at least 4–6 cohorts to distinguish a trend from noise. | Compare across at least 4–6 cohorts to distinguish a trend from noise. |
| Time-to-second-order | Shows how quickly customers return, and whether that window is lengthening.Segment by first-order category and acquisition channel. | Segment by first-order category and acquisition channel. |
| Category purchase frequency (natural repeat cycle) | The benchmark against which repeat rate should actually be judged.Derive from historical data on genuinely loyal customers within the category, not an industry-wide assumption. | Derive from historical data on genuinely loyal customers within the category, not an industry-wide assumption. |
| CRM-attributed vs incremental repeat revenue | Distinguishes reported CRM performance from the revenue CRM is actually creating.Consider a holdout or suppressed-send test to estimate incrementality. | Consider a holdout or suppressed-send test to estimate incrementality. |
| Repeat rate by acquisition channel/cohort | Reveals whether retention issues stem from customer quality at acquisition rather than post-purchase experience.Segment discount-code and promo-acquired customers separately. | Segment discount-code and promo-acquired customers separately. |
| Customer lifetime value (12-month, by cohort) | Connects retention performance directly back to acquisition economics.Use a consistent, agreed methodology — LTV estimates vary widely depending on assumptions used. | Use a consistent, agreed methodology — LTV estimates vary widely depending on assumptions used. |
Measurement traps
What can mislead you
- Looks fineCRM revenue is a growing share of total revenue, so retention is improving
- A growing CRM revenue share can simply reflect more emails being sent, or better attribution capture, rather than genuinely improved repeat-purchase behaviour. An incrementality check against a holdout group gives a more reliable read.
- Looks fineOur repeat rate looks low, but it's a considered-purchase category
- This can be a genuine and valid explanation — but it needs to be checked against the category's actual natural repeat cycle using the business's own top-cohort data, not assumed without evidence.
- Looks fineLTV is healthy, so acquisition spend is justified
- LTV figures are highly sensitive to the time window and margin assumptions used; an LTV calculated over 24 months with optimistic repeat-rate assumptions can justify acquisition spend that a more conservative 12-month, evidenced figure would not.
Outcome
What better looks like
Not a promised number. A clearer basis for the next investment decision.
- Repeat purchase rate is tracked by cohort and benchmarked against the category's actual natural frequency
- CRM's contribution is understood as incremental revenue, not simply attributed revenue
- Retention activity is segmented by customer and category, rather than run as one generic post-purchase journey
- LTV and acquisition spend decisions are connected through an agreed, evidenced methodology rather than optimistic assumptions
Where a Growth Diagnostic would start
A three to four week senior review across demand, discovery, acquisition, conversion, retention, measurement and capability — sequenced so this problem is either confirmed as the constraint or ruled out early. Read alongside the ecommerce & retail model page for how we frame the wider system.
Related problems
Ecommerce & Retail
Rising customer acquisition costs
The cost of winning a new customer keeps climbing, even where individual platforms report healthy performance.
Ecommerce & Retail
Marketing attribution is unclear
Every platform claims the credit and the numbers don't add up. A pragmatic approach to measurement that decisions can actually be based on.
Consumer & Omnichannel
Loyalty and retention don't hold together across channels
Loyalty and CRM activity often only reaches a partial, digitally-biased view of the customer base, weakening retention.
Questions about this problem
- Do we need a full CRM platform migration to fix this?
- Rarely as a first step — most retention problems are diagnostic and strategic before they're a platform limitation. It's worth establishing what's actually driving low repeat rate before assuming the CRM technology itself is the constraint.
- How far back does cohort data need to go to be useful?
- At least 12 months, ideally 18–24, to capture enough cohorts to distinguish a genuine trend from normal month-to-month variation, particularly for categories with a longer natural repeat cycle.
- Can this be assessed without running a CRM holdout test?
- A holdout test gives the clearest incrementality read, but a reasonable estimate can also be built from comparing CRM-attributed revenue trends against organic repeat-purchase trends over time, if a live test isn't feasible immediately.