eCommerce brands do not have subscription churn. They have something mathematically equivalent and operationally harder to detect: purchase-rate decay. A store that loses 20% of its second-purchase rate annually does not feel the loss in month one. It feels it in year three, when the retained customer base has shrunk to half its previous size and acquisition spend is running at double the rate just to maintain flat revenue. The mechanism is identical to SaaS churn. The failure mode is identical. The early warning system almost no operator has built.
The compounding nature of this decay is what makes it dangerous. A single-digit annual decline in repeat purchase rate does not register as a crisis. It registers as seasonality, or a bad quarter, or a product assortment issue. By the time the pattern is unmistakable, the cohort economics have already deteriorated to a point where recovery requires either a fundamental product change or a sustained period of growth below replacement rate.
The Default Assumption (and Why It Fails)
The standard framing in eCommerce analytics treats customer retention as a binary: a customer either comes back or they do not. Retention rate — the percentage of customers who make at least a second purchase — is the metric most operators track. Klaviyo puts the industry median for DTC repeat purchase rates at 27–34% across non-subscription categories (Klaviyo DTC Benchmarks, 2024).
This framing fails because it measures the event of repurchase, not the rate of repurchase decay over time. A customer who made four purchases in year one and two purchases in year two is still counted as "retained." The cohort analytics tool shows a green check. The dashboard shows no churn signal. But that customer's annual purchase frequency has dropped 50%, and if the trend continues for another year, they will be functionally inactive — a lapsed customer the system still classifies as retained because they technically purchased at some point in the trailing 12 months.
The relevant metric is not whether a customer came back. It is the rate at which their purchase frequency is declining across the cohort, and whether that rate is accelerating.
What the Decision Actually Hinges On
The Decay Rate Within Active Cohorts
Purchase-rate decay is not uniform. It concentrates in the first 12–24 months after acquisition. A customer who makes a second purchase within 60 days of their first is significantly more likely to make a third than one whose second purchase arrives at 120 days. Shopify's merchant data shows that customers who repurchase within 30 days have a 53% chance of making a third purchase; those who repurchase at 90+ days drop to a 27% third-purchase probability (Shopify Merchant Insights, 2024).
This means the decay curve is steepest at the beginning. Brands that measure cohort-level purchase frequency in quarterly buckets miss the early decay signal entirely, because the rate of decline is sharpest in the window that quarterly reporting smooths over. Monthly cohort tracking, segmented by time-to-second-purchase, is the minimum resolution required to see this pattern before it compounds.
The Reactivation Cost That Compounds Alongside Decay
As purchase frequency declines within a cohort, the cost to reactivate those customers increases. A customer who has not purchased in 90 days responds to a standard email win-back at a rate roughly 3–5× higher than one who has been inactive for 180 days. Email reactivation sequences for 90-day lapsed customers typically convert at 8–12%; for 180-day lapsed customers, that rate drops to 2–5%.
The implication is that reactivation — the standard tactical response to identified churn — becomes less effective precisely as the churn compounds. A brand that catches decay at 90 days can recover 1 in 10 lapsing customers at low cost. A brand that identifies the same decay at 180 days recovers fewer customers at higher cost per reactivation. The window in which intervention is economically rational is narrow, and it closes faster than most monthly reporting cycles.
The Third Variable: What Repeat Purchase Rate Decay Does to Acquisition Economics
This is the one nobody models explicitly. When purchase-rate decay compounds across your retained base, you need to acquire more new customers to maintain the same revenue. But acquisition cost does not stay constant when you scale volume — in most paid channels, CAC rises 15–30% when you increase spend by 50% or more, because you exhaust the highest-intent audiences first and start reaching lower-probability buyers.
The result is a squeeze that tightens from both sides: the retained base generates less revenue per customer per year (decay effect), and replacing that revenue through acquisition costs more per dollar recovered (scale effect). This is the mechanism by which a brand with a steadily declining repeat purchase rate ends up spending more on acquisition every quarter to maintain flat top-line numbers, until the unit economics no longer support the spend.
The Cost Reality
The following table shows how a starting cohort of 1,000 customers decays over three years at four different annual purchase-rate decay rates. "Purchase rate" here means the average number of purchases per customer per year within the retained cohort.
| Year | 0% Decay (Baseline) | 10% Annual Decay | 20% Annual Decay | 30% Annual Decay | |---|---|---|---|---| | Starting purchase rate | 2.40 | 2.40 | 2.40 | 2.40 | | Year 1 rate | 2.40 | 2.16 | 1.92 | 1.68 | | Year 1 total purchases (1,000 customers) | 2,400 | 2,160 | 1,920 | 1,680 | | Year 2 rate | 2.40 | 1.94 | 1.54 | 1.18 | | Year 2 total purchases | 2,400 | 1,944 | 1,536 | 1,176 | | Year 3 rate | 2.40 | 1.75 | 1.23 | 0.82 | | Year 3 total purchases | 2,400 | 1,750 | 1,229 | 824 | | 3-Year cumulative purchases | 7,200 | 5,854 | 4,685 | 3,680 | | Revenue loss vs. baseline (at $85 AOV) | — | −$114,910 | −$213,065 | −$297,100 |
A 20% annual decay rate — which falls within the range of what brands experience after the novelty effect of initial acquisition wears off — produces a 35% reduction in cohort-level purchase volume over three years. At $85 AOV, that is a $213,065 revenue gap from a single 1,000-customer cohort, before accounting for the acquisition cost required to replace it.
The practical benchmark: if your repeat purchase rate declined by more than 15% from year one of a cohort to year two, you have a decay problem worth modeling explicitly. If you have never measured this at the cohort level, start with your Q1 2023 first-purchase cohort and measure their Q1 2024 vs. Q1 2023 purchase frequency.
The Trade-Off Map
Treating Churn as an Acquisition Problem
The most common operational response to declining repeat revenue is to increase acquisition spend. This works as a short-term revenue fix and fails as a long-term business model for two reasons. First, it does not arrest the decay in the retained base — it layers new customers on top of a deteriorating cohort, creating a larger at-risk population that will begin decaying at the same rate within 18–24 months. Second, acquisition spend efficiency declines as volume scales, as noted above. Brands that respond to repeat revenue decay with acquisition spend alone typically achieve flat revenue with expanding CAC and shrinking margin for 18–36 months before the economics become unsustainable.
Treating Churn as a Retention Problem (Most Common Tactical Response)
Win-back campaigns, loyalty programs, and re-engagement email sequences address symptoms of decay rather than causes. They are worth running — a well-executed win-back sequence at the 60–75 day mark (before the customer reaches the 90-day lapsed threshold where conversion rates drop sharply) recovers 8–12% of at-risk customers at a cost of $3–$8 per recovered customer via email (Klaviyo Win-Back Campaign Benchmarks, 2024). That is a genuine positive ROI intervention.
The limitation is that it does not change the underlying decay rate. If the decay is driven by product fit (customers buying once to try, not finding enough reasons to return), no email sequence corrects it. If the decay is driven by competitive substitution (a competitor has a better loyalty program or faster shipping), no win-back offer overcomes the structural disadvantage. Retention tactics are effective at recovering individual customers; they do not diagnose or fix the mechanism causing decay.
Treating Churn as a Product and Catalog Problem
The highest-leverage response to systematic purchase-rate decay is to identify which products drive repeat behavior and which do not, then restructure acquisition and catalog strategy around that finding. This requires cohort analytics segmented by first-purchase product — a capability that Shopify's native analytics does not provide, but that tools like Lifetimely, Triple Whale, or a custom Supabase query can produce.
Brands that have done this analysis consistently find that 20–30% of their SKUs drive 60–70% of repeat purchases. Customers who enter via those SKUs have measurably lower decay rates. Shifting acquisition creative to emphasize those entry products — even at slightly lower first-purchase conversion rates — changes the composition of new cohorts in ways that reduce decay rates by 8–15% within 12 months.
When to Act (Specific Triggers)
Trigger 1: Build Your Cohort Baseline First
Build a cohort-level repeat purchase rate report if you have never done it and your brand is more than 18 months old. The absence of this data does not mean your decay rate is acceptable — it means you do not know what it is. A Shopify export to a spreadsheet, segmented by first-purchase month and counting purchases by month thereafter, is sufficient to identify the decay curve. This takes 3–4 hours for a first build.
Trigger 2: Intervene in Channel Mix When Cohorts Fall Below Threshold
Intervene in acquisition channel mix if your channel-level cohort analysis shows that any channel's first-purchase cohorts have a second-purchase rate below 22%. Shopify's benchmark for DTC repeat purchase rates puts the minimum viable threshold at 27–34%. A channel delivering below 22% is generating customers who are structurally less likely to return — and those cohorts will decay faster than the blended average, dragging down your overall retention metrics.
Audit your product-level repeat purchase data if your blended repeat purchase rate has declined more than 8 percentage points over two years. This is the signal that the decay is structural rather than cyclical. A catalog analysis — which products drive second purchases, which correlate with long-term customer value — is the correct diagnostic tool at this stage.
Evaluate your win-back timing if your current re-engagement sequences trigger at 90 days or later. Move the first win-back touchpoint to 60 days and measure conversion rate change. If the 60-day sequence converts at 2× the rate of the 90-day sequence (which it typically does), the earlier trigger was capturing recoverable customers that the later trigger was losing permanently.
What Operators Get Wrong Most Often
Mistake 1: Confusing Active Customer Count With Healthy Cohort Behavior
The most invisible mistake is confusing active customer count with healthy cohort behavior. A brand that acquires 500 new customers per month and defines "active" as any customer with a purchase in the trailing 12 months will show a growing active customer count for at least 12 months regardless of its decay rate — because new acquisitions are continuously entering the active window. The dashboard shows growth. The underlying cohort economics are deteriorating.
The number that exposes this is trailing-12-month purchase frequency per customer, calculated for customers in their second or third year, not across the entire base. If that number is declining — even slowly — you have a decay problem that acquisition volume is currently masking. Acquisition is a drug that treats the symptom; measuring year-two and year-three cohort frequency is the diagnostic that finds the disease.
Mistake 2: Treating Lapsed and Lost Customers Identically in Win-Back Campaigns
The second mistake is conflating lapsed customers with lost customers and treating both groups identically. A customer who purchased 4 times last year and once this year is not the same risk profile as a customer who purchased once 14 months ago. Win-back campaigns that treat all non-recent-purchasers as equivalent waste budget on true churned customers who will not respond, and under-invest in the recoverable recently-lapsed segment who will. Segmenting win-back audiences by recency and historical frequency — not just recency alone — typically improves win-back campaign ROI by 25–40%.
The Verdict
Purchase-rate decay is subscription churn wearing different clothes, and it requires the same systematic measurement discipline. Run a cohort-level frequency analysis on your Q1 2023 first-purchase cohort this week, compare their 2023 and 2024 annual purchase rates, and calculate the decay percentage. If the number exceeds 15%, you have a compounding problem worth addressing before it requires acquisition spend to mask.
