Your customers don't live on your lifecycle calendar

Why fixed 30/60/90-day flows reach most customers at the wrong moment, and what to measure instead.

Alex JostAlex Jost
Founder of RetentionX · Nov 19, 2026 · 6-minute read
Hard Margins
Ep 39 · New
Play
  
SpotifyListen on SpotifyPlay episode →

Most retention programs are built around time. 30 days after purchase, send this. 60 days later, send that. At 90 days, call them lapsing. At 120 days, start the winback flow. It is clean and easy to automate. It is also fundamentally different from how customers actually buy.

One customer may naturally reorder every four weeks. Another every eight. A third may buy the exact same product every twelve weeks because they use it more slowly. Put all three into the same 60-day flow and you reach them at three completely different moments. For one customer, you are late. For another, perfectly timed. For the third, you are selling something they probably still have at home.

Three customers in one fixed day-60 flow: one reminder is late, one on time, one early

The calendar belongs to the brand. The buying rhythm belongs to the customer. And a lot of retention performance gets lost in the gap between the two.

The average customer doesn't exist

Imagine a beauty product with an average replenishment time of 60 days. That sounds useful, so the brand builds its replenishment program around day 55. But the average hides the actual behavior. Some customers reorder around day 35. Others around day 60. Others consistently take 90 or 100 days. The average might be 60 days even though relatively few customers actually behave like a 60-day customer.

That matters because retention is ultimately an individual decision problem:

  • Should I contact this customer now?

  • Are they early, on time, or actually late?

  • What should I say?

  • Do they need an incentive at all?

Company averages are useful for understanding the business. They are much less useful for deciding what to do with one customer today.

The next order is not a date. It is a probability curve.

I think the better question is not “When will this customer reorder?” It is “How does the probability of their next order change over time?”

Immediately after a purchase, the probability is usually low. As the customer approaches their normal buying window, it rises. If they pass that window without purchasing, something else starts happening. At first, they may simply be slightly late. Eventually, the absence of the expected order becomes meaningful. Maybe they still have product. Maybe consumption slowed. Maybe they switched brands. Maybe the product disappointed them.

That creates several very different customer states: too early → approaching the buying window → high-probability window → overdue → increasing churn risk.

Probability of the next order over time, with five customer states from too early to rising churn risk

Most lifecycle systems reduce all of that to days since last order. That is too crude.

Timing changes the economics

This is where the idea becomes much more important than simply sending better replenishment emails. Imagine you send the same 15% offer on day 45 to three customers.

Customer A usually purchases around day 40. The incentive might recover an order that was beginning to slip. Customer B normally buys around day 55. You may have just discounted an order that would have happened naturally ten days later. Customer C normally buys around day 90. The offer is probably irrelevant. Same campaign. Same discount. Completely different economics.

The same 15% offer on day 45 for three customers who usually buy around day 40, 55 and 90

That is why I think retention teams should stop asking only “Did this campaign generate revenue?” and start asking “What was the customer expected to do without us?” A retention program can generate a lot of attributed revenue while quietly giving margin away to customers who were already about to buy.

The closer you get to the customer's natural buying moment, the less persuasion you should need. Good retention timing should reduce the economic cost of creating the next order.

Churn should be relative to expected behavior

Individual cadence also changes how I think about churn. Take two customers who have not purchased for 75 days. Customer A normally buys every 30 days. Customer B normally buys every 100 days. The absolute inactivity period is identical. The behavioral signal is completely different. For Customer A, something has clearly changed. For Customer B, nothing unusual has happened yet.

Two customers at 75 days without an order: one is 45 days past expected, the other 25 days before expected

So instead of defining churn as “no order for 90 days”, I would define it relative to expected behavior: how far has this customer moved beyond their normal purchase window?

That makes the first deviation from normal behavior much more interesting than the eventual “lapsed” label. A customer who historically purchases every 45 days and suddenly reaches day 65 is already telling you something. You do not need to wait until day 120 to notice. That is behavioral drift. And catching it early gives you many more options than trying to resurrect the customer months later.

The product changes the rhythm too

Customer history is only part of the prediction. What they bought matters. A cleanser, lipstick, supplement, running shoe and handbag do not create the same next-order journey. Even within one brand, one product might replenish in 30 days while another naturally leads to a different category three months later.

That means the strongest prediction should combine customer rhythm (how this person tends to buy) with product rhythm (what usually happens after customers buy this product). The overlap between those two can tell you much more than either one alone.

Customer rhythm and product rhythm as two likely windows, and the narrower window where both agree

This becomes especially important after order #1. A new customer has no individual purchase history yet. But you do know what they entered through. That product can tell you when healthy customers normally come back, what they tend to buy next, and how quickly the relationship should progress.

So instead of asking “How many days since order #1?” you can ask “Given what this person bought, when should a healthy next purchase happen?” That is a much better starting point for converting order #1 into order #2.

Prediction should lead to less marketing, not more

The goal of better prediction is not to create more automated messages. It should often mean sending fewer. A customer who is highly likely to purchase naturally may need nothing. Someone approaching their normal buying window might need a subtle reminder. A normally reliable customer whose cadence suddenly breaks may deserve much more attention.

And being overdue does not automatically mean sending a discount. Sometimes the right intervention is:

  • a replenishment reminder

  • a relevant next-product recommendation

  • product education

  • a substitute for something unavailable

  • personal outreach

  • or simply no intervention at all

Five customer states relative to the usual next order, with the lightest useful intervention for each

Prediction should not mean “We think they'll buy in seven days, so send them 15% off in five.” It should mean “We understand where this customer is in their journey, so we can use the lightest intervention required to keep it moving.” That is much more economically interesting.

The metric I would change

If brands move from fixed lifecycle schedules to individual buying rhythms, retention reporting should change too. These are the four numbers I would care about.

✓
The Operator Playbook
Report the rhythm, not the calendar
Four numbers to put in the retention report instead of revenue attributed to the flow.
✓
Purchase Window Conversion
How many customers purchased during their expected next-order window?
✓
Overdue Recovery Rate
How many customers who moved beyond their expected window were recovered?
✓
Cadence Change
Are customers returning faster or slower over time?
✓
Margin per Intervention
How much incremental contribution did the intervention create after discounts and other incentives?
✦  Try the AI Analyst  →

Those metrics get much closer to the real objective than “revenue attributed to the flow.”

The takeaway

For years, retention marketing has been built around fixed calendars because fixed calendars were easy to automate. Customers do not buy that way. They have individual rhythms shaped by what they buy, how quickly they consume it, how frequently they shop, and how their relationship with the brand evolves.

So I think the next generation of retention moves from days since last purchase to distance from expected next purchase. Because the important question is not “Has it been 60 days?” It is “Is this customer behaving the way we would expect them to?”

Once you know that, you can intervene when it actually matters. Or, just as importantly, know when to leave the customer alone.

Retention should move from fixed schedules to individual buying rhythms.

– Alex