Customers Don't Always Churn
Sometimes They Fade. And That Changes Everything - (Part 2)
Mark Sage - 7 min read - 19/01/2026
The first article pulled apart a comforting idea — that loyalty is something you can understand by watching who leaves. The leaky bucket, the churn rate, the monthly dashboards all suggest a clean moment of exit, a clear before and after.
But once you stop looking at averages and start following real customers over time, that clarity disappears. What you see instead are very different journeys. Some customers vanish almost immediately, some drift away slowly, and a small group never really seem to leave at all.
Survivorship analysis helped make sense of that mess. It showed us whencustomers disengage, not just how many. But it also revealed something else; it assumes that customers actually fall out of the system.
In grocery — and lots of other sectors — they don’t.
They keep shopping. They keep showing up. They may still look “retained” on every dashboard you have, or possibly keep dipping in and out of lapsed; yet, something is changing. Trips become smaller. Missions narrow. Categories disappear. Frequency drops. The customer hasn’t gone anywhere — but the relationship is slowly thinning.
This is the problem churn can’t see.
Customers not falling out of the loyalty programmes, but fading.
Fading. Not Falling.
When we built yuu Rewards, we deliberately didn’t talk much about churn — certainly not in the way SaaS companies or subscription services obsess over it. Grocery simply behaves differently. The rhythm of shopping is fast, frequent, and habitual, and the customer base is comparatively stable. Most FMCG and CPG manufacturers measure performance in rolling 12-week windows for exactly that reason. So, for yuu, our core health metric wasn’t churn at all — it was retention, measured as 12-week active (P12W) divided by 12-month active (P12M).
At its heart, this metric captured something very simple — are your customers still shopping with you recently?
The 12-month window tells you who is “in your orbit”; the 12-week tells you who is still in motion. It’s less about whether someone has disappeared entirely, and more about whether they’re still behaving like a regular grocery shopper.
You could say that rather than being focused on those who stopped shopping, we focused on our ability to keep people shopping — and eighteen months after launch, yuu still showed a consistent 85%+ retention rate among customers active in both windows. There was no early-life cliff, no sharp drop-off. Even when we compared P12W active customers against the entire member base from the launch day — 18 months later — the retention figure still sat above 80%.
On the surface, that’s an impressively strong result. But the real question is, what does this metric actually tell you inside a grocery ecosystem?
The reason this measure works so well is that grocery does not swing like other sectors. You don’t suddenly get a tidal wave of new customers flooding in the way you do in apps, telco, or SaaS.
In grocery loyalty, something very predictable happens — around half of your entire two-year member enrolment arrives in the first month.
That’s launch month, when the signage goes up in every store, when every cashier is trained to invite every single shopper to join. With yuu Rewards, after the initial surge to two million members in month one, recruitment settled into a steady rhythm, rising at roughly 4–5% per month for the rest of the first year, before gradually tapering as the programme matured.
Unlike DFS — the travel retailer — where tourists flowed in and out on a daily cycle — and where spikes in new customers appeared around Golden Week, Christmas, or major travel seasons — grocery customers are creatures of habit. They shop regularly. They develop routines and anchor missions. Once they establish their repertoire of stores, they tend to stay unless something meaningfully pushes them away.
This stability makes the 12-week/12-month ratio extremely reliable. Even two years after launch, the retention rate remained consistently high.
That however is retention of customers, not necessarily customer spend.
And this is the big difference with an everyday spend sector like grocery — a customer can easily appear retained while quietly leaking value.
Someone might still shop with you — but switch from a weekly top-up shop to a “buy as you need” pattern. That means they return within the 12-week window, but bring a smaller basket, fewer categories, or fewer full-service missions.
In grocery, it’s actually rare to lose a customer entirely. Instead, what you lose is not the customer, but their interest.
To monitor this, we tracked two key measures alongside retention.
RFM Migration (Recency, Frequency, Monetary Value) — Examining how individual customers moved between RFM segments. If they were slipping — shopping less often or spending less — we saw it immediately. A member might still be active, but their value was quietly eroding. RFM migration let us expose the slow leaks that retention alone would never catch. In Harrison’s terms, this is detecting the “decay curve” rather than waiting for a cliff event that will never come.
Cross-Sponsor Rate (Coalition Breadth) — As a coalition programme, we wanted members to shop across partners. A customer could remain active in yuu overall, yet become inactive with a particular sponsor — effectively disappearing from an entire category. Cross-sponsor rate helped catch this. We tracked how many banners each customer shopped and whether that breadth was widening or shrinking.
Two years after launch, 80% of active members either stayed in the same RFM segment or moved up, and the cross-sponsor rate had grown by 57%.
So not only did we have a super strong retention rate, but we’d grown both value and depth.
More interesting than the headline number though, was how cross-sponsor behaviour changed. The shift wasn’t simply “more people shopping one extra brand.” It was a structural broadening of behaviour. Over the first year, the number of members shopping four or more banners nearly doubled. The 5+ banner cohort more than doubled.
That movement — from light to broad participation — is the strongest signal that a coalition is working. Breadth begets depth.
Multi-banner shoppers form habits, stack missions, and feel gravitational pull from the ecosystem rather than individual stores. That’s not promotional traffic; that’s ecosystem loyalty.
The real issue in grocery then is not churn, but slow disengagement. This is the big truth — churn is too simplistic for a grocery-led coalition like yuu Rewards.
By the time a customer stops all activity — and by the time you notice — it’s already too late to intervene. The churn event you can measure is the end of a long journey of fading engagement.
The churn that matters most in grocery isn’t churn at all. Instead, it’s the quiet, gradual dripping away of behaviour.
Customers don’t fall out of the system, they simply fade.
For yuu, retention told us who was still engaged. But the RFM migration told us who was quietly slipping and the cross-sponsor breadth told us whether the coalition was strengthening or weakening. If we had measured churn alone, we would have been almost completely blind.
What Harrison’s work makes clear is that churn only tells a meaningful story when you understand how long customers survive, not simply how many remain.
DFS proved this the hard way, in that the traditional churn metric suggested stability, but survivorship analysis revealed a steep early cliff hidden beneath a rising tide of new customers. Once we grouped customers into meaningful cohorts and tracked how long they stayed, an entirely different picture emerged. Only then could we see which customers were crossing the threshold into loyalty, which channels brought in long-term value, and where the structural limits of the category sat.
It’s worth noting that even with our focus on cohort survivorship, we didn’t abandon traditional KPIs. We still tracked a 12-month ‘Retain’ number, repeat rate, active base, and new sign-ups — the same operational metrics used in almost every loyalty programme. These figures remained essential for forecasting, budgeting, and communicating performance to the organisation.
But the difference was this — the metrics were accurate as outputs, yet less useful as inputs.
They showed overall progress, but not the underlying behaviour or the levers we needed to steer. Survivorship gave us that missing visibility.
yuu Rewards, by contrast, demonstrated the opposite truth. In grocery, customers don’t disappear en masse, so churn stops being useful. Habits are strong, rhythm is predictable, and the real challenge isn’t a sudden drop-off — it’s the slow erosion of value.
Retention, RFM migration, and cross-sponsor breadth gave us the tools to see this. They showed us who was still engaged, who was slipping, and how deeply members were participating across the ecosystem. And, crucially, they helped us spot the drip-drip-drip of behavioural loss that churn metrics are blind to.
Together, these two experiences reinforce a single point — you cannot measure loyalty with a single metric, and certainly not with churn alone.
The job of the loyalty marketer is not simply to shrink the hole at the bottom of the bucket but instead is to understand the shape and placement of the buckets, the pattern of the spray, and the journeys customers take after they join.
If you measure the wrong thing, you will manage the wrong thing.
Churn may be a familiar metric, but loyalty is rarely that simple and so understanding it properly requires looking at how customers live, not just the moment they leave.