Churn in loyalty starts long before you record it

Written by Pavel Los, a loyalty and CRM practitioner with 20+ years in marketing and loyalty, leading large-scale programs across local and international environments (including Shell and Oracle).
I help brands audit, redesign, and activate loyalty programs—improving time-to-value, engagement loops, and measurable ROI through practical frameworks, workshops, and hands-on consulting.

Churn in a loyalty program isn’t just “members leaving”, it’s a message, and the timing tells you what the member is trying to say. Antavo’s research puts hard numbers behind this, 74% of loyalty members “quiet quit” after 2 months, while only 3.4% actively opt out.
Most churn is invisible unless you track inactivity, not just exits.

So how do you make it actionable?

You stop treating churn like one bucket, and you start reading it by phase, and you also acknowledge a critical reality, the signal shows up earlier than the churn record.

The missing angle: when the signal appears vs when churn is recorded

Most teams record churn as a lagging event

• cancellation, opt-out

• “member inactive for 60 or 90 days”

• tier drop, contract end, last purchase outside the window

But the first signal typically appears earlier as behavioural decay

• fewer logins, fewer purchases, fewer offer clicks

• points earned but not redeemed

• slower responses, fewer support interactions, weaker sentiment

This is the difference between leading indicators and lagging outcomes, for example, usage decline can show up weeks before churn is visible in reporting.

Practical takeaway

Your churn dashboard answers, “What already happened?”
Your churn signal system answers, “Who is slipping right now, and why?”

If you only look at recorded churn, you’re doing churn documentation, not churn prevention.

Read loyalty churn by phase, not as one bucket

1) Registration drop-off

They joined, but didn’t reach a first meaningful benefit fast enough.
Think, time-to-first-earn, time-to-first-redeem is too long.

Watch for

• No first earn event, no welcome offer uptake

• Drop-off after sign-up, app install, wallet save

• Confusion signals, “how does this work?” behaviours

Fix it with

• A “first win in 48 hours” plan, make the first earn inevitable, make the first reward feel real

• Fewer steps, fewer conditions, clearer language

2) The honeymoon ends

They tried, but the program feels hard to use, slow to progress, or not worth it.

Antavo’s consumer disappointments map directly to this phase

• 49.1% say it takes too long to earn rewards

• 41.1% are disappointed when points expire before use

• 38.9% find rewards unattractive

Watch for

• Earn without burn, points pi

• Declining purchase frequency among members

• No tier progress plus “not worth it” feedback

Fix it with

• More small wins, fewer “far away” rewards

• Progress that’s visible and motivating

• Expiry design that nudges, not punishes

3) The value check

Long-time members don’t “churn”, they re-evaluate.
“Is this still worth a place in my wallet, especially when alternatives get louder?”

Also watch the perception gap, Antavo reports 82.6% of marketers believe loyalty makes customers feel valued, but only 56.2% of customers agree.

Watch for

• Engagement drops in your best fatigue, price sensitivity, competitor pull

• Lower redemption intent, fewer high-value behaviours

Fix it with

• Recognition that is felt, not just calculated

• Personalised benefits tied to what they actually do

• Reminders of value delivered, “here’s what you gained”, not “here’s what we offer”

How to do it practically: close the “signal lag”

1) Define churn and define the early signal

Pick one recorded churn definition (e.g., inactive 60 days), then define early risk signals that happen sooner (e.g., engagement down 30% week-over-week, no earn in 21 days, no redeem in 90 days).

Leading indicators like declining engagement are explicitly used to predict churn earlier than the churn event itself.

2) Tag every inactive member by phase, then match your playbook

• Registration drop-off → cut sign-up friction, speed up first earn, make “how it works” obvious in the first 48 hours

• Honeymoon ends → make progress feel faster, improve reward appeal, reduce expiry frustration, add reachable moments of magic

• Value check → protect best members with recognition and personalised benefits, remind them of value delivered, make switching feel like a downgrade

3) Build a simple “drift score” (weekly)

Not a fancy model, just a practical early warning

• activity trend (up, flat, down)

• earn to burn pattern

• tier progress velocity

• offer engagement trend

• “days since last meaningful action”

4) Trigger help, not spam

When drift starts, intervene with utility

• “here’s the fastest reward you can get this week”

• “you’re 1 step away from unlocking X”

• “your points are about to expire, here are 3 good redemptions”

Conclusion: churn timing is your loyalty diagnostic

The big loyalty insight is simple, most churn is quiet, and it starts before your churn report admits it.
Treat churn as one problem, and you fix the wrong thing.
Treat it as a sequence, and you get a roadmap, plus am.

Question to think about today
If you split loyalty inactivity into these phases and measured when the first signal appears, where is your biggest blind spot right now, onboarding speed, mid-journey value, or late-stage differentiation?