A subscription business sees each customer it loses, because the customer cancels on a known date. A points program sees only silence, so the operator decides when silence counts as gone. That choice moves the churn rate as much as members do.
What is churn rate?
Churn rate is the percent of customers lost over a period. Jill Avery of Harvard Business School defines it as the percent of customers “who end their relationship with a company in a particular period” (HBR, 2014). For the same group and period, churn and retention add up to 100%.
Churn is easiest to count where leaving is on record, as with a phone plan. In noncontractual settings, such as retail stores, the time a customer leaves “is unobserved by the firm” (Fader, Hardie and Shang, 2010). Loyalty programs sit in that group.
How do you calculate churn rate?
Divide the customers lost during the period by the customers at the start, then multiply by 100. Count only customers there on day one, so new joiners who leave fast do not skew the rate. SiriusXM instead divides average monthly self-pay deactivations by average self-pay subscribers (SiriusXM 10-Q). Pick one method and keep it.
Match the period to how customers buy. Avery says annual is the default, while firms that price by the month track churn monthly (HBR, 2014). Monthly rates compound, so twelve times the monthly rate overstates the annual one, as our guide to customer retention metrics shows.
| Measure | Formula | Inputs to fix first |
|---|---|---|
| Customer churn rate | Customers lost in the period ÷ customers at the start × 100 | Who counts at the start, and what counts as lost |
| Gross revenue churn | Recurring revenue lost to exits and downgrades ÷ recurring revenue at the start × 100 | The same starting customers, with upgrades left out |
| Net revenue churn | (Revenue lost minus upgrades from the same customers) ÷ revenue at the start × 100 | Equal to 100% minus net revenue retention, so it can fall below zero |
| Voluntary or involuntary churn | Customers lost for that reason ÷ customers at the start × 100 | A reason code on every exit |
| Inactivity churn (loyalty) | Active members at the start with no activity in the lapse window ÷ active members at the start × 100 | Window length, and what counts as activity |
What is the difference between customer churn and revenue churn?
Customer churn counts heads. Revenue churn counts money, so a big account that leaves weighs more than a small one. Gross revenue churn counts what existing customers stopped paying. Net revenue churn subtracts what the same customers added through upgrades, so it can turn negative.
Datadog reports the mirror image, net revenue retention. It measures what customers from 12 months earlier pay now, with upgrades in and new customers out. Its rate was about 120% at the end of 2025 (Datadog 10-K), so the same customers paid about a fifth more.
What is the difference between voluntary and involuntary churn?
Voluntary churn is a customer who chooses to leave. Involuntary churn is a customer lost to a failed card or an unpaid bill, and its fix sits in billing. SiriusXM’s self-pay monthly churn fell to about 1.4% in the second quarter of 2026. It credited lower vehicle-related and non-pay churn, partly offset by slightly higher voluntary churn (SiriusXM, Q2 2026).
What is a good churn rate?
No single churn rate is good for every business. Avery calls the belief in a magic number a common mistake, since what is acceptable “varies widely by business model” (HBR, 2014). Compare churn only with firms that count it the same way, and with your own history.
How do you measure churn in a loyalty program?
Points members leave the way store customers do. They “just silently attrite,” in a line Fader, Hardie and Shang quote. The only evidence is “an unusually long hiatus” since the last purchase (Fader, Hardie and Shang, 2010). So program churn is a rule the operator writes: no qualifying activity for a set number of months means lapsed.
Programs that tie points expiration to inactivity already have that rule in their terms. Marriott Bonvoy forfeits all points after 24 months in a row of inactivity. After five years without active status, it can deactivate the account (Marriott Bonvoy terms). Starbucks runs three clocks. A Green member’s Stars expire six months after the month they were earned. Each purchase or redemption keeps them alive another month. An account with no Stars earned for two years in a row can be unregistered (Starbucks Rewards terms). Its January 2025 results counted 34.6 million U.S. members active in the past 90 days (Starbucks, Q1 FY2025). An active count is a size. Churn needs the same members on two dates.
Set the churn window from the member’s own pace. A window shorter than the normal gap between purchases flags members who were never leaving. A window as long as the forfeiture clock reports the loss after the balance is gone. The model of Fader, Hardie and Lee judges this from two facts per member, recency and frequency (Fader, Hardie and Lee, 2005).
For illustration, a grocery program starts the year with 1,000,000 active members. By December, 300,000 have not earned for six months, so a six-month window reports 30% churn. Only 180,000 have been quiet for 12 months, so a 12-month window reports 18%. Next spring, 100,000 of the 300,000 earn again. A third of that 30% was a long gap between shops. If points forfeit at 24 months, the 180,000 still give the program a year to win them back.
What does churn cost a loyalty program?
The famous figures come from Fred Reichheld of Bain & Company. In 1990, he and W. Earl Sasser found that cutting defections by 5% raised profits 25% to 85% in the firms studied. A bank’s branch system gained the most (HBR, 1990).
In 2000, Reichheld and Phil Schefter summed up the same work as 25% to 95% for a 5% rise in retention (HBR, 2000). Both ranges describe the firms studied. Neither is a law for every program. A 2002 study of more than 16,000 customers found little or no evidence that steady customers are always cheaper to serve (Reinartz and Kumar, 2002).
Ulta Beauty books loyalty revenue when points are redeemed or expire (Ulta 10-K). Under rules like that, a lapsed member’s expiring points can lift reported revenue in the year they go. The member’s future margin is gone all the same. Retention economics counts that margin.
What signals warn of churn before it shows?
Churn is a lagging number. Avery quotes HubSpot’s Jonah Lopin: a rise in churn shows up “six or eight months after the point in time when you actually failed the customer” (HBR, 2014). A points program creates its own earlier signals:
- Points earned but never spent, so the member never feels the reward.
- A balance near its expiry date with nothing new on the account.
- A tier drop that takes away perks the member used.
- Opt-outs from program email, which cut the channel a save offer needs.
Target by response as well as risk. Eva Ascarza tested both on data from a phone carrier and a membership group. Picking customers by likely response to a save offer cut churn more than picking those likeliest to leave (American Marketing Association). Members past the window need a win-back campaign.
How GRAVTY handles churn rate
In a points program, churn is a definition, so the data under it has to be complete. GRAVTY runs 500M+ members in production. It keeps each earn, burn, tier change and offer response as a timestamped event on the member’s record. Expiry sits in the same program model as benefits and family accounts.
Member 360 pulls each member’s behaviors and transactions into one view, household included. When the active count drops, a team can ask Agentic AI Compass why the metric changed and compare any segment in plain language. Visual Rules, the patented no-code rules engine, lets loyalty teams put reactivation rules live without an IT ticket. The save offer can then reach a lapsing member while the balance still counts.
Frequently asked questions
What is the difference between churn rate and retention rate?
Churn rate is the share of a starting group that left, and retention rate is the share that stayed. For the same group and period, the two add up to 100%. Avery notes that many executives prefer to report retention, and that both figures look at the same thing (HBR, 2014).
Should churn rate be measured monthly or annually?
Measure churn on the cycle customers buy and pay on. Avery says annual is the default for most companies, while firms that price by the month track it monthly (HBR, 2014). Convert between the two by compounding. Multiplying a monthly rate by twelve overstates the annual rate.
Can churn rate be negative?
Customer churn cannot fall below zero, but net revenue churn can. It goes negative when existing customers add more through upgrades than leavers and cutbacks take away. Datadog kept about 120% of year-earlier recurring revenue from the same customers at the end of 2025 (Datadog 10-K). That is net revenue churn of about -20%.
How long should a member be inactive before counting as churned?
Longer than the normal gap between that member’s purchases, and shorter than the points forfeiture clock. Real programs run very different clocks. Starbucks reported members active in the past 90 days (Starbucks, Q1 FY2025). Marriott Bonvoy forfeits points after 24 months in a row of inactivity (Marriott Bonvoy terms).
Is a high churn rate always bad?
No. Avery says some business models thrive despite high churn. She also warns that heavy upfront price deals attract deal seekers, who leave fast when they find a better deal elsewhere (HBR, 2014). For a program, that points at rich sign-up bonuses as a source of churn.
What happens to a churned member’s points?
They stay on the books until the program’s terms expire them. Marriott Bonvoy forfeits points after 24 months in a row of inactivity and does not reinstate them (Marriott Bonvoy terms). When points expire, the revenue held against them is recognized (Ulta 10-K).