# Loyalty Programs and Retention: What Transfers to SaaS

> The research on loyalty programs is good, large, and almost entirely about retail. Here is the part that survives the move to software.

- Author: Rishikesh Ranjan · Published: Aug 8, 2026 · Updated: Aug 8, 2026
- Type: Playbook
- Tags: Retention, Frameworks
- Growth levers: Retention (primary)
- ~1138 words

---

The evidence on loyalty programs is better than most growth topics get. There are meta-analyses, effect sizes, and decades of retail data. The honest summary is that loyalty programs work **conditionally**, and one of those conditions should change what you build.

The condition is who moves. Reviewing that literature, [Loyalty and Reward Co](https://loyaltyrewardco.com/do-loyalty-programs-work-a-review-of-scientific-evidence/?utm_source=productgrowth.blog) reports a finding from Liu (2007) that programs showed no effect on heavy buyers, while light and moderate buyers improved and kept improving. Read that next to how most loyalty schemes are designed and the problem is obvious: **the tiers pay your best customers, who were not leaving**.

> **Where these numbers come from:** The studies cited on this page are as reported in that review, which is published by a consultancy that sells loyalty program design. Their conclusion, that programs work when designed well, is what a seller would conclude, so the numbers below are the ones that survive that reading: the failure modes and the segment findings, which cut against the sales pitch.

## The evidence, including the parts a vendor would not lead with

| Finding, as reported in that review | What it means for you |
| --- | --- |
| Liu (2007): no effect on heavy buyers; light and moderate buyers improved | Rewarding your top tier is the least efficient thing the program can do |
| Bombaij and Dekimpe (2020): direct rewards stopped working once competitor penetration passed about 75% | In a category where everyone runs one, a program is table stakes, not an advantage |
| Lin and Bowman (2022): launch spikes dissipated in roughly six months | Judge a program on month nine, never on the launch quarter |
| Nishio and Hoshino (2024): birthday rewards cut lifetime value from $125 to $72 | A badly aimed reward can actively destroy value, not merely waste budget |
| Wallström and colleagues (2024): monetary rewards reduced emotional loyalty | Paying for loyalty can buy out the thing you were trying to build |
| Ferguson and Hlavinka (2007): about 39.5% active participation across 1.3 billion US memberships | Enrolment is not a metric. Most members are inactive |

> “A loyalty program that only moves people who were already staying is a discount with extra steps.”

## What does not transfer from retail to software

Nearly all of that research is retail: grocery, airlines, quick service, beauty. Software breaks three of its assumptions at once, which is why copying a points scheme into a SaaS product usually goes nowhere.

- **Purchase frequency is not the behaviour you want. **Retail loyalty buys another visit. Subscription software is already paid for this month, so rewarding a purchase rewards nothing. The behaviour worth reinforcing is usage depth, and that is a product problem.
- **The buyer and the user are often different people. **Points earned by an end user do not reach the person who signs the renewal. Consumer loyalty design assumes one wallet and one decision maker, and B2B has neither.
- **Switching costs already exist. **A grocery shopper can defect this afternoon. A team with data, integrations and workflows in your product cannot, so the marginal value of a points balance is small next to what they would lose by leaving.

## B2B customer retention tactics that use loyalty rewards well

The transferable idea is not points. It is that recognition and status change behaviour more cheaply than discounts do, and that the effect concentrates in the middle of your base rather than the top.

1. **Aim at the middle, not the top. **Find the accounts using you weekly but shallowly. They are the ones the evidence says move, and they are usually invisible because nobody escalates a healthy-looking account.
2. **Reward the behaviour that predicts renewal. **Pick the usage that correlates with staying, then make more of it worth something. Your [retention metrics](https://www.productgrowth.blog/p/customer-retention-metrics) already tell you which behaviour that is.
3. **Use status, access and expertise before money. **Early access, a named contact, a certification, a customer advisory seat. All are cheap, none is a discount, and the research on monetary rewards reducing emotional loyalty is a reason to keep it that way.
4. **Do not touch price. **A loyalty discount is a permanent price cut that repeats every renewal, which is the same trap as save-discounts in [customer retention cost](https://www.productgrowth.blog/p/customer-retention-cost). It books as retention and behaves as churn you paid for.
5. **Judge it at month nine. **Launch spikes dissipate in about six months, so a program measured in its first quarter always looks like a win.

## When not to run a loyalty program at all

Two of the findings above are really disqualifiers. If most of your category already runs a program, the evidence says yours will not differentiate you, so the budget belongs somewhere with a gradient. And if your churn is concentrated in the first month, a loyalty scheme is aimed at a stage your users never reach: that is an activation problem, and the [twelve retention plays](https://www.productgrowth.blog/p/customer-retention-strategies) put activation first for exactly this reason.

The cheapest test before building anything is to check whether loyalty is your problem at all. Split your [churn rate](https://www.productgrowth.blog/calculators/churn-rate) by tenure. If it is flat across cohorts, you have a product problem that no reward fixes. The rest of the [retention writing](https://www.productgrowth.blog/retention) here is aimed at that case.

> **Steal this:** Before you design tiers, list the ten accounts your program would reward most and ask what the odds were that any of them churned this year. If the honest answer is low, you are about to spend the budget on people who were staying anyway, which is what the evidence says most programs do.

## FAQ: loyalty programs, retention and rewards

#### Do customer retention loyalty programs actually work?

Conditionally, and the condition matters more than the headline. Reviews of the research report that loyalty programs show little effect on heavy buyers while light and moderate buyers improve, which means a scheme built around rewarding your top tier is aimed at the customers least likely to leave. The same literature reports that launch effects dissipate in roughly six months and that direct rewards stop differentiating once most competitors run a program, so both the timing and the category matter when judging one.

#### Do loyalty programs work for B2B SaaS?

Points-based programs usually do not, because software breaks three assumptions the retail research rests on. The subscription is already paid, so rewarding another purchase rewards nothing; the person earning rewards is often not the person who signs the renewal; and a team with data and integrations in your product already faces switching costs that dwarf a points balance. What does transfer is recognition and status, meaning early access, a named contact, certification or an advisory seat, which change behaviour without cutting price.

#### What is the difference between customer retention and loyalty?

Retention is a measured behaviour, the share of customers still with you at the end of a period. Loyalty is the preference behind it, the reason someone would stay even when a cheaper option appears. The two come apart in practice: a customer locked in by a contract or by migration cost is retained but not loyal, which is why one study reported that monetary rewards can raise repeat purchase while reducing emotional loyalty. Measure retention, but do not assume it proves loyalty.

#### When should you not build a loyalty program?

When most of your category already runs one, since the research reports direct rewards stop differentiating past roughly 75% competitor penetration, and when your churn is concentrated in the first weeks, because a loyalty scheme rewards a stage those users never reach. Split your churn by tenure first. If it is flat across cohorts, the problem is the product and no reward structure will fix it.

Last reviewed August 2026. The studies cited here are as reported in the review linked above rather than read in the original, which is worth knowing if you plan to quote any of them in a business case.

---

All posts: https://www.productgrowth.blog/archive · Site: https://www.productgrowth.blog