Loyalty program ROI: how to calculate it without fooling yourself

Most loyalty program ROI numbers you’ll find online are revenue dressed up as return. Case studies quote 400% and 500% because they add up everything members spent, compare it to what the program cost, and call the difference profit. Your CFO will spot the problem in about a minute.

The same program can show 295% or 30% depending on what you count. This guide walks through the formula, a worked ecommerce example that puts four versions of the number side by side, and the measurement setup that gets you to the one you can defend in a budget meeting.

What loyalty program ROI actually measures

Loyalty program ROI measures how much incremental profit your program earns for every dollar it costs. You take the extra gross profit the program brings in, subtract what you spent to run it, and divide by that spend.

Two words in that definition do most of the work. “Incremental” means profit you wouldn’t have earned without the program, so spending by customers who’d have bought from you anyway doesn’t count. “Profit” means you apply your gross margin to that revenue before you compare it to cost, because a dollar of revenue on a 55% margin product only gives you 55 cents to pay for points, software, and staff.

Skip either word and the number inflates. Skip both and you get the figures that show up in a lot of loyalty case studies.

Getting this right matters because results vary so much from program to program. McKinsey found that top-performing programs can lift revenue from customers who redeem points by 15% to 25% a year, while about two-thirds of established programs fail to deliver value. A loose ROI method can’t tell you which group you’re in.

The loyalty program ROI formula

Here’s the formula:

Loyalty program ROI = (incremental gross profit − total program cost) ÷ total program cost × 100

Say your program brings in $825,000 in incremental gross profit and costs $632,500 to run. Your ROI is ($825,000 − $632,500) ÷ $632,500 × 100, which comes to about 30%. The worked example below shows where those numbers come from.

A result of 0% means the program paid for itself and nothing more. Anything below zero means it cost more than it earned.

What counts as incremental profit

Start with incremental revenue: the extra spending from members compared with similar customers who weren’t in the program. That extra spending usually comes from a few places, and you’ll want to see which ones are moving:

• Retention lift: members stay active longer, so fewer of them churn in a given year

• Frequency lift: members buy more often

• AOV lift: members spend more per order, often to reach a points threshold or a tier

Multiply incremental revenue by your gross margin and you have incremental gross profit. If your margin varies a lot by category, use the margin on what members actually buy, since loyalty members often skew toward full-price or higher-margin products (or toward discounted ones, depending on how you’ve designed rewards).

What counts as program cost

Program cost is everything you spend to run the program for the period you’re measuring:

• Loyalty platform fees

• Reward cost, net of expected breakage (more on this below)

• Staff and operations time to run the program

• Marketing to recruit and activate members

• Any discounts or perks you give members that you wouldn’t give otherwise, such as free shipping tiers or early access pricing

Count rewards at what they cost you. A $10 discount costs you $10. A free product costs you the cost of goods, not the retail price, so a $40 candle with a $12 landed cost is a $12 reward.

One cost people often miss: you pay rewards on all member spending, including the spending that would have happened without the program. If a member spends $310 a year and earns 5% back, you’re rewarding all $310, even if only $30 of it is incremental. That’s a big reason honest ROI numbers come out so much lower than the marketing ones.

A worked example: same program, four ROI numbers

Take an illustrative DTC brand that sells home goods online. Its loyalty program looks like this:

• 50,000 active members

• Members spend an average of $310 a year

• 55% gross margin

• Members earn 5% back as dollar-off rewards, so each reward costs its face value

• 70% of issued rewards are expected to be redeemed

Program costs for the year:

• Platform: $30,000

• Rewards: 5% of $15.5 million in member spending is $775,000 issued. At 70% redemption, that’s $542,500 in expected reward cost.

• Operations and marketing: $60,000

• Total program cost: $632,500

The lift. When the brand compares members to non-members, members spend $50 more per year. That’s the apparent lift, and it’s the number most teams report. But the brand also ran a holdout group, randomly keeping a slice of eligible customers out of the program for a year. Comparing everyone who was offered the program with that holdout put the true lift at $30 per member. The other $20 was selection bias: the people who joined were already the brand’s better customers.

Now apply the formula four ways:

The first row is the number you’ll see on a lot of slides. The last row is the one that holds up when finance checks the work.

A 30% return is still a good result: the program earns back its cost plus $192,500 in a year. It just isn’t 295%, and if you’d budgeted next year’s expansion on 295%, you’d be in for a rough review.

Three inputs move the honest number most. If redemption climbs from 70% to 80%, reward cost rises to $620,000 and ROI drops to about 16%. If it falls to 60%, ROI rises to about 49%. If true lift is $20 per member instead of $30, the program loses money, with ROI around −13%. Reward rate, redemption, and lift matter far more than platform fees.

How to handle breakage and reward liability

Redemption is the input finance will press on hardest, because its flip side is breakage: the share of points or rewards that members never redeem. It’s tempting to count breakage as profit, since you collected the spending and never paid out the reward. Finance usually won’t let you, at least not right away.

Under most revenue recognition approaches, when a member earns points, part of the sale is treated as a promise you still owe them. That portion is generally deferred as a liability until the points are redeemed or expire, and companies recognize expected breakage gradually, based on how members have actually behaved.

For your ROI model, the safe approach is the one in the worked example. Estimate an expected redemption rate from your own data, count the cost of rewards you expect to be redeemed, and don’t add breakage back as a separate gain.

Redemption tells you something beyond cost, too. Very low redemption usually means members don’t value the rewards or forget they have them, so the program isn’t changing behavior. Very high redemption can mean rewards are generous enough to squeeze your margin. There’s no universal right number, since it depends on your reward design and category. Track the trend and watch how it moves ROI.

Accounting rules vary by company and situation, so walk through your program’s treatment with your finance team before you finalize the model.

How to measure loyalty program impact

The worked example rests on one number: how much of the members’ extra spending the program actually caused. There are two practical ways to get at it.

Holdout groups

The holdout from the worked example is the most reliable way to measure lift. Because the customers you keep out are chosen at random, the only systematic difference between them and the customers you offered the program to is the program itself. Compare the two groups’ spending, retention, and frequency, and the gap is your incremental lift.

A few things to get right:

• Randomize before people choose. Hold out customers at the point they’d be invited or eligible, not after they’ve signed up. Then compare everyone you offered the program to, joiners and non-joiners alike, against the holdout. Comparing only the people who joined brings back the selection bias you were trying to remove.

• Size it to detect the lift you care about. A holdout that’s too small produces noise. Ask your analyst to run a power calculation based on your customer count and how much lift you expect. A brand with 50,000 eligible customers has much more room to work with than one with 2,000.

• Run it for at least one full purchase cycle. If customers in your category buy two or three times a year, a six-week test won’t tell you much. Many brands keep a small holdout running permanently so they always have a baseline.

Holdouts are the most reliable method, but they cost you something: the customers in the holdout don’t get rewards, and some brands aren’t comfortable with that. If that’s you, use the next approach.

Matched cohorts and pre/post comparisons

When you can’t hold anyone out, build a comparison group that looks like your members. Match members to non-members on what you know before they joined: purchase history, order count, AOV, tenure, acquisition channel, and category. Then compare how each group’s spending changed after the member joined.

A pre/post comparison on the same customers helps too. If a customer spent $200 in the year before joining and $260 in the year after, that $60 is a starting point. Compare it to how much a matched non-member’s spending changed over the same period, since seasonality and price changes affect everyone.

Matched cohorts are less reliable than holdouts because you can only match on what you can measure. Customers who join a loyalty program are often more engaged in ways your data doesn’t capture. Treat the result as an upper bound.

Watch for selection bias and pull-forward

Selection bias is the biggest source of inflated ROI. Your best customers are the most likely to join, so members will outspend non-members even if the program does nothing. The holdout in the example above cut the apparent lift by 40%.

Pull-forward is the second trap. A points expiration or a tier deadline can push members to buy sooner, which looks like lift in the short term. If they’d have made that purchase a month later anyway, the program shifted revenue without adding any. Check whether purchases spike around deadlines and then dip, and measure over a long enough window to see the dip.

Loyalty program metrics that predict ROI

These are the metrics that feed the formula. Track them for members and for your holdout or matched group so you can see what’s driving the result.

• Active member rate: the share of members who bought in the period. Inactive members cost you platform fees and add nothing to lift.

• Redemption rate: the share of issued rewards that get used. This drives your largest cost line.

• Repeat purchase rate: the share of members who bought more than once, compared with the holdout.

• Purchase frequency: orders per member per year, members versus holdout.

• Average order value: members versus holdout, and watch for orders sized to hit a points threshold.

• Retention and churn: how many members are still buying 12 months later compared with the holdout.

• Incremental customer lifetime value: the difference in lifetime value between members and comparable non-members. A one-year ROI misses value that shows up in years two and three, so this is where long-term programs earn their keep.

• Cost per active member: total program cost divided by active members.

• Breakage rate: the share of rewards you expect will never be redeemed.

If you want to go deeper on which customer signals forecast long-term value, we’ve covered the metrics that predict customer lifetime value separately.

How to build a loyalty program business case

Once you know your numbers and what drives them, a business case turns them into a decision: should you launch, expand, or cut this program? Finance teams tend to trust a business case that shows a range and explains its assumptions over one that shows a single big number.

Model three scenarios

Use the formula to build conservative, base, and upside cases. Change the inputs you’re least sure about, which are usually lift and redemption. Using the example brand:

• Conservative: $20 per member in true lift. Incremental gross profit is $550,000, and ROI is about −13%.

• Base: $30 per member in true lift. Incremental gross profit is $825,000, and ROI is about 30%.

• Upside: $40 per member in true lift. Incremental gross profit is $1,100,000, and ROI is about 74%.

Then stress-test redemption. At the base lift, moving redemption between 60% and 80% swings ROI from about 49% down to about 16%. If the conservative case loses money, say so, and explain what you’d change (a lower reward rate, a tighter tier structure) if early results land there.

Show payback period and the cost of not doing it

Payback period is how long it takes for cumulative incremental gross profit to cover cumulative program cost, including one-time setup costs like implementation and launch marketing. First-year results usually trail later years because membership ramps up gradually, so show payback month by month.

Also show the cost of saying no. If you’re deciding whether to launch and competitors in your category all offer rewards, skipping a program can make you the easier brand to leave. If you’re deciding whether to cut an existing program, the risk is bigger: members lose points and status they’ve earned, and some will take that as a reason to shop elsewhere. Both are hard to measure, so present them as risks and not as numbers in the ROI.

Agree on the measurement plan before launch

The fastest way to lose a budget fight is to argue about methodology after the results come in. Before launch, agree with finance on:

• How big the holdout is and how long it runs

• Which metrics count toward lift, and over what period

• What margin you’ll apply

• How you’ll treat reward cost and breakage

• When you’ll review results, and what result would trigger a change

Write it down. When the six-month numbers come in, everyone will be reading them the same way.

Why service belongs in your loyalty ROI model

A measurement plan covers the levers you already track. Most loyalty ROI models count points, tiers, and emails, but few count what happens when a member contacts support, even though that’s often when a loyal customer decides whether to stay.

A bad service experience can undo years of points. In Zendesk’s 2023 CX Trends report, 73% of customers said they’d switch to a competitor after multiple bad experiences, and more than half said they’d leave after one. PwC found that 32% of customers would stop doing business with a brand they loved after a single bad experience. Your most loyal members are the customers you can least afford to lose that way, and their retention is a big part of your program’s incremental value.

Service shows up in the ROI formula in two places. It protects retention lift, because members who get a fast, informed answer don’t churn over one bad order. And it can save members who are about to leave, which turns a lost customer into continued incremental revenue.

To make that happen, your support team needs to know who they’re talking to. With Gladly, loyalty data from platforms like Yotpo and Okendo shows up right in the conversation, including VIP tier and points balance. You can use Yotpo data like VIP tier to route top-tier members to a priority team, and Gladly AI can use loyalty status to answer points questions or personalize a resolution in self-service. When a team member needs the full picture fast, Team Assist’s “Summarize this customer” pulls loyalty status and lifetime value into one snapshot.

Ollie, the fresh dog food subscription brand, built its service operation around this idea.

Ollie’s team now saves 45% of at-risk customers who reach out about canceling, and proactive SMS campaigns cut churn by 7%. Read the full Ollie story to see how they set it up.

If you want to put a number on the service side of retention, the CX ROI calculator can help you model it.

Get the number you can defend

Loyalty program ROI comes down to four choices: revenue or profit, members versus non-members or members versus a true comparison group, rewards at face value or at what they cost you, and breakage as a gain or as an estimate inside your reward cost. Make the honest choice on each, and you’ll get a smaller number that you can stand behind in a budget review.

Then use the metrics to find out what’s moving it, and don’t leave service out of the model. Your loyalty members will contact support eventually, and how that goes affects how long they stay.