If you’re wondering how to launch a loyalty program, the first step is not “copy Starbucks” or “find a software vendor”. Here’s how to do it right.
This content was first published in the No Best Practices newsletter on 2.25.2024.
I throw a lot of shade at points-based loyalty programs. But I realized that I’ve never done a proper write-up about the right way to build a loyalty program. So here it is 🙂
Step 1: Define Your Loyalty Program’s Objective
“Loyalty” is one of the fuzziest terms in marketing and eCom. I’ve never seen a brand define it with a hard KPI.
Similarly, loyalty programs rarely have quantifiable objectives. If the objective of your loyalty program is to “increase loyalty” but you have no yardstick to measure loyalty…then what are you doing?
Another popular objective for loyalty programs is “make sales go up”. Not even “make repeat purchases go up”. These programs are often a band-aid slapped on a business with customer acquisition issues.
I like to define loyalty as the stage when the average customer has a greater than 50% likelihood of making their next purchase. For most brands, this happens between purchases 3 and 5.Â
Likelihood of proceeding to a subsequent purchase generally improves with each purchase a customer makes. So once they hit the loyalty point, they’re relatively “locked in”.
Observing this, you can break repeat customers into two segments: loyal, and not yet loyal. Both of these segments respond to different incentives. A 1x buyer thinks and behaves more like a prospect than a loyal buyer. So you can’t assume that one set of offers or incentives will work uniformly across your repeat customer base.
Some sample objectives for a loyalty program:
- Grow our loyal customer base by improving the percent of customers who progress from purchase #1 to purchase #3.
- Increase LTV from loyal customers (4+ lifetime orders) by increasing AOV.
A good objective contains (1) a highly targeted audience and (2) a KPI you’re trying to change.
Step 2: Estimate The “Size Of The Prize”
Another reason loyalty programs fail, or at least fail to impact the bottom line: the launch and maintenance of the program costs more than it could ever earn back in incremental revenue.
Before you decide to run with an objective, you need to calculate a realistic upside. Here’s how you do that:
- Determine how many customers were in the segment L12M, what they spent during that period, and the resulting order frequency and AOV.
- Estimate a realistic lift for the metrics in your objective. For example, you might lift AOV from $150 to $160. You’re probably not going to double it.
- Net out the impact of any discounts or promotions you plan to run to accomplish this lift.
- Subtract the “lift case” from the “base case” (what actually happened L12M). Those are the 12 month “earnings” of your loyalty program.
- Imagine you delivered those earnings for 2024. How big is that number within the context of your brand’s topline revenue and the growth you hope to achieve.
- Discount the lift by your probability of achieving it. 50-70% is fair.
Loyalty programs often overestimate the incremental impact they’ll be able to realize.Â
Your most loyal customers are probably buying from you every 30-60 days. Improving on that frequency in a meaningful way is hard (unless you’re selling drugs). Improving AOV is more realistic, but if you do it with promotions, you often wind up with less gross margin vs doing nothing at all.
If you want to make the juice worth the squeeze, you’re probably better off designing a program that leads more prospects down the path to loyalty. Points programs don’t do this. Prospects and 1x buyers won’t jump through hoops; they’re still in “what have you done for me lately” mode.
Step 3: Brainstorm & Test Loyalty Incentives
This is the hardest part of the process because you have to get creative. The best incentives come from two places: (1)Top decile brand equity and/or (2)Deep understanding of your customers’ needs and motivations.
Only 10% of brands have #1 by definition. Kith (and their recently launched loyalty program) is a great example of a brand with top decile brand equity. Unless you are at this level, you’ll have to lean on #2–customer knowledge.Â
You also have to take stock of all the other incentives a customer may be exposed to on your site: sitewide offers, affiliate offers, welcome offers, post-purchase offers, etc. Literally make a list of all the offers you offer, how they’re triggered, and what channel(s) are used to distribute them.
If your brand has been leaning on promos as an incentive for a long time, you might find that it’s hard to create an incentive that cuts through the noise. So if you really want to go “all in” on your objective, you might have to clean up and pull back on incentives offered elsewhere.
Once you have a list of 3-5 incentives you want to test, the next step is to perform a true holdout test on the customer segment you’re targeting and measure the 30-90 day impact of the incentive.
Email is a great venue for this test, because it’s relatively easy to segment an audience and keep your comms about this test private. That’s right–you don’t want this to leak. Don’t call it a loyalty program. Don’t publicly commit to anything until the offer has been validated in testing.
If you have a big target segment (from step 2), you can split it into 4 sub-segments: the holdout group (gets nothing/”business as usual” email comms) and one segment for each offer (assuming you’re testing 3 offers).
Message the offer over a period of one to two weeks. Send 1-2 emails reminding the segment to use the offer. And then analyze segment purchasing behavior over the next 30-90 days.
If one of your segments shows a stat sig lift in number of active buyers and/or spend per customer over the 30-90 week period, then congrats: you probably have a winning offer. If not, go back to the drawing board and test another batch of offers.
If your target segment is too small to split into 4 sub-segments, your brand is probably too small to realize benefit from a loyalty program. Each sub-segment should contain 3-5k members minimum.
But if you want to run the test anyway, test one offer at a time, each for 30-90 days. Do not start the next offer test until the initial 30-90 days have elapsed.
Why do I recommend analyzing behavior over 30-90 days? Because you want to be sure that your incentive is really changing behavior. If you see a lift in week two, but that lift is wiped out by a sitewide sale you run in week 4, then the incentive isn’t doing anything. You’re just paying to give away margin using more technical complexity than a sitewide sale.
Side note: you do not need software to perform a proxy test on a points-based incentive. Tell customers they’ll receive $x in credit for every $y they spend during a date range. Then pull the totals yourself and send qualifying customers a custom promo code.
This is a pain to execute, but much less painful in the long run vs publicly committing to an unprofitable loyalty program.
Step 4: Vendor Selection
If your incentives are simple, you might not even need a vendor to scale them. Just start messaging them to the entirety of your target segment with a formalized name. You can consider messaging them publicly and letting everyone participate as long as there is some barrier to entry.
Software becomes useful in scenarios where you need to track customer LTV to qualify program participants and make that info accessible to members. You want to confirm a few things about each software platform you consider:
- Does it have the ability to carry out your program’s specific objectives? Or will they try to shoehorn you into some pre-existing points/incentive framework?
- How much will it cost per year?
- Are there any usage-based fees? Ex. Do you get charged more when more folks opt into the program?
- Does it integrate with your existing ESP and eCom platform?
You also want to understand how the costs of software might eat into the incremental revenue your program will generate.
Take the actual customer-level lift you drove in the test in step 3. Estimate how many times you’ll be able to drive that lift over 12 months. Then multiply that by the number of people in your target audience and discount that by 10-20% (to be safe).
This is your new baseline revenue lift. Subtract the annual cost of software. And subtract the cost of any new hire(s) you might need to manage the program. If this calculation produces a meaningful positive number, then proceed to step 5.
Note that I didn’t say “launch the program if the number is positive”. It’s not worthwhile to jump through these hoops to grow topline sales by 1% YoY. Go focus on customer acquisition instead.
Step 5: Launch Your Loyalty Program & Monitor Impact
It’s the moment we’ve all been waiting for: the launch. Because you’ve done the work to thoroughly validate your assumptions (right? RIGHT?), you can feel confident that the program will actually drive a lift in sales.
Even so, you’ll want to keep an eye on the following:
- Who is opting into the program vs your target segment.
- Who is active/redeeming rewards?
- Is the level of redemption roughly aligned with what you saw in the pilot?
- Is frequency and/or LTV increasing for your target segment (check after 30-90 days)?
- Has anyone figured out a way to scam the program and unfairly claim rewards?
After the program has been live for three to six months, you’ll want to see a lift in AOV or frequency for your target segment. Use a cohort analysis tool to do this.
You’ll also want to compare real opt-in rates for your target segment to the opt-in rates you forecasted in step 2 & 3. Low participation can tank expected sales lift. This might be a marketing problem, or your incentive might not be compelling at scale.
Do NOT compare the purchase behavior of program members vs non-members. This is a faulty analysis. Your most loyal, active customers are going to be more likely to join the program. So all you’re saying with this comp is “my best customers are better than my other customers”.
Make sure to check in on these metrics every quarter throughout the first year of the program. If the results don’t meet expectations, ask your software provider for advice. If their advice is “just do points bro”, drop them as soon as you can.
IMO it is better to do more private offer testing before launching a public program. Launching your loyalty program publicly is “burning your boats”. If you find out that the program is losing money 6 months in, changing or eliminating the program will anger your customers. And TBH, it will also negatively impact brand credibility.
Do not rush a loyalty program implementation. Following “best practices” will have you paying more to retain your worst customers–a money-losing formula.