Loyalty program strategy should be gated by revenue stage. Under $3M you build no program at all, just a reorder reminder timed to when the bottle runs out, a one-click skip on subscriptions, a reorder link that skips the login, and email answered by a person within a day. From $3M to $10M a structured tier or membership starts paying for itself, with earn rules tied to behaviors that predict retention and redemption capped to protect contribution margin. Past $10M you build membership with utility beyond discounts, partner and referral mechanics, and segment-specific programs run by someone whose actual job it is.
The failure mode I see most often is a brand at $2.4M installing a points platform because retention feels soft. A year later they’re carrying an unmeasured discount liability, paying $400 to $1,500 a month for it, and their 60-day repeat rate hasn’t moved. I’ve run this sequencing inside brands from $1M through $60M+, and the ones that got value built the program at the stage where the mechanics had something to multiply. For campaign-level tactics, customer loyalty campaigns for supplement brands covers what to send and when.
$1M to $3M: build the reorder, skip the platform
Your customer count is too small for points balances to reach anything a person wants, and your team is too small to maintain rules. Every hour configuring earn tiers is an hour not spent on the reorder mechanics that decide whether there’s a customer to reward.
Build these four things:
- A reorder prompt at 70 to 75% of actual supply duration. Count capsules and dose, not calendar months. A 60-count bottle at two a day is a 30-day supply, so a day-30 email lands three days late.
- Subscribe and save at 15 to 20%, with skip and delay one click deep. Hiding the skip button converts a delay into a cancellation.
- A no-login reorder link in every email. Password resets kill more reorders than price does.
- A human on the inbox with a same-day reply target. At $2M that’s an advantage you’ll lose later, so use it now.
Refuse these: a points platform, a tier structure, two-sided referral rewards, a branded rewards page, anything with a per-order transaction fee. None earn back their setup cost below $3M.
The metric that gates the next stage: 60-day repeat rate, by first-purchase cohort. In health and wellness DTC I look for 15 to 25% at 60 days as workable and above 28% as strong. Under 12% is a product, expectation, or usage-compliance problem, and a program layered on top reports enrollment numbers while the business keeps leaking. The cohort table is in the ecommerce metrics that predict scale.
$3M to $10M: the first structure that pays for itself
You now have enough order volume that a percentage point of retention is real money. Two decisions here matter more than the platform you pick.
Tie earning to behavior
Points per dollar spent is a delayed discount with extra steps. Every customer earns it, including the ones buying anyway, who are most of your revenue.
Tie earning to the behaviors that predict a customer sticking around:
- Reaching shipment three on a subscription
- Completing a usage check-in at day 21
- Writing a review with a photo
- Adding a second SKU to an existing routine
- Referring someone who converts and reorders
Each is something you want more of and something the customer has to choose. Spend-based earning pays for a choice already made.
Design redemption so it can’t eat your margin
Point cash value: Reward dollar value ÷ points required to claim it
Issuance rate: Point cash value × points issued per dollar of revenue
The common default of 1 point per dollar with 100 points redeeming for $5 puts point cash value at $0.05 and issuance at 5% of revenue. On a $70 order carrying $20 of contribution margin, that 5% is $3.50, or 17.5% of the margin funding your entire business. So a 5% effective redemption rate costs you a fifth of your operating room.
Point liability: Outstanding unredeemed points × point cash value × expected redemption rate
Run that monthly where your bookkeeper sees it. Four caps bound the exposure:
- Expire points after 12 months of inactivity, disclosed at signup
- Exclude already-discounted subscription orders from earning
- Cap redemption at 20% of any single order’s value
- Cap earn per order so a bulk buyer can’t bank a free case
Rewards paid in product cost you COGS rather than revenue. A $5 credit costs $5; a trial size retailing at $12 might cost $2.80 and put a new SKU into the routine.
$10M to $20M and up: membership with utility
Above $10M, discount-based loyalty stops interesting your best customers. They’ve been buying for two years, they aren’t price sensitive, and 5% back is noise. Access is what they’ll notice.
- Early access to new formulations, with a window that means something. Members get 72 hours before anyone else sees the launch.
- Formulation input. Members vote on the next flavor or SKU and you tell them the result. The response data is product research you’d otherwise pay for.
- Education with a real practitioner attached. Dr. Gabrielle Lyon built a $6M practice on an audience over a million people because access to the expertise was the product and the supplements followed it.
- Community, if you have an identity worth belonging to. Most brands don’t.
- Partner benefits. Trade value with non-competing brands your customer already buys from.
- Segment-specific programs. Practitioners and high-frequency multi-SKU buyers need different mechanics from a single-SKU customer on autoship.
The part nobody budgets for is the operating cost. A program at this scale runs 0.5 to 1 FTE once you count rule maintenance, redemption tickets, tier disputes, partner coordination, and the liability review. Without a named owner it degrades into a rewards page nobody has opened since launch. Ancestral Supplements went from $28M to $60M+ on ownership like that being explicit rather than assumed.
Which structure fits which stage
| Structure | What it rewards | Stage it fits | Margin impact | Operational load | How it fails |
|---|---|---|---|---|---|
| Points program | Cumulative spend | $5M+, 4 or more SKUs | 3 to 6% of revenue at typical issuance | Platform fee plus monthly liability review | Becomes a delayed sitewide discount nobody responds to |
| Tiers | Sustained behavior over 12 months | $3M+ | 1 to 3%, mostly paid in perks | Moderate; rules need upkeep | Top tier unreachable, or everyone lands in it and it signals nothing |
| Paid membership | Upfront commitment | $10M+, benefits worth paying for | Positive if the fee covers delivery | High; every benefit is a promise | Benefits are just discounts, so you sold a coupon book |
| Subscription as loyalty | Reorder consistency | $1M+ | 15 to 20% off subscription orders | Low; already in your stack | Cadence mismatched to consumption, so churn hides in it |
Subscription is the highest-return structure at every stage under $10M, and most brands treat it as a checkout option rather than a program. Benchmarks: subscription retention benchmarks for supplements.
The four numbers that tell you whether it’s real
Effective redemption rate: Points redeemed in period ÷ points issued in period
Point liability: Outstanding unredeemed points × point cash value × expected redemption rate
Incremental program margin: (Contribution margin per member − contribution margin per holdout customer) × number of members
Net program contribution: Incremental program margin − rewards redeemed by customers who would have repurchased anyway − platform fees − loaded staff cost to run it
That last one decides whether the program deserves to exist, and almost nobody calculates it, because the middle term requires knowing what would have happened anyway.
How do you know the program caused anything?
Loyalty platforms report enrollment, points issued, and redemption rate. All three rise whether or not the program works.
The trap is comparing enrolled members to non-members. Members self-selected, meaning they were already your most engaged customers before joining. That comparison shows a 40 to 60% lift every time, at every brand, including brands whose program does nothing.
A holdout group is the only credible answer, and you can run one without a data team.
Pick a rule that’s effectively random. Customer ID ending in 7 gets excluded from enrollment and from every program email. Roughly 10% of customers, assigned by something with no relationship to buying behavior.
Hold it for 90 days minimum, 180 if your purchase cycle is long. Anything shorter measures noise.
Compare one number: cumulative contribution margin per customer, holdout versus enrolled, over the same window from the same starting cohort. Not revenue, not orders.
Size it honestly. Under about 2,000 customers per arm, a difference below 8 to 10% isn’t distinguishable from randomness. Run it anyway and treat a small lift as “no evidence yet” rather than a win.
Write down what result would make you kill the program before you look at the data. That commitment is the whole discipline.
Most first honest holdouts land between slightly negative and modestly positive. Useful either way, and it costs one email segment.
When to kill a loyalty program
Kill it when the holdout shows no incremental margin after two full quarters, when redemption sits under 15% of issued points for a year, when fewer than a quarter of repeat customers engage with it, or when whoever maintains it can’t name the behavior it changes.
Winding one down without a revolt takes 90 days and four moves:
- Stop issuing new points immediately, quietly. No announcement yet.
- Email balance holders with a deadline and a conversion better than what their points were worth. Someone sitting on $6 of points gets a $10 credit or a free product. Costs less than another year of platform fees and lands as a gift.
- Honor every outstanding balance for 90 days. Publish the date and don’t shorten it later.
- Move your top tier onto something with utility. Permanent free shipping, early access, a direct line to support. Your top 200 customers are the only people who’ll notice the program is gone, and they should end up better off than they started. Complaints come almost entirely from that group, which is what step two is for.
Common questions
When should I launch a loyalty program? Not before $3M in revenue and not before your 60-day repeat rate clears roughly 25% by cohort. Below either threshold, the same effort spent on reorder timing, subscription cadence, and onboarding returns considerably more.
Are points or tiers better? Tiers, for most health and wellness brands under $10M. Tiers reward sustained behavior and pay out in perks that cost you less than their perceived value. Points reward spend every customer was already doing, and create a cash liability you carry on the books.
Should I charge for membership? Only above roughly $10M, and only if the benefits stand on their own without a discount attached. A paid membership whose main value is a percentage off is a coupon book, and customers price it accurately. Early access, education, and practitioner support can justify a fee.
How much does a loyalty program cost to run? Platform fees run $400 to $1,500 a month at $3M to $10M in revenue, and $2,000 to $6,000 above that. Redemption is the larger cost at 3 to 6% of revenue, plus 0.5 to 1 FTE once the program has tiers, partners, and support volume.
How do I know if my loyalty program is working? Run a holdout. Exclude about 10% of customers by a rule unrelated to their behavior, keep them out for at least 90 days, and compare cumulative contribution margin per customer against enrolled members. Comparing members to non-members without a holdout always flatters the program, because members self-selected as your best customers before they joined.
If you’re at $3M to $10M staring at a loyalty platform demo and unsure whether it’s the right buy this quarter, that’s a 30-minute conversation and a cohort table away from an answer. Here’s how engagements work, and the button below books the time.