A loyalty program is not a feature you install. It is a liability you operate. Here is the routine that keeps it profitable instead of quietly eating your margin.
Your loyalty program minted a few million points last quarter. At a penny each, that is tens of thousands of dollars in discounts your customers can redeem whenever they feel like it, and most brands have no idea the number is that big until a finance review trips over it. Points are not a marketing feature. They are a liability sitting on your balance sheet, growing a little every time someone checks out.
Here is the trap. A loyalty program is the rare growth lever you can switch on in an afternoon. Smile.io, Yotpo, and Rivo all promise a live program by lunch, so most brands set 1 point per dollar at launch, wire up a referral code, and walk away. Eighteen months later nobody has looked at the earn rules once. A sitewide sale has been stacking points on already-discounted orders, a referral link is circulating on a coupon site, and no single person owns any of it.
This is the SOP for running the program, not installing it. The platform handles the mechanics: the points math, the referral tracking, the redemption widget at checkout. The SOP handles the decisions the platform will happily let you get wrong, and the monthly maintenance that keeps a loyalty program from draining the margin it was supposed to protect.
The upside is real and worth being precise about. Returning customers drive roughly 60 percent of DTC revenue, loyalty members carry 2 to 3 times the lifetime value of non-members, and a mature refer-a-friend program brings in 5 to 10 percent of revenue at the median, more at the top end. Referral acquisition costs run well below paid social because a friend's recommendation converts where an ad does not. Most brands that run a program well see it pay for itself inside a year.
The costs are just as real and almost nobody budgets for them. Outstanding points are deferred discount you have already promised. Referral programs are a standing invitation to fraud. And a generous earn rule that made sense on a $40 first order gives away margin when it runs on a $200 restock during a 20 percent-off weekend.
So this is a two-sided problem, not a set-and-forget one. Run the program too loose and you bleed margin and invite gaming. Run it too stingy, with a redemption threshold nobody reaches and earn rates that feel like a rounding error, and you get a dead program that still charges you a monthly platform fee for shelfware. The job of the SOP is to hold the middle: generous enough to change behavior, disciplined enough to stay profitable.
The platform is the machine. The SOP is the operating manual, and the two are not the same thing. Smile or Yotpo will calculate points perfectly and still let you run a promotion that mints a quarter's worth of liability in a weekend, because it does not know your margins. The decisions live outside the dashboard.
The usual failure is the loyalty hero. One person knows which promos exclude points, remembers that gift cards should not earn, and eyeballs the referral report for fraud when they think of it. Then they go on leave during BFCM, the busiest and riskiest week of the year, and nobody else knows the rules. The second failure is consistency: without written guardrails, every promo becomes a fresh argument about whether points should stack, and three people give three different answers.
A loyalty platform runs the mechanics. It does not make the decisions or catch the drift. The SOP is the small set of rules that keep the program profitable, plus the monthly review that keeps those rules true as your catalog and promo calendar change underneath them.
Start with what a point is worth, and hold it. A common setup is 1 point per dollar spent, with 100 points redeeming for $5, which values each point at five cents. The exact number matters less than picking one and knowing your real redemption cost as a percentage of the order it discounts. If you cannot state that percentage, you do not yet know whether the program makes money.
Then pick the earning actions that actually drive behavior, and stop there. Purchases, a first-time account creation, leaving a review, referring a friend, and a birthday reward cover most brands. Five well-chosen actions beat fifteen that dilute the currency and confuse the customer.
The part platforms let you skip is the guardrails, and the guardrails are where the margin lives. Write these down explicitly:
Every unredeemed point is a discount you have already handed out and not yet paid for. That balance belongs in a number you look at monthly, the same way you watch inventory. Some of it will never be redeemed, and that breakage works in your favor, but planning your economics around customers forgetting their rewards is not a strategy.
An expiration policy keeps the liability bounded without punishing your best customers. A clean version: points expire after 12 months of inactivity, and any new earning resets the clock. Active buyers never lose a thing, and dormant balances do not pile up forever.
A sitewide sale, points that earn on discounted orders, and a double-points weekend stacked together can mint a full quarter of points liability in three days. This is the single most expensive loyalty mistake, and it is invisible until redemption season. Your SOP decides which promos cap or suspend earning, in advance.
The step-by-step for configuring earn rules, redemption thresholds, and the exclusions that protect your margin.
Referrals are the highest-return part of the program and the easiest to abuse. The economics are straightforward: a give-and-get of $10 to $25 on each side is standard, and it works because the acquisition cost still lands far below paid social. The discipline is in the fraud rules, because a referral perk with no guardrails is just a public coupon waiting to be found.
Write the guardrails before you turn it on:
Attribution should run on unique links and codes the platform tracks, not a checkout survey asking how they heard about you. If you cannot tell which referral drove which order, you cannot pay out honestly or measure the channel.
Configure give-and-get rewards, fraud guardrails, and payout timing so the channel stays profitable.
The same referral playbook for brands running loyalty and reviews on Yotpo.
Tiers only work when the next one is reachable and the perk is worth chasing. A top tier that 2 percent of members will ever hit is decoration, not a lever. Set thresholds a real repeat customer can climb to in a year, and make each tier unlock something people actually want, early access and free shipping tend to beat another handful of points.
The program earns most of its money through the comms around it, not the widget on the account page. Loyalty-triggered flows, a points-balance nudge, a you-are-200-points-from-a-reward email, a tier-up congratulation, an expiring-points warning, earn several times the revenue per recipient of a standard promotional blast because they are timely and personal. Run them through Klaviyo and keep the cadence disciplined. A loyalty program that emails like a promo calendar trains people to unsubscribe from both.
Set earning actions, tier thresholds, and redemption options inside Yotpo without giving away margin.
Build the loyalty-triggered flows, points-balance nudges, and tier-up sends that make the program pay.
We turned the guardrails into a one-page pack: the earn-exclusion list, the referral fraud checks, and the monthly liability review. Hand it to whoever owns the program. Grab it from the Smile.io points program SOP linked above.
Enrollment is the vanity number every platform puts front and center, and it means almost nothing. Nobody's lifetime value went up because they clicked join. Measure the things that tell you whether the program changed behavior and paid for itself:
Loyalty rules are the fastest-rotting configuration in your marketing stack, because three separate things move underneath them. Every new SKU ships with no earn or exclusion rule attached. Every promotion is a fresh chance for points to stack somewhere you did not intend. Every price change shifts the reward-cost math that made your point value profitable in the first place.
The danger is that a stale loyalty rule does not throw an error. An earn rule you wrote last year, now running on a product line you sell at half the old margin, keeps minting points that lose money on every order, silently, while everyone keeps trusting the settings because they are written down somewhere. Documented and wrong is worse than undocumented, because it carries authority nobody thinks to question. That is drift, and a loyalty program is where it costs the most.
Review the liability and the referral report monthly. Review the rules themselves quarterly, and immediately after any sitewide promotion, catalog expansion, or price change. Then keep the whole thing in a living library instead of one person's memory.
The pillar behind the drift problem this SOP is built to survive: documented and wrong is worse than undocumented.
How this loyalty SOP fits into a maintained system of procedures instead of rotting in a shared drive.
Do not redesign the program. Start by pulling one number: your current points liability. It will almost certainly be larger than you expected, and that surprise is your baseline.
Then write three things down. Your point value and the earn-exclusion list. Your referral fraud guardrails. And the name of the one person who runs a monthly liability and fraud review. That single page protects more margin than any redesign, because it works the same whether you have 500 members or 50,000.
The hard part is not writing it. It is capturing what your best operator actually checks each month, the small rules they carry in their head, and keeping the doc honest as your catalog and promos shift. That is the exact problem we built ReccordSOP to solve. Record your loyalty manager running one monthly review end to end, and ReccordSOP turns it into a screenshot-by-screenshot SOP the rest of the team can follow. When your Smile or Yotpo settings drift from the doc, or a new promo breaks an earn rule, drift detection flags it before it costs you a season of margin.
Record your loyalty review once and generate the SOP free at reccordsop.com.
Yes, and arguably more than a large one, because a small team is where the loyalty hero problem bites hardest. If one person carries the earn rules and fraud checks in their head, the program breaks the moment they are out. A one-page SOP, point value, earn exclusions, referral guardrails, and a monthly review owner, is enough. You do not need a document for every screen.
Work backward from margin rather than copying a competitor. Whatever earn-to-redeem ratio you land on, the figure that decides whether the program is healthy is the discount each redemption represents as a share of the order it lands on, measured after product cost and shipping. Set the ceiling you can live with there, then choose a points value that respects it. A rate that looks generous on the storefront can still be unprofitable once you do that math.
Verify the reward against more than an email address, because a fresh inbox is free. Checking payment method, shipping address, and device catches most self-referrals. Hold the referrer's payout until the new order actually ships and survives the return window, limit how many referrals a single account can drive, and scan the report for the tell-tale sign of a code that has escaped onto a deal site. You want a reward that only fires on genuine new demand.
The absolute number matters less than the gap between members and everyone else, since that spread is the only figure that proves the program did something a plain discount could not. In practice, members repeating ten to fifteen points above the non-member baseline is a healthy sign, which often puts member repeat rates somewhere in the 25 to 40 percent band and higher for consumables. Report the two cohorts separately or a blended average will flatter you.
On two clocks. A short monthly pass over the liability balance and the referral report catches fraud and runaway points while they are still cheap to fix. A deeper look at the earn and redeem rules themselves belongs on a quarterly cycle, and you should pull that review forward any time a sitewide sale, a catalog expansion, or a price change lands, because those are the moments that turn a once-profitable rule into a leak.
I built ReccordSOP after watching too many DTC ops teams lose months to undocumented workflows. These SOPs are battle-tested with Shopify operators running $1M to $50M brands.
Last reviewed July 13, 2026
Most brands ask for reviews with a blast and a discount. The brands with ten times the reviews run it as a timed, repeatable process. Here is the SOP.
The cancel button isn't an exit. It's the last conversation you get, and most brands waste it on a blanket discount.
Most SOPs are wrong within 90 days of publishing. Here's how to detect it before it costs you a customer.
We use essential cookies for sign-in and a small amount of analytics to improve the product. Privacy policy.