A loyalty program has one job: to make the second and third order come sooner. It isn't there to build community or boost engagement. If your members don't order more often than everyone else, you've just put a friendly name on a cost line. What follows is nine mechanics we've watched work in real restaurants, plus an honest note on where each one falls apart.
The test every mechanic must pass
Three filters before the list. Can a tired customer understand the deal in five seconds? "10% back on every order" clears that bar; "earn 3 stars per $2 in qualifying categories" doesn't. Does the reward pull the next order, or only discount this one? And can you watch redemption week to week in your reports? A program you can't measure is one you'll never fix.
Mechanics 1–3: the workhorses
1. Percent cashback on every order. The customer gets 5–10% of each order as credit toward the next one. Everyone understands it, it applies to every ticket, and you only "pay" the reward when a fresh order comes in, so the redemption is itself revenue. The downside is emotional: credit feels less exciting than a free item, and it needs a wallet inside your app or ordering site to run smoothly. It's our default recommendation, and the margin math is in cashback vs. discounts.
2. Points per dollar. Earn 1 point per $1, redeem 100 points for $8 off. Flexible enough to run double-point days on slow weekdays. The catch is that the exchange rate hides the real value, so lighter spenders never reach a reward and stop caring. If you go with points, make the first reward reachable inside 3–4 average orders.
3. Digital punch card, buy 9 and get the 10th free. Perfect for a one-product shop (coffee, poke, burgers), where progress is visible and visible progress drives behavior. It's useless across a broad menu, though, and a 10-order finish line is too far for someone who orders once a month. Shorten it: buy 5, get a free side.
Mechanics 4–6: for the regulars
4. Spend tiers. Bronze at signup, Silver at $200 a quarter, Gold at $500, with each step raising the cashback percent or adding free delivery. Your top 10% of customers often carry 40%+ of revenue, and tiers give them a reason to route all their ordering through you. The risk is complexity, so stop at two or three tiers and never demote anyone mid-quarter without telling them.
5. Birthday reward. A free dessert or $7 in credit, good for the seven days around the date. Redemption runs among the highest of any campaign, because the occasion supplies the reason to order and birthday orders are rarely for one person. The limit is obvious: it fires once a year, so treat it as a layer on top of a program rather than the program itself.
6. Referral bonus. Give $5, get $5 in credit. It's the only mechanic here that brings in new customers instead of retaining old ones, and you only pay out on a completed first order. Watch for self-referral fraud; one reward per phone number and device covers most of it.
Cashback, tiers, referrals, and push campaigns run natively inside Dots apps and ordering sites.
Mechanics 7–9: the differentiators
7. Free-delivery subscription. $6.99/month, delivery fees waived. Once the fee is paid, subscribers stop comparison-shopping, and order frequency usually jumps hard in the first month. You need real delivery volume for the math to hold, and your own courier costs have to be under control before you start.
8. Order streaks and challenges. "Order 3 weekends in a row, get a free appetizer." Good for building a habit on a schedule you pick, and for propping up your weakest day. The danger is that one missed week resets the streak and can sour the whole thing, so keep the windows forgiving.
9. Surprise rewards. A random free item or a double-cashback drop lands in someone's account, announced by push. At small scale it costs almost nothing, it generates screenshots and word of mouth, and it gives you a reason to send a notification that isn't "please buy." Because it's unpredictable, it can't be the backbone, so use it as seasoning.
Count redemptions, not sign-ups
A program with 5,000 members and 2% redemption is worse than one with 800 members and 40%, because only the second is actually moving behavior. Most restaurant loyalty members go quiet within a few months of joining, usually because the first reward sat too far away. So audit it monthly: what share of active customers earned a reward, what share redeemed one, and whether the redeemers went on to order more often. Redemption under 20% means the reward is too far off, so lower the threshold before you touch the program.
Where the program should live
None of this survives on paper cards or a marketplace listing; the aggregators won't even hand you the customer's name. A loyalty program needs a channel you own: your branded app or ordering website, where every order writes itself into the customer's history. That's also where the follow-up lives. On our platform, marketing tools fire the "your cashback expires Friday" push on their own. Across 3M+ orders processed on Dots, branded apps convert up to 35% of visitors into buyers, and a loyalty program is a big reason people install them at all.
One honest warning before you go: don't launch three mechanics at once. Make cashback the base, run it for two months, and read the redemption numbers. Then add a birthday layer, then referrals. Any mechanic you bolt on before the last one is measured is a mechanic you'll never be able to judge.
Start this week. Pull your last 90 days of orders, count the customers who ordered exactly once, and price what a 10% credit on their first order would have cost you. That number is your program budget, and it's almost always smaller than what you already spend on discounts.
Your branded app and ordering site with a loyalty engine built in — shown on your own menu and numbers.