Restaurant loyalty gets judged by the coupon a guest redeems, and that is the wrong moment to watch. A redeemed coupon proves that an offer reached someone who was already on her way in. The moment that actually tells an operator something is quieter and easier to miss: the visit that should have happened and did not.
Take a regular who orders the same breakfast burrito at 7:40 every weekday, from the drive-thru on her way to the train. In March her office moves to a four-day week, so the Friday visit stops. A few weeks later the Wednesday visit goes too, because on busy mornings a delivery membership now brings breakfast to her desk. Through all of it, the coupons keep arriving for the mornings she still drives through.
Nothing in the loyalty report flags her. Her account is active, her coupons get redeemed, and every dashboard calls her loyal. What the report cannot see is that she is a regular in the middle of leaving, and that the program is discounting the last of her visits.
Restaurant loyalty is a habit business. Too many programs still run it like a coupon book.
The loyalty report is watching the wrong moment
McDonald’s already says, in its own words, that the habit is the prize. It describes its goal as becoming “the first choice for more customers, more often.” That is a goal about visits, and it sits awkwardly beside a loyalty report built around redemptions.
Visits are harder to count than they look. McDonald’s reports members who were active at some point in the past 90 days. That kind of measure, an activity window, answers one question well: does this guest still belong to the program? It cannot answer a second one: is her rhythm breaking? A weekly guest who stops coming stays inside a 90-day window for a full quarter, so roughly a dozen visits can go missing before the headline number moves. By the time it does, the restaurant is no longer having a retention conversation. It is writing a win-back offer.
There is a second blind spot, and it sits at the counter. A program can only see the guests who identify themselves, and many guests who walk in never do. Chipotle links nearly nine in ten of its digital orders to its Rewards program, yet only about one in five of its in-restaurant transactions. The counter, where so many habits form, is exactly where the record goes quiet.
Put those two gaps together and an uncomfortable picture emerges. A chain’s most dependable regulars are often the guests its loyalty report understands least.
A coupon aimed at a habit is a price cut
A discount does its job when it reaches a guest who is still deciding where to eat. Aim the same discount at a guest who has already decided, and it becomes a price cut on an order that was coming anyway. That price cut comes out of a margin that was thin to begin with: more than four in ten operators told the National Restaurant Association that their restaurants were not profitable in 2025.
Loyalty Juggernaut separates two kinds of program on exactly this line. Transactional loyalty gives points for purchases and tries to make the guest buy more. Behavioral loyalty rewards what a customer does, and a missed Friday is precisely the behavior it is built to read.
Rewards can also grow faster than the visits they are meant to buy. When a guest earns a reward, the restaurant holds back part of that sale’s revenue until she uses it, because it still owes her the free item. In the quarter Chipotle relaunched its program with richer benefits, the revenue it held back this way rose by about a third on the year. Over the same quarter, comparable transactions rose 1.0%. The filings cannot say how much of that extra reward bought new visits and how much landed on visits that would have happened regardless. The points liability accounting guide explains the accounting behind that held-back revenue.
There is a simple way to find out, and it is the most useful habit a loyalty team can build. Keep a small, randomly chosen group of members off each offer, then compare their visits with everyone else’s over the same weeks. The difference is what the offer actually caused. That group is a holdout, and it turns an argument about whether a promotion worked into a number.
Taking a discount away carries its own risk. In 2025 Sweetgreen moved members off its paid Sweetpass+ plan, which its filings tie to higher discounts, and onto a free points program. Traffic at its established restaurants then fell, and the company’s annual report names the switch as one reason, alongside a more selective consumer. The filing suggests that some of those discounted visits were real, and that they left when the discount did. The question was never whether to discount. It is which visits each reward is buying.
A guest’s rhythm is worth planning around
Once the habit is treated as the asset, loyalty stops being a calendar of promotions and becomes something a finance team can plan around. A regular’s rhythm is revenue that can be forecast with some confidence, and every reward issued against it is a cost the business chose to carry. The promotions team tends to watch redemptions. The profit and loss statement watches traffic and margin, and those are the numbers that decide whether the program earns its keep. So the monthly loyalty report should lead with three of them: the share of visits tied to a member ID, the share of regulars visiting below their usual rhythm, and the visits each offer added against its holdout. Redemptions belong underneath, as the cost of getting there. Reported that way, the program shows up as what it should be: an enterprise growth engine measured in visits kept, rather than a coupon budget measured in coupons redeemed.
The delivery companies understood this early, and they now compete for the same weekday habit. Uber says members of Uber One bring in more than 70% of its delivery bookings, and DoorDash says its DashPass membership typically raises how often members order. Each of those memberships is a loyalty program in its own right, run by the app that carries the restaurant’s food to the guest’s door, and each one teaches her to open someone else’s app first.
The next shift is in who takes the order at all. Taco Bell now takes drive-thru orders by voice AI in hundreds of restaurants, and DoorDash has put an ordering app inside ChatGPT. The drive-thru is still the brand’s own lane, but a voice agent that takes the order without attaching the member ID leaves that visit off the guest’s record. The kitchen still cooks the meal. Whoever takes the order keeps the habit.
Five practices turn a coupon book into a habit program
- Know each guest’s rhythm. The hour she orders, the days she comes in, what she orders, how she orders, and how long it has been since her last visit all belong on her record, because every later decision starts from that baseline.
- Treat the missed visit as the trigger. A weekly guest who skips twice is sending a signal long before a 90-day window would drop her from the count. Reaching out at that point costs a message. Waiting until she has gone costs a discount.
- Hold rewards back from the orders that would come anyway. A standing discount on a steady habit is a price cut, so spend rewards where a guest’s behavior can still change, and prove each one against a holdout. The loyalty program ROI guide walks through the method step by step.
- Put a member ID on the in-restaurant visit. A scan at the counter, a linked card, or a photo of the receipt turns an anonymous meal into a visit on her record. Chipotle answered its own gap with an in-restaurant sign-up campaign and incentives for its crews.
- Price any paid layer against the habit it buys. A paid plan suits guests who already come several times a week, as Uber One shows at scale. A plan sold mainly on discounts can lose its members when the discounts end, which is the lesson Sweetgreen’s filing suggests.
In GRAVTY®, a missed visit can start the next move
On GRAVTY®, the breakfast regular’s routine becomes data the brand owns. Patented multi-dimensional behavior tracking records the hour she orders, the lane she uses, the burrito she repeats, and the weekdays she comes in.
When the Friday visit stops, a rule built in GRAVTY Visual Rules, the patented no-code rules engine, can catch it. The rule watches for a regular whose weekly visits fall below her usual count, and it can fire weeks before a 90-day window would register anything. The same engine can keep the standing coupon off the mornings she still comes in, so the program stops paying for visits it already has.
A flag is the start of a question, and Agentic AI Compass is where the team asks it. The team can ask why visits fell among weekday breakfast regulars and simulate a win-back offer’s outcome for that group before the offer runs. A holdout in the campaign then shows what the offer caused. On the day she pays at the window without the app, GRAVTY® AI-Scan can read a photo of her receipt and put the visit on her record. The offer she eventually receives is shaped for her by GRAVTY’s patented mass individualization.
The guest should feel remembered
The regular herself does not want to be managed, and she would not describe herself as a loyalty member. She thinks of the drive-thru as her place, and what she wants is for her place to notice that her week has changed. The best version of this program greets her on a Thursday with something new from the morning menu, at the moment her routine is being rewritten. The worst version sends the same burrito coupon it has sent her all year.
A redeemed coupon tells a restaurant that a guest came back. A missing visit tells it she is deciding whether to.
Points reward the visit she made. Attention earns the next one.
The numbers behind restaurant loyalty
These are the figures behind this essay. Each comes from a company’s own filing, terms page, or release, or from the National Restaurant Association’s annual survey.
Program types and the risk in each
Each common program design suits a different visit pattern, and each carries a risk the operator has to manage.
| Type | How it works | Where it fits the business | The risk to the operator |
|---|---|---|---|
| Points | Guests earn per dollar and trade points for food | Brands with bigger, more variable checks | Pays heavy spenders for orders they would place anyway |
| Visits | A free item after a set number of orders | Daily habits like coffee and lunch | Counts visits and misses the gaps between them |
| Tiers | Status and faster earning past a yearly threshold | Recognizing the top regulars | Moves the few guests near the next threshold |
| Paid | A monthly or yearly fee buys perks or free delivery | Guests who already come several times a week | Members who joined for a discount can leave when it ends, as Sweetgreen found in 2025 |
| App-first | Rewards live inside the ordering app | Brands whose orders already run through the app | Misses in-restaurant guests who never open the app |
| Cash back | A share of each check returns as credit | Brands competing on everyday value | Works like a standing discount on every order |
Eight restaurant programs, benchmarked by design and disclosure
What eight chains chose to build, and what each one tells investors about the result. Every cell comes from the brand’s own terms page, filing, or newsroom. The last column is the one to copy: a program that cannot say how many members are active has no baseline for the habit.
| Program | Earn design | Status design | Paid layer | What it discloses | Source |
|---|---|---|---|---|---|
| McDonald’s loyalty | A loyalty program and personalized offers in the app, across 70 markets | None stated | None stated | Nearly 220 million active in the past 90 days, June 2026 | Q2 release, Aug 2026; 10-K, Feb 2026 |
| MyPanera | 10 points per $1 | MyPanera; MyPanera+ at $300 spent in a calendar year; Sip Club | MyPanera+ Sip Club subscription, up to 30 drinks a month | More than 70 million members, no activity window stated, Aug 2026 | Relaunch release, Aug 2026 |
| Chipotle Rewards | Points on spend, in the app or at the counter; challenges add points | None in the terms | None in the terms | More than 21 million active, Apr 2026 | Terms, Apr 2026; relaunch release, Apr 2026 |
| Starbucks Rewards | 1 Star per $1 at Green, 1.2 at Gold, 1.7 at Reserve | Green; Gold at 500 Stars in 12 months; Reserve at 2,500; relaunched March 2026 | None in the terms | No count in its three 2026 quarterly releases | Terms, Mar 2026; FQ2 release, Apr 2026 |
| Chick-fil-A One | 10 points per $1 on qualifying orders, rising as members climb tiers | Member; Silver at 1,000 points; Red at 4,000; Signature at 10,000 | Free to join | Not disclosed | Chick-fil-A One page, Sep 2026; support page, Sep 2026 |
| Dutch Rewards | 3 points per $1, in the app only; 250 points for a free medium drink | None stated | None stated | No count; Rewards transactions were 73% of the total in Q2 2026 | Rewards page, Sep 2026; Q2 release, Aug 2026 |
| Domino’s Rewards | Points on qualifying orders, redeemed for menu items | None stated | None stated | Not disclosed | 10-K, Feb 2026 |
| Club Wingstop | Points redeemable on future orders, with points sharing | None stated | None stated | No count; its first loyalty program, launched May 2026 | Launch release, May 2026; 10-Q, Jul 2026 |
Of the eight, only McDonald’s and Chipotle publish an active count, and only McDonald’s names its window: 90 days. A total with no window is a sign-up count. It says nothing about who still comes in.
What the filings and surveys show
| What it shows | The figure | Source |
|---|---|---|
| Member sales rise while visits fall | McDonald’s sales to loyalty members rose more than 20% to $40 billion over twelve months; U.S. comparable sales rose 0.8% in Q2 2026, held back by fewer guest visits | Q2 release |
| The dining room is hard to see | Chipotle links nearly 90% of digital transactions to Rewards, about 20% of in-restaurant ones | Relaunch release |
| Rewards can outgrow visits | Chipotle deferred $60.2 million for rewards earned in Q2 2026, up from $45.7 million; revenue rose 9.3%, mainly from new restaurants; comparable transactions rose 1.0% | 10-Q; Q2 release |
| Removing a discount has a cost | Sweetgreen tied Sweetpass+ to higher discounts; established-restaurant traffic fell 10.4% in 2025, partly on the switch to SG Rewards | 10-K, FY2025 |
| Delivery apps sell frequency | Uber One: 50 million members, May 2026; more than 70% of delivery gross bookings in Q2 2026. DoorDash: more than 35 million members at the end of 2025, trials included; DashPass “typically drives an increase in average consumer order frequency” | Uber Q1; Uber Q2; DoorDash Q4; DoorDash Q2 |
| Marketplace orders cost margin | Domino’s warns that marketplace orders risk lower store-level profit than its own channels | 10-K |
| AI takes the order | Taco Bell’s voice AI was live in more than 890 U.S. restaurants across 38 states in July 2026, per its provider. DoorDash opened an app inside ChatGPT in December 2025 | Provider; DoorDash |
| Guests and operators | Loyalty membership matters to 70% of delivery customers and 56% of sit-down diners, a deal to 81% and 70%; 75% of delivery customers used a third-party service in six months, and 56% would rather order direct; 46% would order while talking to an AI chatbot; 42% of operators were not profitable in 2025 | NRA 2026 report |
How an independent restaurant can start
- Match the type to the visit pattern. Daily coffee and lunch suit visit rewards. Bigger checks suit points.
- Put the member ID on every channel. Pick software that ties the point of sale to online ordering, so every visit lands on one record.
- Price the reward from the margin. Set it against food cost, at a threshold a regular reaches in a few weeks.
- Keep the earning in direct orders. Chipotle’s new-member reward excludes orders placed through third-party delivery platforms.
- Train the counter to ask. One question at the register reaches guests who never open the app.
- Hold back a test group from day one. Keep a random slice of members off each offer and compare their visits. The loyalty program design guide covers the full sequence.
Frequently asked questions
What is the best loyalty program for restaurants?
For a restaurant brand, the best loyalty program is the one that ties the most visits to a known guest and can prove, against a holdout, that it adds visits. The public benchmarks show how far apart brands are. McDonald’s counts nearly 220 million members active in the past 90 days across 70 markets. Dutch Bros runs 73% of its transactions through Dutch Rewards. Chipotle links nearly 90% of digital orders to Rewards but only about 20% of in-restaurant ones, which is where most programs still lose sight of the guest.
What are some examples of loyalty programs for restaurants?
The restaurant programs worth studying are McDonald’s, Chipotle Rewards, Starbucks Rewards, Chick-fil-A One, Dutch Rewards, SG Rewards, and Club Wingstop, each for a different design decision:
- McDonald’s: reports members active in the past 90 days, a reach measure that lags a guest who is drifting away.
- Chipotle Rewards: relaunched in April 2026, keeping points alive with one qualifying purchase a year.
- Starbucks Rewards: three levels, Green, Gold, and Reserve, so earning rises with status.
- Chick-fil-A One: four tiers, from Member to Signature, built to recognize the most frequent guests.
- Dutch Rewards: app only, so every Rewards transaction carries a member ID.
- SG Rewards: Sweetgreen’s 2025 switch from the paid Sweetpass+ to free points, which its annual report names as one reason for lower traffic.
- Club Wingstop: Wingstop’s first loyalty program, launched in May 2026, with points members can share.
The best restaurant loyalty programs guide profiles more of them.
How do restaurant loyalty programs work?
A restaurant loyalty program trades a reward for an identified visit. The guest shows a member ID in the app or at the counter, earns points, visit credits, or status on the order, and redeems them later for food or perks. The restaurant gets a record of who comes back and how often, which is the data it needs to spot a regular who is slipping. It also carries every unredeemed reward as a liability until the guest uses it, which the points liability accounting guide explains.
How many restaurants have a loyalty program?
About half of US restaurants have a loyalty program. In the National Restaurant Association’s 2025 report, 52% of operators said they run one. The share was 66% at limited-service restaurants and 38% at full-service ones. Coffee and snack shops led at 80%.
Do restaurant loyalty programs increase visits?
Yes, by operators’ own account: 70% of restaurants with a program said it helped boost customer traffic in 2024, per the National Restaurant Association’s 2025 report. That figure is self-reported. McDonald’s shows why it needs a test. Its sales to loyalty members rose more than 20% to $40 billion over twelve months, while US guest counts fell in the second quarter of 2026. A holdout settles it. Keep random members out of an offer and count only the gap in visits.
Should a restaurant charge for its loyalty program?
Yes, when the fee buys something a frequent guest uses every week, such as free delivery or a daily drink. Uber One members made up more than 70% of Uber’s delivery gross bookings in the second quarter of 2026. The risk comes when the paid perk goes away. Sweetgreen tied part of a 10.4% traffic drop in 2025 to its switch from Sweetpass+ to free SG Rewards. The paid loyalty programs guide covers the fee math.
