Customer loyalty: what it actually costs, tier by tier, in 2026

What does a customer loyalty program cost in 2026?

As of August 2026, restaurant loyalty software runs between 79 and 1,200 USD per month, and the band where roughly 80% of independent operators actually live sits between 149 and 349 USD monthly.

The Guadalajara steakhouse that opened this piece paid 640 USD and showed 11,400 members on its dashboard, of whom barely 1,180 came back; the list price was the only accurate number in that operation. On top of the subscription you must add the redeemed discount, which in that case ate 12% of every bonused check, plus the floor time nobody invoices and everybody pays. With sector net margins of 3 to 9% according to Statista, giving away twelve points of ticket to inflate a redemption counter is mathematically worse than running no program at all. The license is 30% of the bill. The rest leaks out through the service door. Between 79 and 149 USD monthly you get spend-based points, a basic member dashboard and capture by kiosk or QR; that covers a single-site coffee shop and little else.

What each price tier includes, minus the fine print?

Move up to 149-349 USD and frequency segmentation appears, along with SMS or push messaging billed separately —worth remembering that 97% of SMS are read within the first fifteen minutes, per Tabular—, POS integration and cohort reports you can actually decide on.

From 349 to 640 USD you unlock a branded app, multi-site support, configurable reward rules and an account manager who answers inside 24 hours. Above 640 and up to 1,200 USD sits chain territory: open API, campaign attribution, delivery integration and annual contracts. Every tier jump sells features. None of them sells floor behavior, and floor behavior is what gets redeemed. Four variables explain almost the entire gap between what you sign and what you spend. First comes the billable member count: nearly every vendor charges per stored record, so a kiosk harvesting fake emails multiplies your invoice by 1.4 to 2.2 times in year two.

Four factors that move the real price, not the quoted one

Second is messaging, priced between 0.01 and 0.06 USD per SMS, which for an 8,000-member base with two sends a month adds 160 to 960 USD annually. Third, and by far the costliest, is the structural discount: shifting the reward from 15% to 8% of the check frees three to five points of gross margin outright. Fourth comes POS integration, ranging from 0 to 1,800 USD in setup depending on whether your terminal sits in the vendor's catalog. Get those four numbers in writing before you sign anything. Floor time is the invisible cost, and it can be calculated. Asking for an email with judgment, noting the guest's preference and logging the occasion takes two to three minutes per table; in a 240-cover-a-day venue with eight servers, that comes to roughly 20 hours a week already sitting in payroll, yielding about 29% valid captures when nobody trained the script.

The cost no vendor puts in the proposal

That same time, with a four-line script, pushes validity past 60%. It is the difference between 11,400 dead records and 3,000 live ones, and the live ones are worth more: 47% of loyalty members use their membership several times a month and 32% several times a week, according to LoyaltyPass. A record with no real name and no second visit is not a member. It is a line in a database you rent by the month. Negotiate three things, in this order: billable base, contract term, discount. On the base, demand billing only for members with at least one transaction in the last twelve months; a steakhouse with 11,400 records and 1,180 actives that applies that clause drops from 640 to around 210 USD monthly, and no vendor walks away over it. On term, swap the penalty-laden annual for a renewable quarterly and ask for the first two months at 50% against a valid-capture target; if the system works, the target hits itself.

How to negotiate the subscription and cut the real spend?

On discount, cut the reward to 8% of the check and compensate with recognition: the preferred table, the dish the chef is testing, the greeting that gets the surname right.

That costs 0 USD in food cost. Your vendor will never suggest it, because their dashboard measures redemptions. Diego F. Parra sets one rule at Masterestaurant that rewrites the arithmetic of the whole program: the metric you pay and reward is not the registered member but the VALID CAPTURE, defined as a real email, a noted preference and a second visit inside 90 days. When the floor gets paid for registrations, the floor inflates registrations; that is incentive design, not dishonesty. Suppose tomorrow your platform billed you only for members who came back: the 640 USD subscription would fall to roughly 66, your server would stop chasing emails at the door, and the 12% discount would land on the guest genuinely choosing between you and the place across the street.

Why the Masterestaurant method pays for valid captures?

You can build that scenario today without switching vendors, because the validity criterion belongs in your report, not in the seller's dashboard. No tool will teach your team how to ask.

Two truths here look like they collide, and they do not. A platform delivers data, and data is necessary: without a frequency log you have no idea that your Thursday regular has been missing for six weeks. Yet a dashboard holding 11,400 members changes nothing if Tuesday's server does not know that table 7 is celebrating the son's birthday, and that knowledge does not live in the cloud, it lives in a seven-minute pre-service ritual. The bridge between them is operational rather than technological: the data enters the shift briefing or it does not exist. For years I recommended the platform first and the training afterwards, and that order cost real money; today I flip it, with the floor script running two weeks before the software is switched on.

The tension almost nobody resolves: data against behavior

At net margins of 3 to 9% (Statista), the order of operations changes the result. Use this rule and save yourself three months of comparison shopping. Below 25,000 USD in monthly revenue, buy no platform: work the POS you already own, keep a sheet with name, phone and preference, and train the script; license spend is 0 USD and your 90-day return rate will tell you whether scaling is worth it. Between 25,000 and 80,000 USD a month, the 149-349 USD tier with one trained server beats any branded app, because 75% of traffic already happens off-premise according to Circana and your competitive edge sits precisely at the table. Above 80,000 USD with two or more locations, moving to 349-640 USD earns its keep through multi-site consolidation. Write down your definition of a valid capture today and count how many your team produced last week.

Where the two models genuinely part ways?

The traditional program bills by registered members, so the floor learns to inflate that number; the Masterestaurant method scores VALID captures, meaning a real email, a logged preference, and a second visit inside 90 days.

In a points model the discount is the currency, so 10-15% of every redeemed check comes out of gross margin and usually lands on the guest who was coming back anyway. The floor model pays in recognition instead: the preferred table, the dish the chef is testing, the greeting with the surname pronounced correctly. That costs zero additional food cost. Platforms deliver data; the Training Kit delivers BEHAVIOR. A dashboard with 11,400 members changes nothing if Tuesday's server does not know table 7 is celebrating a son's birthday, and closing that gap between data and behavior is exactly what the automated preshift handles in ninety seconds before doors open. Customer acquisition cost for a new restaurant guest sits at 18-34 USD through paid advertising in 2026; winning back someone already in your system costs under 2 USD.

Where the two models genuinely part ways — in practice?

When an owner asks me how to increase restaurant sales without raising ad spend, the answer lives in that order-of-magnitude gap. Traditional programs report redemptions, a metric that climbs when you give away more.

The MR method reports member frequency and member average check against non-members, the only two variables where customer loyalty touches the P&L.