A program you do not measure is a program you are running on faith, and faith is expensive. The good news is that loyalty is one of the few things in a restaurant that measures itself: the moment you enrol people, the numbers you need start collecting on their own. This last topic is the short list of what to watch, what each number is telling you, and where the line sits between reading your program and the deeper customer analysis that is its own course.
📈 The three numbers
You could track a dozen things. Three carry almost all the signal, and you can pull them from a punch-card count or a POS report without any special tools.
| Number | What it is | What it tells you |
| Repeat rate | The share of customers who come back at all | Whether the program is doing its one job: turning first visits into second ones |
| Redemption rate | The share of earned rewards that actually get claimed | Too low means the reward is out of reach; very high means you may be rewarding regulars |
| Incremental visits | Extra visits from members beyond what they did before | The real payoff — the visits the program actually caused |
Of the three, incremental visits is the one that answers topic 3's only question, and it is the hardest to see, because it requires knowing what members would have done without the program. You get at it by comparison: members against their own past, or members against similar customers who never joined.
🔍 Did frequency actually change?
The single most honest test of a loyalty program is whether people come more often than they used to. Everything else can look healthy while this stays flat — you can have thousands of members, a pile of stamps, and a busy-looking program that changed nobody's behaviour. So make this the number you defend.
Measure it simply. How often did a typical customer come back before the program, and how often now? If the frequency rose after launch, the program is earning its keep. If it is unchanged and all that happened is your existing regulars now collect rewards, you have confirmed the most common failure from the last topic — a discount wearing a loyalty badge — and you know to change the reward, not the software.
🧮 The numbers name the mistake
Every failure from the last topic leaves a mark in these figures, which is what makes measuring worth the trouble. A redemption rate near zero says the reward is too far away. A reward-earners list full of daily regulars says you are paying for visits you had. Flat frequency despite a full membership says the program is not changing behaviour at all. You do not fix a program by staring at it; you fix it by reading which number is wrong and changing the one rule behind it. Where this stops is deep segmentation — grouping customers by value and behaviour, predicting who churns, building targeted offers. That is real and it is worth doing, and it is the customer-data course later in this track. Here, three numbers and an honest before-and-after are enough to run a program well.
🧭 What you have built
Put the six topics together and you have a program, not a giveaway. You understood that bringing back a customer you have beats chasing a stranger (topic 1), picked the kind of program that rewards the behaviour you actually want (topic 2), and checked the math so it pays instead of leaks (topic 3). You built it to be joined without friction and heard at the register (topic 4), learned the expensive mistakes well enough to spot them in your own (topic 5), and now you measure it with three numbers and a frequency test rather than hope (topic 6). Finding new customers will always matter. This course was about the cheaper, quieter half most restaurants leave on the table: keeping the ones you already earned.