How to get repeat customers at your restaurant

Repeat customers are made at the first visit and counted at the second. Advice about them usually skips the counting, and without it there is no way to tell whether anything you tried actually worked.

Count it before you try to change it

"Repeat customers" sounds like a feeling, and it gets managed like one. It is a number, and your own records can produce it.

1. Pick a starting group. Every identifiable guest whose visit falls in one period, for example one calendar month. Identifiable means you know who they are: a reservation name, an online order, a card, a guest list entry.
2. Pick a return window and fix it in advance. For example, the ninety days after that month ends. The window should fit how often people normally eat with you, and it must stay the same every time you measure.
3. Count how many of the starting group came back inside the window. Divide that by the size of the group. That is your repeat rate for the period.

Split the result the way the second half of that table does. The overall rate mixes your regulars with people who ate with you once, and regulars keep it looking healthy while first-time guests quietly fail to return. The rate among first-time guests is usually the one worth working on.

Two mistakes that make the number lie

Matching guests loosely. The same person can appear as a reservation name, a different card and a delivery account. If those are counted as three guests, your repeat rate falls for reasons that have nothing to do with the guest. Match on the most reliable identifier you have and apply the same rule every time.

Changing the window. A rate measured over sixty days and one measured over one hundred and twenty are different numbers. Comparing them shows you a change that did not happen.

Why first-time guests do not come back

Sometimes it is the meal. Often nothing went wrong at all, and the guest simply never thought of you again. Three gaps explain a lot of it, and each one has a fix that does not involve advertising:

• No memory. A good meal fades quickly when there are other good places nearby. A short thank-you from the restaurant within a few days puts a name to the memory before it goes.
• No reason. "Come back soon" is not a reason. A dish arriving on the menu, a season changing, a night built around something specific: those are reasons, and they do not shrink the check the way a discount does.
• No invitation at the right moment. Contacting a guest the day after is too early to invite them back and too late months later. The right moment depends on how often that kind of guest eats out, which is covered in when a restaurant guest is actually lapsed.

If your recent reviews point at a real problem with food, service or price, fix that first. Inviting people back to an experience that let them down reminds them why they stopped.

Recognition beats points

Being known is the thing a regular gets that a stranger does not: the host who remembers the table they like, the server who knows they skip dessert, the owner who says hello. Guests notice it, and a points balance cannot imitate it.

Recognition can be designed rather than left to whoever happens to be on shift:

• Notes on the guest record. Many reservation systems let staff record a preference, an allergy or an occasion. A note written once can be read by every shift after.
• A pre-shift look at who is booked. Two minutes before service to point out returning guests, and anything worth remembering about them, changes how the first minute at the table feels.
• A small unannounced gesture. Something brought to the table because the guest is back, not because they reached a threshold. A reward someone earned is a transaction. A gesture nobody asked for is hospitality.

What a loyalty program can and cannot do

Points and punch cards are filled fastest by guests who were coming every week anyway. That makes them good at thanking regulars and weak at creating new ones, because much of what they give away pays for visits that would have happened without them.

That does not make a program useless. It makes it something to measure: compare how often members and similar non-members return, or introduce the program to part of your guests first. If the difference does not cover what the program gives away, it is a cost dressed as a strategy. The method for measuring any offer this way is on how to tell if a win-back campaign actually made money.

Measure again, the same way

Once something changes, run the same count over the same window and compare it with the period before. For anything you send to guests, such as a thank-you or a second-visit offer, go further: hold back a random slice of first-time guests and compare the two groups, because the season alone can move a repeat rate.

Keep what moved the number. Drop what only felt like it should have.

Questions owners actually ask