Most restaurant loyalty reports die in the inbox. They run to six pages, mix vanity totals with real signals, and nobody opens them twice. The fix is a short list. The restaurant loyalty program metrics worth tracking every week come to five: repeat-visit rate, lapsed guests, sign-ups per shift, share of orders direct, and share of sales tied to a known guest. This post defines each in one sentence, shows where the data comes from, and says what to do when a number moves. If you are still choosing a tool, start with the pillar on restaurant loyalty program software and own-your-list basics. Then come back here.
Why five numbers beat a dashboard
A loyalty dashboard shows everything the software can count: points issued, tiers, app opens, email clicks. Most of it does not change what you do on Tuesday. A weekly page for an owner needs a different test: if this number moves, is there a specific action someone takes this week?
Each of the five below passes that test. Each also comes from data you should already be able to pull from your POS (point of sale) system, your online ordering tool and your sign-up form. If you cannot pull one, that is the first finding in your report.
One honest limit up front. We are not giving you industry benchmarks for any of these. We did not find a current, sourced figure we could stand behind for a South Jersey restaurant, and a made-up "good" number would only mislead you. Compare each metric to your own last eight weeks instead. Direction and trend matter more than a target someone else set.
There is research behind the idea that return visits are worth watching. A Cornell University research recap from January 2020 describes a study of 80 full-service restaurants in one chain. Guests’ reported intention to return explained 55 percent of the variance in sales between units. That is one chain and one study, so treat it as a reason to measure returns, not as a rule for your dining room.
The five numbers, defined
1. Repeat-visit rate
Of the identified guests who visited in the last 90 days, the share who came two or more times in that window. “Identified” means the check is tied to a phone number, email or loyalty ID. Anonymous cash checks do not count here, which is why metric five exists.
Pick one window and keep it. Ninety days suits a full-service restaurant. A coffee counter might use 30. Changing the window every month makes the trend useless.
2. Lapsed guests
The count of guests who visited at least twice before and have not been back in a set number of days. Start with 60 days for a place people visit often and 120 for a special-occasion restaurant, then adjust once you see your own visit gaps.
Report the count and the change from last week, not a list of names. The names live in the system that sends the win-back message. Our post on welcome, win-back and birthday emails a restaurant can run without a marketing hire covers what that message looks like.
3. Sign-ups per shift
New loyalty or list sign-ups divided by checks (or covers) in that shift, shown by shift. A raw weekly total hides the problem. The rate shows that Friday dinner signs up guests and Sunday brunch does not, which tells you where to fix the ask: a table card, a prompt on the receipt, a line in the host script.
If you collect phone numbers for texting, the wording on the form matters. We cover consent language and record-keeping in the post on text and email sign-ups for restaurants. Run the final wording past your own lawyer.
4. Share of orders direct
Orders placed on your own website, app or phone line divided by all online and delivery orders, including DoorDash, Uber Eats and Grubhub. Count orders first. Dollars can be a second line, because marketplace baskets and direct baskets are often different sizes.
This number matters because a direct order arrives with a guest you can contact again. A marketplace order usually does not. The money side is worked through in direct online ordering versus DoorDash, the monthly math for one restaurant.
5. Share of sales tied to a known guest
Net sales on checks linked to an identified guest divided by total net sales. It is the quality check on the other four. If only a small share of checks are identified, repeat rate and lapsed counts describe a sliver of your business.
A low number here is not a failure; it is a to-do. It usually means staff are not asking, the sign-up is too long, or the POS is not attaching the guest to the check.
Where each number comes from
You do not need new software to start. You need four exports and a spreadsheet. The table shows the usual source and the first thing to check when it looks wrong.
| Number | Usual source | First check if it looks off |
|---|---|---|
| Repeat-visit rate | POS check history with guest ID | Are guests duplicated under two phone numbers or emails? |
| Lapsed guests | Same export, last-visit date per guest | Did the cutoff window change? |
| Sign-ups per shift | Sign-up form or loyalty tool, plus checks per shift | Was the table card or prompt missing that week? |
| Share of orders direct | Your ordering tool plus each marketplace’s order report | Are phone orders counted in the direct column? |
| Share of sales from known guests | POS sales summary and the identified-check export | Is the POS attaching guests to open tabs at the bar? |
The most common failure is the join. Your POS knows the check, your loyalty tool knows the guest, and your delivery apps know neither. If you cannot export guest-level records from each, you will be hand-matching rows every Monday. Our guest-data export audit walks through what each system will and will not give you, and it is worth doing before you build any report.
The fields you capture decide what you can report on later. Fewer is better, and the guest data fields worth capturing is a useful companion if your sign-up form has grown past three questions.
Laying out the one page
Keep the page to a single screen or a single printed sheet. For each of the five numbers, show four things and nothing else:
- This week’s value.
- The trailing four-week average, so one bad Tuesday does not set off alarms.
- The change versus the same week last year, where you have it.
- One sentence from the operator saying what moved and what they are doing about it.
That last line is the part owners actually read. A number without a sentence invites guessing. A sentence without a number invites arguing. Together they take about two minutes.
Put the page in the same place at the same time every week, such as Monday at 9 a.m. in a shared folder or an email to the owner. The habit matters more than the format.
Handle small numbers and seasons honestly
A single-location restaurant has small counts. A week with 40 identified guests can swing the repeat rate by a lot on a handful of people. Use the trailing average and say so on the page.
Seasons distort things too. A shore-town summer fills the room with guests who live elsewhere and will not return in 90 days. If that is you, report repeat-visit rate for local-zip guests separately, or treat the off-season weeks as your clean read. We do not have data showing how large this effect is for any specific shore town, so look at your own prior year before deciding.
What to do when a number moves
A metric is only useful if it has an owner and a default response. Write these down once so nobody debates them each week.
- Repeat-visit rate falls for three weeks: check service and menu changes first, then whether a welcome message stopped sending.
- Lapsed guests climb: confirm the win-back message is live, then read its send count. A climbing count with no sends is a broken automation, not a guest problem.
- Sign-ups per shift drop on one shift: look at who worked it, whether the card was on the table, and whether the form loaded on a phone in the dining room.
- Direct share stalls: check that the order link is on your Google Business Profile, your website and your receipts, and that it loads quickly on a phone.
- Known-guest share is low: shorten the sign-up and rehearse the one-line ask with staff.
Notice that three of the five responses start with “check that the thing is running.” A number that moves suddenly is often a broken connection rather than a change in guest behavior. Rule that out before you change the menu.
When a spreadsheet stops being enough
For one or two locations, a spreadsheet fed by weekly exports is fine. It stops being fine when the Monday rebuild takes more than an hour, when two people keep different copies, or when the numbers disagree with the POS.
At that point the report should build itself from one guest record. That is the idea behind the restaurant CRM we design for restaurants and bars, which connects reservations, guest profiles, follow-up and operator insights. Whether to build that or keep subscribing to a tool is a cost question we cover in build versus subscribe for a custom loyalty app and guest CRM.
Two cautions. A custom system does not fix a weak sign-up ask. And reports built on guest-level data should be reviewed by a person each week, because a duplicated record or a changed field name can skew a figure quietly. Automating the pull is sensible. Automating the judgment is not.
Frequently Asked Questions
What are the most important restaurant loyalty program metrics to track?
Start with five: repeat-visit rate, lapsed guests, sign-ups per shift, share of orders direct, and share of sales tied to a known guest. Each one links to a specific action, such as fixing the sign-up ask or triggering a win-back email. Points issued and app downloads are easy to count but rarely change what you do this week.
How often should an owner review loyalty numbers?
Weekly is enough for most single-location restaurants. Weekly catches a broken automation or a missing table card within days, while a trailing four-week average smooths out slow nights and weather. Daily checks mostly show noise at small volumes. Monthly reviews let a problem run for weeks before anyone notices.
What is a good repeat-visit rate for a restaurant?
We have not found a current, sourced benchmark we trust for independent restaurants, so we do not quote one. The more useful comparison is your own repeat-visit rate over the last eight to twelve weeks, using the same window and the same definition of an identified guest each time.
Can I track these numbers without loyalty software?
Yes, if your POS can export checks with a guest phone number or email. Pull the export weekly into a spreadsheet and calculate the five numbers by hand. The limit is the join: delivery marketplaces often do not share guest-level data, so share of orders direct will be a count, not a guest list.
Build the page before the spring rush
Pick your windows, pull the exports, and produce the first page this week. Then you will have a baseline before spring bookings arrive. If the Monday rebuild is already eating time, or your POS, ordering tool and sign-up form will not share a guest record, tell us how your front of house runs. We can scope a custom CRM around it, with hosting, third-party subscriptions and ongoing support priced separately in the proposal.
Sources
Every outside claim in this article links to where it came from. These open in a new tab.
- Cornell Universitybusiness.cornell.edu
- can swing the repeat rate by a lotwww2.census.gov