Before you launch a restaurant loyalty program, record four things for a defined period: the dates, the number of identifiable eligible guests, their eligible visits, and the revenue from those visits. Write down the rules behind each number. Use the same rules after launch so you can see what changed without mistaking every change for an effect of loyalty.
Add a short operating log, too. A holiday, a new ordering channel, or a price increase can move the numbers even if the loyalty offer has no effect. Your baseline is a starting point for a decision, not proof that the program caused a result.
Choose periods you can compare
Pick a pre-launch period long enough to represent normal service and a post-launch period with the same number of weekdays and weekends. Four weeks before and four weeks after can be a practical starting point for a restaurant with regular traffic. If your business is highly seasonal or guests visit less often, you may need a longer view.
Mark the launch date and decide what to do with a partial launch week. For a clean first comparison, you might leave that week out of both periods and record it separately. Do not quietly include a promotion-heavy opening weekend in one period while leaving an equivalent event out of the other.
Use the following hypothetical worksheet to see how the pieces fit. It assumes the restaurant can identify guests consistently across the included orders in both periods. The figures are invented and are not a claim about any software product.
| Measure | Four weeks before | Four weeks after | Counting rule |
|---|---|---|---|
| Identified eligible guests | 100 | 110 | Unique guests with at least one eligible visit |
| Eligible visits | 150 | 176 | Completed qualifying orders tied to those guests |
| Eligible visit revenue | $3,000 | $3,696 | Net food and drink sales on those orders, after discounts |
| Visits per identified guest | 1.50 | 1.60 | Eligible visits divided by identified eligible guests |
| Revenue per eligible visit | $20 | $21 | Eligible visit revenue divided by eligible visits |
In this example, visits per identified guest rise by 0.10, and revenue per visit rises by $1. The guest counts are unique within each period; they do not mean the same people appeared in both periods. The revenue figures exclude tips, delivery charges, and tax. They are sales measures, not profit, and do not include the cost of running the program.
Save the underlying report or export along with its date range and filters. A total written on a sticky note is hard to reproduce when you return to it a month later. Ask any vendor whether you can export the relevant records and whether reports can be filtered to match your chosen dates and channels.
Define who and what counts before launch
An eligible guest is not automatically a loyalty member. For the baseline, decide which guests you can identify using the same method before and after launch. That might be a customer profile connected to an order. If anonymous counter guests cannot be matched to a profile, exclude them from the identified-guest measures and note that the results cover only part of your business.
Keep the definition separate from enrollment. Comparing all identifiable guests before launch with only loyalty members afterward would change the population as well as the period. Record enrollment separately if you want to track adoption. Access to a guest's contact details also does not, by itself, establish permission to send marketing messages.
Next, define an eligible visit. One workable rule is a completed food or drink order tied to an identified guest, excluding canceled or fully refunded orders. Decide how to handle partial refunds and record the rule. If one guest places two orders on the same day, decide whether that is one visit or two. The hypothetical worksheet counts each qualifying completed order as a visit; another restaurant may prefer a guest-day rule.
Check the practical details with the vendor before relying on the report:
- How are repeat orders matched to one guest, and what happens when a guest uses different contact details?
- Which ordering channels can be included in the same report?
- Can canceled orders, refunds, discounts, and split checks be handled consistently?
- Can you count guests and visits under the same rules before and after enrollment begins?
Do not assume that every eligible visit earns points. Your measurement definition and the program's earning rules are separate choices. Fleksa's loyalty documentation is a starting point for discussing the setup; confirm which orders are eligible in your intended workflow.
Log changes that can explain the comparison
Put a short note beside each period covering anything that could affect traffic or average spending. Include holidays, weather disruptions, local events, opening-hour changes, menu price changes, and unusually large catering orders. You do not need a perfect diary. You do need enough context to avoid making a buying decision from a number you cannot explain.
Channel mix matters as well. If the pre-launch report covers only direct online orders but the later report also includes counter orders, an increase in visits may reflect broader coverage. Either limit both reports to the shared channel or show each channel separately. If a new channel cannot be separated, label the comparison as incomplete rather than treating it as a like-for-like result.
Record other marketing activity alongside the loyalty launch. A discount sent to a large guest list, a new local advertising push, or a gift card promotion may affect visits and sales in the same period. Note the dates, audience, and offer where you can. Review your campaigns alongside the baseline rather than assuming all returning guests came back for points.
Also note changes in identification coverage. If staff begin asking more guests for a phone number after launch, the identified population may grow even if total traffic does not. Keep a separate count of all completed orders, including anonymous orders, when available. That gives you context for whether the tracked group represents more or less of the restaurant's activity. Do not add anonymous orders to the identified-guest visit rate.
Use a few measures to make a decision
Start with the worksheet's guest count, visits, and revenue. From those, calculate visits per identified guest and revenue per eligible visit. Look at both: a higher visit rate is useful to investigate, but a lower spend per visit could change the business picture. A larger identified guest count alone may reflect better data capture rather than more repeat business.
For the hypothetical restaurant, the first decision could be to continue the program for another comparable period while checking whether the increase holds. That follows from the modest rise in both calculated rates, assuming the restaurant's program costs are acceptable and no major operating change explains the difference. It would not justify a claim that loyalty caused the increase, or a decision to expand rewards without reviewing their cost.
Track program costs separately from sales. Record fixed software or setup charges, variable reward costs, and exception costs such as manual adjustments when they occur. State which costs your review includes and excludes; staff time and payment processing, for example, may sit outside a basic program-cost report. A quoted processing fee is not the restaurant's all-in operating cost. For help thinking through reward economics, use the guide to calculating restaurant loyalty reward cost.
Decide in advance what would prompt a closer look: perhaps a drop in visits per identified guest, costs above the amount budgeted, or a sharp change in the share of orders that can be identified. These are review triggers, not universal targets. If the periods differ substantially, fix the reporting definition or collect another period before drawing a conclusion.
Before launch, fill out the worksheet with your own dates and definitions, then ask Fleksa to compare the reporting and loyalty workflow with your needs and current availability. Start with its restaurant marketing automation information and bring your channel and export questions to the conversation.
Frequently asked questions
How long should a loyalty baseline run?
Four weeks can be a useful starting point when it includes a normal mix of weekdays and weekends. Use a longer period if visits are infrequent or the shorter window is unusually affected by events.
Should I count only loyalty members after launch?
Not for a direct comparison with all identifiable guests before launch. Keep the eligible population consistent, and track member enrollment as a separate measure.
Does an increase in repeat visits prove loyalty worked?
No. Prices, promotions, seasonality, channel coverage, and changes in guest identification can also affect the result. The baseline helps you spot a change and decide what to investigate next.
