A guest with an old last visit and a recent last order qualifies for a last-visit campaign if the visit meets your cutoff. That same guest may be excluded by a last-order rule. If you want to reach people who have not done either recently, both dates need to meet the cutoff under an AND rule.
The important buying question is whether your marketing tool uses the same definitions you do. Before sending, agree on what counts as a visit or order, how cancellations and missing dates are handled, and which records can update each date. Then compare a small set of expected results with the audience preview.
Define the activity before choosing a cutoff
A visit might mean a completed dine-in transaction, a checked-in reservation, or another event your restaurant records. An order might mean an accepted order, a completed transaction, or a fulfilled pickup or delivery. Those definitions are not interchangeable. A reservation alone does not necessarily prove that someone came in, and an order placed does not necessarily mean it was completed.
Choose the event date that matters for your campaign. For an order placed on one day and fulfilled the next, do you want the placement date or fulfillment date? For a visit, does the date come from check-in, payment, or another event? Ask the vendor which dates its filters actually use and whether your preferred choice is available.
Decide how cancellations affect the record. A canceled order may appear in a guest's history without counting as a qualifying order for a win-back campaign. A refund or partial cancellation may need a separate rule. Do not assume that an order's visibility in a profile means it updates the date used for segmentation.
Also ask which channels feed the dates. Dine-in, direct online orders, phone orders, and other transactions may not all be connected to the same guest record. Your restaurant POS system and marketing tool need a clear way to identify the same person if you expect activity from both to affect one audience.
Build a small set of conflicting records
The following example is hypothetical. The owner wants a campaign for guests inactive before October 1. A date before October 1 is old; October 1 or later is recent. For this exercise, a visit means a completed dine-in transaction, an order means a fulfilled order, and canceled orders do not update the qualifying order date. All dates refer to the same year.
The table shows four deliberately different guest records. The last column records the intended treatment before any audience rule is applied; it is not a claim about how a particular product behaves.
| Guest | Last qualifying visit | Last qualifying order | Other detail |
|---|---|---|---|
| Ana | September 10 | October 12 | Recent fulfilled order; old visit |
| Ben | October 8 | September 15 | Recent visit; old fulfilled order |
| Cara | September 20 | September 22 | October 5 order canceled; ignore it for this rule |
| Dev | No visit date | September 25 | Missing visit date is not proof of an old visit |
Cara's record is especially useful because a system might store the canceled order but treat it differently when calculating the last order date. Dev exposes a separate issue: a missing date is not automatically old or recent. For this exercise, a missing date does not satisfy a date condition. If the tool offers a separate “no recorded visit” condition, evaluate it as its own audience choice.
Make the records resemble your operation. Add an order placed before the cutoff but fulfilled afterward if that distinction matters. Add a guest whose reservation was canceled and another whose reservation was completed without a linked payment. The point is to identify which event changes the date, not simply to create a long list of guests.
Compare the audience rules
Apply each condition to the same four records. This makes the consequences of the operator's choice visible:
- Last visit before October 1: Ana and Cara qualify. Ben has a recent visit; Dev has no visit date.
- Last order before October 1: Ben, Cara, and Dev qualify. Ana has a recent fulfilled order.
- Last visit before October 1 AND last order before October 1: Cara qualifies. Both conditions must be true, and Dev's missing visit date does not satisfy the first condition.
- Last visit before October 1 OR last order before October 1: Ana, Ben, Cara, and Dev qualify. Each has at least one qualifying old date.
The OR audience is broader here. It includes Ana even though she ordered recently and Ben even though he visited recently. That may be appropriate if the message specifically invites guests back to dine in or place another order. It is a poor fit for a general “we haven't seen you lately” message if your intention is to exclude anyone with recent activity in either channel.
For that general win-back goal, the AND rule is the better match under these assumptions. It still leaves an open decision about guests like Dev. If you want to include people with no recorded visit and an old order, ask whether you can add that condition separately. Review that resulting audience rather than treating a blank as an old date by default.
Check the cutoff boundary too. In this example, October 1 is recent, so a date on October 1 does not qualify as “before October 1.” A filter labeled “before,” “on or before,” or “within the last” can produce a different audience. Confirm the time zone and whether the tool compares calendar dates or timestamps, especially when activity happens close to midnight.
Ask for evidence and run a safe comparison
Ask the vendor to show the filter definitions and the data behind the audience preview. A useful demonstration should answer these questions:
- Which event and timestamp update each last-visit and last-order field?
- Do canceled, refunded, or partially canceled transactions change those fields?
- How are missing dates treated, and can they be selected separately?
- Can visit and order conditions be combined with AND and OR? Are cutoff dates inclusive?
- Which channels contribute activity, and how are records matched when a guest uses different contact details?
- Can staff inspect an individual record and explain why it is included or excluded?
- When is membership calculated: at preview, when a campaign is scheduled, or at send time?
Use the four records above as a simulated check, with clearly identified test profiles or a vendor demonstration. The expected membership should match the lists in the previous section. A simulation must not send a guest message, take a real card payment, or dispatch a courier. Ask how the vendor can demonstrate the result without triggering those actions.
If you later choose to run a live service pilot, treat it as a separate step with real operational consequences. Set the audience, channel, timing, and approval process first. Do not use a practice scenario to trigger a live campaign. Also ask what happens if a guest places an order after the preview but before the scheduled send.
Access to an email address or phone number is not the same as permission to market to it. Ask how the tool records and applies each guest's channel preferences and suppression status. For questions about your obligations and contract terms, consult a qualified adviser.
Record the owner's acceptance criteria
Write down the decision in terms a manager can check. For the hypothetical general win-back campaign, an acceptance record could read:
- Purpose: Reach guests with no qualifying recent visit and no qualifying recent order.
- Events: Count completed dine-in transactions as visits and fulfilled orders as orders. Ignore canceled orders when identifying the last qualifying order.
- Boundary: Both dates must be before October 1. A date on October 1 does not pass.
- Missing data: A missing visit or order date does not pass that date condition. Review guests with missing dates separately.
- Expected preview: Cara is included; Ana, Ben, and Dev are excluded from this four-person example.
- Release: A manager checks the preview, channel permissions, message, and scheduled audience behavior before approving a send.
If the preview differs, do not simply adjust the list until the count looks right. Check the event definition, cancellation treatment, date boundary, matching, and missing-data rule. The vendor may need to explain a field or show a different configuration. If the desired rule is unavailable, decide whether a narrower campaign is useful rather than assuming the tool can produce it.
Keep this record with the campaign setup so another manager can repeat the decision. If your restaurant adds a new ordering channel or changes how visits are recorded, revisit the definitions. The same filter name can become less useful if the underlying guest activity is incomplete.
Use the worksheet to compare your requirements with Fleksa's restaurant marketing automation. Ask to see how your visit and order definitions, audience preview, and send controls would work before buying. For the operational records behind those questions, also review your restaurant management software workflow.
Frequently asked questions
Does a recent order exclude a guest from a last-visit campaign?
Not necessarily. A rule based only on an old last visit can include a guest who ordered recently. Add an order condition if recent orders should exclude them, and confirm how the tool combines the conditions.
Should a canceled order count as the last order?
Choose a definition that fits the message. For a campaign aimed at guests who have not received an order recently, excluding cancellations may make sense. Confirm whether the system can apply that definition.
What should I do with a missing visit date?
Treat it as unknown, not automatically old. Decide whether guests with no recorded visit belong in a separate audience, then check whether the tool can select them without changing the intended rule for guests with known dates.
