Before buying a POS, ask the vendor to show whether you can isolate the same sales period by location, device, and staff member without counting an order twice. The answer depends on more than whether those filters appear on a screen. You need to know what each filter means, how filters work together, and which transactions the report leaves out.
A useful demo starts with a small sample whose answer you already know. Then you can compare the report with your own count, ask about differences, and decide whether the results are reliable enough for your close and management routines.
Choose a two-device sample shift
The following example is hypothetical. It is a controlled reporting exercise, not a claim about any POS. Use a vendor's demo or sandbox data, and confirm that practice transactions cannot create a real payment, guest message, or courier dispatch.
For one location, use a shift from 5:00 p.m. to 9:00 p.m. on a single date. Two devices, A and B, record six completed orders. Two staff members, Ana and Ben, both use the devices. The amounts below are order sales amounts, not deposits or payment-processing fees.
- Device A: Ana records orders of $10 and $20; Ben records one order of $15.
- Device B: Ana records one order of $25; Ben records orders of $30 and $40.
There are six orders totaling $140. Device A accounts for three orders and $45; device B accounts for three orders and $95. Ana accounts for three orders and $55; Ben accounts for three orders and $85. Each order has one device and one staff member assigned for this exercise.
The figures exclude fixed costs, variable costs such as processing fees, and exception costs. They also exclude refunds, discounts, taxes, tips, and voids. The exercise checks report counts and sales amounts; it does not calculate profit or all-in operating cost.
This small set can expose a problem that a daily total alone would miss. A report could show $140 for the shift but still assign an order to the wrong device or person. That matters to the owner reviewing sales, the manager closing the shift, and staff whose activity may be discussed using the report. A sales attribution report should not be treated as a measure of an employee's overall performance without understanding how shared work is recorded.
Ask how filters and groupings work together
Have the vendor show the unfiltered report first, then add location, time, device, and staff filters one at a time. A filter narrows the records included; a grouping divides the included records into rows. Ask whether you can filter to one location and one shift while grouping those same results by device, staff member, or both.
Pay particular attention to the word device. Does it mean the terminal where an order was opened, sent, paid, or last edited? If an employee starts an order on A and another closes it on B, which device gets the sale? Likewise, ask whether the staff filter uses the person who opened the order, the person who took payment, or another role.
Ask the vendor to show the definitions in the report or its documentation and record the answer beside the demo result. If an answer depends on configuration, note the setting used. A label such as “staff sales” is not enough to tell a manager how shared orders are allocated.
Also establish which time the date range follows. An order opened before 9:00 p.m. but paid after 9:00 p.m. can appear in different periods depending on whether the report uses opening time, closing time, or payment time. Ask whether the end of the range is inclusive and which time zone applies. For a multi-location business, ask whether the location filter follows the location where the order originated or where payment was taken.
These details are worth checking even if you already like the rest of the restaurant POS system. A report that cannot combine the dimensions you need may require a separate export or manual reconciliation at every close. Ask who can access the report, whether the same filters work in an export, and whether saved report settings are available for the manager who will use them.
Compare the filtered results with the whole shift
Start with the six-order, $140 result for the location and shift. Next, apply the device filter. A and B should each show three orders, and their $45 and $95 totals should add back to $140. Clear the device filter before testing staff: Ana's $55 and Ben's $85 should also add back to $140.
Then combine dimensions. Device A plus Ana should return two orders totaling $30. Device B plus Ben should return two orders totaling $70. The four device-and-staff combinations should cover the six orders once: A and Ana, two orders and $30; A and Ben, one order and $15; B and Ana, one order and $25; B and Ben, two orders and $70.
If a result differs, inspect order-level details before deciding the report is wrong. The report may use a different timestamp or attribution rule than the one you expected. Ask the vendor to identify every included order and explain the difference. A total that happens to match is not sufficient evidence if the underlying orders are different.
Keep exception handling separate from the clean sample. For example, ask the vendor to add one simulated order that two employees work on, then show it in each employee's filtered view and in a grouped report. If the full order appears under both people, adding the staff rows may double count it. That may be acceptable for an activity view, but it is a problem if your managers intend to sum those rows as sales. Ask whether there is a report that assigns each order once, and how it chooses the person.
Run a similar discussion for refunds, voids, discounts, tips, and orders that cross the shift boundary. Do not assume a sales report, payment report, and deposit record should match line for line: they may measure different events or periods. For the wider close process, see payments, processing cost, cash close and POS reports.
Use a filter test matrix before selecting a POS
The matrix below uses the hypothetical shift above. Every row uses the same date and 5:00–9:00 p.m. period at the same location. The expected count is a count of distinct completed orders, with one device and one staff assignment per order. Record the actual result and the report's attribution rules alongside a copy or export of the demo output.
| Period and location | Filter or grouping | Expected orders | Expected sales |
|---|---|---|---|
| Sample shift, one location | No device or staff filter | 6 | $140 |
| Same shift and location | Device A | 3 | $45 |
| Same shift and location | Device B | 3 | $95 |
| Same shift and location | Ana | 3 | $55 |
| Same shift and location | Ben | 3 | $85 |
| Same shift and location | Device A and Ana | 2 | $30 |
| Same shift and location | Device B and Ben | 2 | $70 |
The combined-filter rows check intersections, not extra sales to add to the unfiltered total. To reconcile the whole shift using combined filters, include all four device-and-staff combinations, including A and Ben and B and Ana. Compare order IDs as well as amounts so two offsetting mistakes do not hide behind a correct dollar total.
For each vendor, keep the sample inputs, the exact date range, screenshots or exports, the actual counts and amounts, and an explanation of any mismatch. Ask whether exports preserve order IDs and selected filters. If a result requires a workaround, write down the steps and who must perform them at each close.
Set acceptance criteria before the demo: the clean sample must return the expected distinct order counts and amounts under the agreed definitions, or the vendor must explain a difference you are willing to accept. For shared orders and other exceptions, decide whether the available report gives your team a usable way to reconcile without adding overlapping rows. Do not mark an unresolved question as a pass because a different report looks promising.
Apply the same matrix to each product on your shortlist. Bring your manager into the review: a technically correct report may still be awkward if the person closing has to rebuild filters every night. The POS questions page is a useful starting point for broader questions, but the sample results should drive this particular decision.
Make the next demo specific
Request a demo of Fleksa's restaurant POS using the sample scenario. Ask the vendor to document the report used, its filter and attribution rules, the results for each matrix row, and any unresolved exceptions. Then compare that record with your acceptance criteria before making a selection.
Frequently asked questions
Can I add staff sales rows to get the shift total?
Only if the report assigns each included sale once across those rows. Shared orders or multiple staff roles can make rows overlap. Check order IDs and the report's attribution rules before adding them.
Should filtered sales equal card deposits?
Not necessarily. Sales and deposits measure different things and may cover different periods. Ask how the POS defines each report and how payments, refunds, and processing fees appear in the reconciliation.
Should I run the sample during a live shift?
Start with a simulation in a demo or sandbox where no real payment, guest message, or courier dispatch can occur. If you later assess reporting during live service, agree on a separate pilot plan and reconcile actual orders under normal operating procedures.
