Choosing a point of sale system is a decision you make in an afternoon and live with for years. The switching cost is not the subscription - it is retraining staff, remapping products, re-integrating hardware and the fortnight where nobody quite trusts the numbers. So it is worth being slightly more systematic than watching three demos and picking the friendliest salesperson.
We build POS software, so treat everything below with appropriate suspicion. But we have also watched a lot of businesses migrate away from systems that looked fine at the demo, and the failure patterns repeat.
Start from your worst hour, not your average one
Every POS looks capable when you are clicking through it calmly. The real test is your busiest, most chaotic hour: the Saturday evening queue, the lunch rush with three tables waiting to split a bill, the delivery van arriving while a customer is arguing about a return. Write down what happens in that hour, then ask the vendor to demonstrate exactly that. Not the feature list - the hour.
- How many keystrokes or taps does a normal sale take, from first scan to drawer open?
- What happens when a customer wants to pay half in cash and half by card, and then changes their mind?
- How does a return of two items out of five from a receipt three weeks old actually work?
- Can a supervisor approve a discount without logging the cashier out?
The offline question is not a checkbox
Almost every cloud POS claims an offline mode. There is an enormous difference between a system designed to be local-first and one that has an emergency fallback bolted on. Ask specifically:
- 1Which functions stop working offline? Loyalty lookup? Customer credit? Returns? Reports?
- 2How long can it stay offline before something breaks?
- 3If two tills are offline at once and both sell the last unit, what happens on reconnection?
- 4Can I see what is queued, and prove it was sent later?
Understand what you are actually paying
POS pricing is where the real surprises live. The headline number is rarely the number. Build a three-year total and insist on written answers to each of these:
| Cost | What to ask |
|---|---|
| Per register | Is a back-office computer a register? Is a waiter's phone? Is a kitchen screen? |
| Per user | Does adding a part-time cashier cost money? (If yes, expect shared logins and a useless audit trail.) |
| Per location | What happens to the price the day you open a second shop? |
| Transaction fees | Any percentage of sales, online orders, or payment processing? |
| Modules | Is accounting extra? Loyalty? E-invoicing? Kitchen display? API access? |
| Support | Is phone support an upgrade? What are the hours in your timezone? |
| Setup & migration | One-off fee? Who does the data import - you or them? |
| Exit | Can you export everything in a usable format? Is that in the contract? |
Ask about the boring modules early
Almost everyone evaluates the checkout screen carefully and the back office barely at all. Then eight months later they discover that stock adjustments do not post to the ledger, purchase orders cannot handle a partial delivery, and the tax report needs manual correction every month. The checkout screen is the part you will spend the least time thinking about after week two.
- Does a stock adjustment create an accounting entry, or just change a number?
- Can I receive 40 of 100 ordered units and keep the balance open?
- Does the system understand batches and expiry dates if I ever need it to?
- Can I see gross margin by item, or only revenue?
- How do I correct a mistake made three weeks ago without deleting history?
Compliance is not optional, and it is not free
If your market requires structured e-invoicing - KSeF in Poland, Factur-X in France, Peppol in Belgium and the Netherlands, Verifactu in Spain, and a growing list elsewhere - this is not a feature you can bolt on later. Find out whether the integration is included or an add-on, whether it queues safely when the platform is unreachable, and who runs the sandbox testing. A vendor who describes this as trivial has probably not done it.
Check the exit before you check in
The single most useful question in a POS demo is: if I decide to leave in two years, what exactly do I get? A good answer is a structured export of products, stock, customers, transactions and ledger entries, available on demand, written into the contract. A bad answer is a stack of PDF reports and a phone number for the retention team.
“Software you cannot leave is not a partnership. It is a hostage situation with a monthly invoice.”
The one-page checklist
- 1Demo your worst hour, not their happy path.
- 2Test offline for real, on the device you would use.
- 3Get a written three-year total cost including every module you need.
- 4Confirm the back-office modules work the way your business actually operates.
- 5Verify tax compliance is included, tested and queue-safe.
- 6Confirm hardware compatibility with the specific models you own.
- 7Ask who does the data migration, and see a sample of their work.
- 8Confirm the data export path and get it in writing.
- 9Speak to two customers of a similar size and type - not the vendor's favourite reference.
- 10Run a two-week parallel trial before you cut over.
If a vendor is comfortable with all ten of those, you are probably in good hands. If they are uncomfortable with three or more, keep looking - there are plenty of good options, and the one that survives this checklist is the one that will still suit you in eighteen months.
Want this handled by software rather than a spreadsheet?
SumPOS does most of what this article describes automatically - and the free plan is genuinely free, so you can test the idea before committing to anything.