Choosing a POS is mostly a decision about inventory accuracy, offline behaviour and who owns your data. The checkout screen is the least important part.
Most retailers evaluating a point of sale system start by comparing checkout screens. It is the part you see, so it is the part that gets demonstrated.
It is also the part least likely to determine whether the system works for you in a year. The decisions that matter are about what happens when stock counts drift, when the internet drops mid-transaction, and when you eventually want your data somewhere else.
Inventory accuracy is the whole game
A POS that records sales accurately but loses track of stock will slowly become something staff work around. Once the on-screen quantity and the shelf disagree, people stop trusting the number, and from that point the system is a till with extra steps.
Drift comes from ordinary events, not exotic ones:
- Goods received but not entered before they are sold
- Returns processed as new sales
- Damages and expiry written off inconsistently, or not at all
- Transfers between branches recorded on one side only
- Promotions and bundles that decrement the wrong item
None of these are prevented by better software alone. What good software does is make them visible quickly and cheap to correct — a variance report that someone actually looks at, an adjustment flow that records who changed what and why, and a stock take process that does not require closing the shop.
Ask any vendor to demonstrate a stock take and a variance investigation. It is far more revealing than the checkout demo.
Offline behaviour is a business decision
Ask what happens when the connection drops, and listen carefully to the answer.
"It keeps working" is not sufficient detail. The questions that matter: can you still sell? Can you still take payment? What happens to stock levels while offline? What happens when two branches reconnect having both sold the last unit of the same item?
There is no universally correct answer — a supermarket and a pharmacy have genuinely different tolerances here. But there must be a defined answer. A system whose offline behaviour is undocumented will surprise you during your busiest hour.
Mobile money changes reconciliation
In this market a meaningful share of takings arrives by mobile money, and that makes end-of-day reconciliation more involved than counting a drawer.
Payments confirmed on the customer's phone but not yet reflected in the till, partial payments split across methods, and reversals processed after close all have to be handled. A POS that treats mobile money as "cash with a reference number" will produce a daily total that does not match the statement, and someone will spend an hour every evening working out why.
What to look for: payments matched automatically against the transaction, unmatched payments surfaced as an explicit queue rather than buried in a report, and a daily position that reconciles against the payment statement rather than against itself.
Multi-branch changes the requirements sharply
Everything above gets harder with more than one location. Stock transfers need to be a tracked movement with a sending and receiving confirmation, not an adjustment at each end. Pricing needs a clear rule about what is central and what is local. Reporting needs to roll up without hiding a branch's problems inside the aggregate.
Many systems that work well for a single shop degrade badly at three. If growth is plausible, it is worth testing that scenario before committing.
Ownership of your data
This is the question retailers most often skip and most often regret.
Your sales history is the basis of every forecast, every supplier negotiation and every credit application you will make. If it can only be extracted as a PDF report, it is not really yours in any useful sense.
Before signing anything, establish: can you export complete transaction-level data, in a structured format, on demand, without paying extra? If the answer is unclear, treat that as the answer.
Practical evaluation checklist
- Demonstrate a stock take and a variance correction, not a sale.
- Unplug the network mid-transaction and watch what happens.
- Reconcile a day that includes mobile money, cash, and a reversal.
- Export a month of transaction-level data and open it yourself.
- Add a second branch and move stock between them.
A system that handles those five cleanly will serve you. One that handles only the first will not.
Where we fit
We build point of sale and inventory systems, e-commerce platforms, and the integrations between them — including mobile money reconciliation and multi-branch stock control. We also work with retailers who already have a system and need it to stop disagreeing with reality.
See our retail and e-commerce page, or tell us about your operation.