A new point of sale is the most visible system a retail or hospitality business can replace. Every customer sees it, every cashier uses it every few minutes, and if it fails on a Saturday morning the queue is out the door. We have been part of enough POS go-lives, good and bad, to know that the problems rarely come from where people expect.

Most POS go-lives are planned around the software. Most of the trouble comes from three other places: hardware that does not quite work with it, staff who were shown it once, and a back office that nobody was made responsible for.

Hardware: the problems you find on install day

POS software is often sold as "works with any hardware". In practice, compatibility is a list, and the list matters. The issues we see most often:

  • Receipt printers without supported drivers. The old printer works with the old till but not the new software, or only prints in a fallback mode that cannot cut paper or open the drawer.
  • Cash drawers wired to the wrong place. Most drawers open through the receipt printer. Change the printer and the drawer goes silent.
  • Scanners that do not read your labels. Barcode types, scanner settings and the quality of in-house printed labels all matter. A scanner that misreads one item in twenty turns every busy period into manual entry.
  • Card machines that do not integrate. The card terminal works on its own, but the amount must be typed in by hand, so the till and the bank disagree by the end of the day. Check whether your bank's or provider's terminal integrates with your chosen POS before you sign either contract.
  • Scales and label printers for deli, butchery or bulk goods, which use their own protocols and are often the last thing tested.
  • Underpowered devices. Old PCs or cheap tablets that were fine for the old system and are painfully slow with the new one.
  • Network and power. A till on unreliable Wi-Fi, no UPS, and a router on the same plug as the kettle. In South Africa, a POS plan needs a power plan and a connectivity fallback.

The fix is unglamorous: build one complete till, with every device you intend to use, and test it end to end before buying the rest. Sell, refund, print, open the drawer, scan your worst labels, weigh, take a card payment, cash up. Do it a few weeks before go-live, not the day before.

"Compatible" on a brochure means it worked in someone's test lab. Compatible in your shop means it worked on your counter, with your labels and your card machine.

Ask what happens offline

Pull the network cable during your test. Can the till keep selling? Does it queue the sales and sync them later? Can card payments still go through on the terminal's own connection? A POS that stops when the internet does is a POS that stops your trading. We wrote about this in offline-first software, and a till is the clearest case for it.

Training: the till and the people at it

The typical training plan is a one-hour demonstration the day before go-live. It is not enough, and it shows on day one: long queues, voided sales, managers called over every five minutes, and staff quietly writing sales on paper to "put in later".

What works better:

  • A practice till set up in the back a week or two before go-live, with real products, where staff can make mistakes without consequences.
  • Scenarios, not menus. A normal sale, a return without a slip, a split payment, a staff discount, a price override, a void, an account customer, a gift voucher, and the end-of-day cash-up.
  • The bad days too. What to do when the card machine is offline, the printer is out of paper, or a product will not scan.
  • Different depth for different roles. Cashiers need speed on the common tasks. Supervisors need overrides, refunds, cash-up and the reports that show whether the day balances.
  • A floor champion. One staff member per shift who has had extra training and is the first port of call, so the manager is not the only person who can help.

Plan for extra staff and slower service on the first trading days, and avoid going live on your busiest day of the month. A Tuesday morning go-live is dull. That is the point.

Prices and products are part of go-live

The till can only be as right as the product file behind it. Before go-live, someone must check every active product's code, barcode, description, price and VAT setting, and remove the hundreds of discontinued lines. A wrong price at the till becomes a customer complaint in seconds. It is the same principle as capturing data at the source: get it right once, where it starts.

The back office: where the value actually is

The till is what everyone sees, but most of the value of a modern POS lives in the back office: product setup and pricing, promotions, stock receiving, purchase orders, stock takes, supplier returns, user permissions and reports. In many go-lives this part is barely touched. Staff use the new till exactly like the old cash register, while stock is still counted on paper and orders are still placed by memory.

That usually happens because the back office has no owner. Name one. Train them separately and in more depth, and give them a list of what should move into the system in the first month:

  1. All stock received through the system against a purchase order or invoice, not just added.
  2. Price changes and promotions made in the back office, not by override at the till.
  3. A first full stock take, so that stock figures start from the truth.
  4. Daily reports reviewed: sales by product, voids and refunds by user, cash variances.
  5. User permissions set so that each person can do their job and nothing more, with their own login. Shared till logins make every void and refund untraceable.

Once stock is received and counted in the system, the POS starts to answer questions the old till never could: what to reorder, what is not selling, where shrinkage is happening. That is the step that turns a till replacement into a better-run business, and it is what drove the results in our manual-till-to-POS case study.

What we would tell anyone about to go live

  • Test one full till, with every device, weeks before go-live.
  • Confirm card-machine integration before signing with either provider.
  • Clean the product file before importing it.
  • Train on scenarios, at a practice till, with a champion on every shift.
  • Go live on a quiet day, with extra hands and support on site.
  • Name a back-office owner and give them a first-month checklist.

Get those right and go-live day becomes an ordinary trading day that happens to have a new screen at the counter. That is the best result a POS rollout can have.