Most chain rollouts die in the pilot. Almost never because the device was wrong — usually because the pilot was designed as a miniature rollout instead of a test of the decision that has to be made next.
A chain of forty stores does not want to evaluate a rugged tablet. It wants to know three things, in this order: will our app run on it without changes, will the store staff break it, and what does it cost per store over three years. A pilot that answers those three questions in five weeks is worth more than one that produces a 60-page comparison matrix after four months.
The pilot should test one thing: your software's assumptions
Write down what your POS or inventory app assumes about the device before you order anything. In practice it is always four or five of these:
- A specific Android API level, or a Windows version
- Whether Google Mobile Services are present
- How the scanner delivers data — keyboard wedge, intent broadcast, or SDK call
- Whether the camera is used as a fallback scanner
- How the app behaves when the network drops mid-transaction
Every one of these can kill a rollout after purchase. None of them can be resolved by reading a datasheet. They are resolved by putting the actual app on the actual unit for two weeks.
The failure pattern we see most often: the app uses camera-based scanning in a demo, then the retailer wants a hardware scanner for speed, and the app has no path to the hardware engine. That is a software problem discovered at hardware evaluation, which is the most expensive possible moment to discover it.
Five units is enough, if the five units are the right five
A common mistake is to pilot with five identical devices in five similar stores. That tests durability and nothing else.
A better shape is five units across the range of conditions the chain actually has:
| Store type | What it stresses |
|---|---|
| Flagship, high traffic | Throughput, battery drain, heat |
| Small format, low ceiling | Wi-Fi dead spots, sync failures |
| Suburban with outdoor receiving | Sunlight readability, drop risk |
| Store with a stockroom two floors up | Portability, battery swap |
| The store the regional manager dislikes | Staff who will not read instructions |
The last one matters more than the others. If the device survives the store with the least careful staff, it will survive the chain.
Accessories are half the pilot, and always the part that ships late
A five-store pilot needs five charging cradles, five grip straps, five spare batteries and one spare device. Not a carton of each.
This is where trading companies and distributors diverge sharply. A distributor sells in boxes of ten and treats a five-unit order as an inconvenience. If your pilot needs accessories to be useful, confirm accessory availability with the pilot order, not after it — otherwise week one of a five-week pilot is spent chasing cradles.
Ask for the accessory list as a line-item quote alongside the devices. If a supplier cannot quote cradles and batteries in single units, they cannot support a pilot, and they will not support a rollout either.
The three numbers that decide the rollout
When the pilot ends, the decision is not "did we like the device". It is:
- Support tickets per store per month. Two or more means the device or the app is wrong. One or fewer means it is working.
- Scan-to-confirm time on the busiest task. Measure it on day one and day twenty. If it has not improved, the hardware did not solve the original problem.
- Devices that came back. Not failed — came back. Devices that were left in a drawer or handed in because staff preferred their phone are a stronger negative signal than breakage.
Run those three numbers for five stores and you can extrapolate a forty-store rollout with reasonable confidence. Run a 60-page evaluation and you will still be guessing.
What to ask a supplier before the pilot starts
Four questions, all answerable in one email:
- Can you supply five units and five cradles on one order, at sample pricing?
- Can you pre-configure kiosk mode and Wi-Fi before dispatch, so the pilot units arrive ready to test?
- If the pilot fails on a software assumption, what is your returns position?
- Do you hold the same model and firmware line for at least 24 months, so a rollout in quarter three matches the pilot in quarter one?
The fourth question is the one buyers forget and later regret. A pilot on a model that is discontinued before the rollout completes turns a successful test into a re-qualification project.
A note on price per store
Chains compare total cost across two hundred stores, and device price is only one line. The others are cradle, spare battery, expected replacement rate and support hours. A device that costs 15% more but returns 30% fewer units is cheaper, and a device that needs a new cradle design for every store format is more expensive than its sticker.
Model the three-year cost per store before the pilot, not after. The pilot then becomes a check on your model rather than a substitute for it.
If you are at the stage of choosing five units to test, tell us the store formats and the app's assumptions and we will shortlist two or three models rather than send a catalogue.
