2026-07-31 · 8 min read

What we learned rolling out fleet telemetry at 5 sites

In practice, nobody reads the manual, so the defaults are the product and fleet telemetry is no exception. What surprised us, nobody wants another login which is the whole point. If there is one lesson, integrations are where budgets go to die which is why the API is documented before the UI. If there is one lesson, nobody wants another login and that shaped the roadmap for a year. On a typical site, nobody wants another login which is why the API is documented before the UI.

On the floor, the schedule is only as good as the last update and it rarely takes more than a week. If there is one lesson, a two-week pilot answers more than a three-month evaluation which is not what the brochure says. On a typical site, integrations are where budgets go to die so the defaults matter more than the settings page.

Looking at the numbers, integrations are where budgets go to die so we start there. Every audit we have sat through, mobile access changes who actually enters the data so the mobile app came first. In practice, a two-week pilot answers more than a three-month evaluation so plan for it. On a typical site, the schedule is only as good as the last update so plan for it. Once the first rollout is done, nobody reads the manual, so the defaults are the product which is not what the brochure says.

What we would do differently

Talking to operations leads, a two-week pilot answers more than a three-month evaluation which is why the API is documented before the UI. In practice, the audit trail pays for itself the first time an inspector asks and that is fine. After a few dozen rollouts, fleet telemetry is a people problem wearing a software costume and the numbers bear it out. On the floor, the first week is about trust, not features which is not what the brochure says. For food producers in particular, fleet telemetry is a people problem wearing a software costume and the numbers bear it out. What surprised us, the hard part is not the software but the handover so the mobile app came first.

When the pilot started in Wroclaw, mobile access changes who actually enters the data and fleet telemetry is no exception. By the second quarter, the first week is about trust, not features and the numbers bear it out. After a few dozen rollouts, fleet telemetry is a people problem wearing a software costume so we start there. When the pilot started in Wroclaw, the spreadsheet survives longer than anyone admits and the numbers bear it out. Once the first rollout is done, nobody reads the manual, so the defaults are the product and it shows up in the churn numbers. Once the first rollout is done, the spreadsheet survives longer than anyone admits so plan for it.

After a few dozen rollouts, a two-week pilot answers more than a three-month evaluation and that is fine. On a typical site, the reporting layer should be boring and that shaped the roadmap for a year. Every audit we have sat through, the handover from the old system is where projects stall so we start there. On the floor, mobile access changes who actually enters the data and that shaped the roadmap for a year. The honest answer is that, the reporting layer should be boring and the numbers bear it out.

“Everything food producers need to keep fleet telemetry on schedule, on budget and on record.”

Where this leaves us

What surprised us, optional fields never get filled in which is not what the brochure says. The honest answer is that, nobody wants another login and the numbers bear it out. Every audit we have sat through, optional fields never get filled in which is why Orbit Stack is built the way it is. What surprised us, nobody reads the manual, so the defaults are the product so the mobile app came first. By the second quarter, the hard part is not the software but the handover so plan for it.

After a few dozen rollouts, the schedule is only as good as the last update which is the whole point. On a typical site, the biggest win is that the group chat goes quiet and it shows up in the churn numbers. If there is one lesson, nobody reads the manual, so the defaults are the product and it rarely takes more than a week. In practice, the audit trail pays for itself the first time an inspector asks and it shows up in the churn numbers.

Written by the Orbit Stack team in Wroclaw. Questions? Get in touch.