A report is a snapshot. A breaker has a life.

Every test a device has had, in order

Customer, site, switchgear, device, and at the end of that chain a record that outlives every job that touched it.

A single insulation resistance reading is almost meaningless on its own: a number against a limit that passes or does not. The same reading as the sixth in a series that has been sliding for four years is what tells you to act before the failure rather than after it.

• History per device, spanning every job, crew and year
• Every reading kept corrected and raw, so a comparison across years is of the equipment and not of the weather
• Every cycle opens back into the form that produced it
• Equipment that moves site takes its history with it

Finalize, lock, and keep the trail

There is a moment when a record stops being working data and becomes evidence. Finalizing locks it, and the lock is enforced at the server rather than in the interface: a locked cycle refuses edits and refuses deletion. Reopening one is a separate, deliberate act.

A locked cycle is never written over. A re-test (or an import carrying new values) opens a new cycle alongside the old one, so the history gains a row instead of losing one. What you cannot show is a field-by-field diff of an edit made before the lock: while a cycle is unlocked, edits overwrite what was there and the protection begins at the finalize. Tamper-evident signing of every change is an Enterprise capability.

• Explicit finalize step, so “draft” and “issued” are different states
• The lock refuses edits and deletes at the server, not merely in the interface
• Re-tests and re-imports open a new cycle rather than overwriting the finalized one
• Every cycle carries its test date and the technician it belongs to

What got missed, before somebody asks

The gap on a lineup is rarely the device that failed. It is the device nobody got to: the cubicle locked out that morning, the one behind the scaffold. It goes unnoticed because a device with no result looks exactly like a device nobody has written up yet.

Coverage is counted from the records themselves, so the gap list is a consequence of the data rather than a document somebody maintains.

• Passed, failed and untested counted per project and at every level of the tree
• Untested kept distinct from failed, because they are different problems
• Every device carries the date it was last tested and who did it

It is your data, and you can take it

A database of your test history is exactly the kind of asset that gets used to make leaving difficult. So, plainly: full export, at any time, in formats you can read without us. Not a filtered subset, not a support ticket, not a fee, not a thirty-day window after cancellation. Device records, readings raw and corrected, history, attached photographs and the audit trail, as CSV and PDF, in a structure documented well enough that a competitor could import it.

We would rather you stayed because the product is worth paying for. A retention strategy built on hostage-taking is a bet that it is not.

• Everything, not a report of everything, including photographs and the audit trail
• Self-service, immediate, unlimited, at no cost
• Open formats with a documented structure
• Available while you are cancelling, not only before you decide to

What it saves, and how to work it out

Neither is the real number. The real number is the failure you saw coming: and the trend is the only mechanism by which you get to see it early.

Who it changes the day for