The test results nobody ever opens again

Every structured cabling job ends the same way. The testing gets done, the engineer connects the tester to a laptop, and out comes an export. Sometimes one enormous file, sometimes a folder of hundreds. It goes into the handover pack, the client accepts it, and it is never opened again.

Thousands of individual test results, every one of them evidence that a specific outlet in a specific room performs to specification, reduced to a single attachment nobody will read.

Why it ends up like that

Because the export is the last step, and by the time you get there the job is finished and everyone has moved on.

The tester holds results in its own memory as the engineer works. Nothing leaves it until somebody sits down at the end with a cable and does the dump. At that point the results are a batch — a thing produced in one go, filed in one go, and treated as one object.

So the natural home for them becomes a folder, because a folder is what a batch fits into. And a folder is exactly where information goes to stop being useful.

What that costs you

A question about one outlet becomes an afternoon. Three years later a client reports a dead socket in a meeting room. Was it tested? What was the margin? You have the file. Somewhere in it is the answer, indexed by a cable ID you now have to cross-reference against a drawing you also have to find.

Failures get fixed but not tracked. A link fails, the engineer re-terminates, it passes. Good. But the only record of that is a superseded result in a batch nobody reviews. There is no task, no status, no visible trail of what needed remedial work and whether it all got closed out.

Nobody checks the coverage. Four thousand outlets installed, three thousand nine hundred and sixty results in the export. Is that forty untested, or forty duplicates, or forty labelled differently? On most jobs nobody knows, because comparing a device schedule against a test export by hand is a job nobody has time for.

The evidence is weakest where it matters most. The whole point of certification is being able to demonstrate, later, that the work was right. A format that makes retrieval hard undermines the reason you did the testing.

The obvious alternative

Open the task for an outlet. See its test result.

Not a link to a file. The actual result — the wire map, the margins, the pass or fail, the date it was tested and by whom — sitting on the task for that outlet, on the drawing, where the work was.

Everything follows from that. The client question becomes one click. A failure becomes a task with a status like any other work, which means it gets tracked and closed out rather than assumed. And the coverage gap becomes obvious, because an outlet with no result is a task with a missing field rather than an absence you would have to go looking for.

Why this is closer than it sounds

The interesting part is that it does not require anything new from the tester.

The export is already structured data. Every result already carries a cable ID, because the engineer typed it in before pressing test — that is how the tester knows what it is measuring. That identifier is the join.

If your outlet references and your cable IDs follow the same convention, and on a well-run job they do, then matching a result to the task for that outlet is mechanical. Parse the export, match on the label, attach each result to its task. The bulk file stops being the destination and becomes the transport.

It is the same shape as connecting a network management platform, where the serial number does the joining. Different field, same principle: two systems describing the same physical object, waiting for somebody to line them up.

The wider point

This is not really about cable testing. It is about what happens to structured data produced on site.

Test results, commissioning sheets, certificates, photographs — all of it is generated at a specific place, about a specific device, by a specific person, and almost all of it ends up in a filing structure that throws away two of those three facts. The device schedule knows what. The drawing knows where. The export knows how it performed. Three systems describing the same socket, and no link between them.

Put the result on the task and the link is permanent, because the task already knows the location, the installer and the date.

That is the job. Not more data. The same data, attached to the thing it describes.

See what we are building connectors for · Graftix for data and network contractors