Business

Where the work done on site ends up

In companies working away from base, data is born on site and reaches the back office days later, reconstructed from memory. The result is that margin per job is known only once the job is finished. Capturing data where it is born — access, hours, materials — and attaching it to analytical accounting changes the moment when you can still act.

29 Apr 2026 · 8 min read · updated on 27 Aug 2026

In a company working away from base — construction, installations, maintenance — the most important data is born in the most awkward place: a site, a roof, a plant room. It is born as a number on a sheet of paper, a delivery note tucked into a dashboard, a sentence said to the foreman.

Then the journey starts. The sheet reaches the office on Friday, or Monday. Someone interprets it, asks for confirmations, reconstructs. Meanwhile the work has moved on. By the time the data finally reaches accounting it is a week old and has lost pieces.

The problem isn't accuracy: it's delay

Anyone who has managed jobs knows the feeling: you discover a site was badly margined once it is finished. At that point the data is perfect and completely useless — nothing can be done.

The value of knowing today that a crew is using forty per cent more hours than quoted is not in the accounting: it is in the fact that today you can go and see why. Tomorrow it is an explanation; at the end of the job, a loss.

Hours reconstructed at the end of the week are always rounded, and always in the same direction.
Materials taken from stock "for that site" and not recorded immediately end up, in accounting terms, nowhere.
Variations agreed verbally with the client on site are remembered when the dispute arrives.
Access records of who came and went, without a register, become a problem on the day of an inspection or an accident.

Data shouldn't be chased at month end: it should be captured at the moment and in the place where it is born. Everything else is reconstruction.The principle behind every field project

Capturing data where it is born

The solution isn't asking people to be more precise: it is removing the transcription step. If whoever is on site records the entry, the crew, the time and the materials from their phone, on the spot, that data doesn't have to be reconstructed by anyone.

For it to work, the app has to respect three constraints that don't exist in an office:

Few taps. The most frequent action of the day must fit in two or three gestures, with large type and wide buttons: it is used with gloves, in sunlight and with one free hand.
No long forms. Whatever can be inferred — who you are, where you are, what time it is — is not asked.
Tolerant of connectivity. In a basement plant room there is no signal. The data is recorded anyway and sends itself when the signal returns.

If one of those three is missing, the app is used for two weeks and then the paper comes back. That is not resistance to change: it is that paper, in those conditions, works better.

The piece that changes everything: attaching it to the job

Recording access on an app, by itself, moves the problem a few metres: you would have precise data in a system that doesn't talk to accounting. The leap happens when the data enters the business system already assigned to the right job.

That is what analytical accounting is for: every hour and every material carries the site it belongs to, and cost accumulates there while the work proceeds. Margin per job stops being a year-end calculation and becomes a number you look at on Wednesday morning.

On a project for a construction company we did exactly this: the app records site access, the data flows into the business system and attaches to the job's analytical accounting. Nobody re-keys anything, and the back office has stopped chasing foremen on Friday afternoons.

What you find out later

Two effects clients report after a few months, and that nobody would put in a quote.

The first: quotes get more accurate. With two years of real data on what a certain type of job actually costs, the next quote stops being a bet based on the memory of the last site that went well.

The second: arguments change tone. With the client, because a variation recorded on site with a date and a photo is not an opinion. With the crews, because the conversation is no longer "you took too long" but a number looked at together.

If someone in your company spends Friday afternoon chasing sheets and reconstructing weeks, that is the cost to measure. It is not a people problem: it is data travelling the wrong way.

Will the crews really use an app?

If the daily action fits in two taps, yes — they use it because it removes the sheet they had to fill in at the end of the day. If it asks them to fill in fields, no: they will go back to paper and you will have spent money for nothing. That is the whole story.

Does it work without a connection?

It has to. Data is recorded on the device and sent when the network returns: it is a requirement, not an option, and it must be tested in the worst conditions you have — not in the office.

Can it be used for safety and inspections?

The access register serves that too: knowing who was on site and when is useful in an inspection or after an accident. Mind the boundary though: recording attendance means processing workers' data, so purpose, retention and notice must be defined. It is not an obstacle, it is part of the project.

Do we need to change our business system?

Almost never. If it has analytical accounting and a way to receive data from outside, it can be connected. The case worth reconsidering is a system that doesn't handle jobs at all: there the problem is different and bigger.