Daily operations report from five vendor systems

The client

The same Hungarian contracting company, which also handles ongoing operation of the systems it delivers: monitoring technical equipment across dozens of sites, watching it daily and dealing with faults. The system was not built as an external engagement — it was built from the inside, as the company's process manager. The company name is not published.

The starting point

The equipment comes from five different manufacturers, and each manufacturer shows its data on its own portal. There was no single picture anywhere.
Every morning someone had to work through those interfaces, site by site. It was routine work, but it could not be skipped.
A fault was noticed if somebody happened to be looking at the right page. Whatever nobody looked at never surfaced as a problem.
Partial failure — where equipment runs but runs badly — is the hardest to spot. It does not raise an error; it produces a lower number that looks unremarkable on its own.

The solution

A daily routine that collects data from every site, compares it against what was expected, and sends one report by morning.

Five manufacturers, one format — each vendor system has its own adapter, but all of them translate what they return into the same unified data shape. The practical benefit is that connecting a new manufacturer does not rewrite the report, the email or the register — it only needs a new adapter.
Automatic daily report — ready by morning, by email. Nobody has to log in anywhere to find out whether yesterday went fine.
Missing data treated separately — the system distinguishes equipment that reported zero from equipment that reported nothing. These are not the same thing, and conflating them makes one of the two faults invisible.
Underperformance monitoring — the system flags not only outright stoppage but also anything running persistently below its usual level.
The site list stays with the company — a site can be added to or removed from the daily and monthly report from a single cell. That does not require a developer.
Monthly summary — alongside the daily picture, a monthly overview that shows the longer-term trend.

Outcome

It found two faults nobody had seen for months.

One unit sent no data for eight weeks while it was running — the gap only showed up in the aggregate, as a persistent shortfall of a few per cent. In another case, three units ran significantly below their usual level for two weeks and then recovered on their own. Neither raised an error. Under manual checking, these are exactly the kind of fault that slips through.

The alerts do not need silencing. Back-tested over 59 days, the detection produced not a single false alarm. That is the difference between a report people read and one everyone filters into junk after two weeks.

The morning round is gone. Nobody has to log into five interfaces to establish that there is nothing to do.

Where else this works

The core of the solution is not the type of equipment but a general situation: a single, consistent picture has to be produced from several supplier systems, regularly, without manual work. That looks the same in many places:

Companies operating plant and equipment fleets — where machines come from different manufacturers, each with its own interface.
Property and facility management — several sites, several systems, one management view.
Logistics and fleet management — vehicle and asset data from multiple sources.
Any service provider reporting regularly to clients — a monthly client report is exactly the same pattern: gather data from several places, send it out in a consistent form, and flag anything that deviates.

The precondition is that the data must be reachable by machine somehow — via an API or an export. If it only exists on a screen and nowhere else, that is where to start.

Operation

The system runs on a dedicated server in the EU, on a daily schedule. Collection and sending are separate steps, so if a vendor system happens to be unavailable, the report still goes out — and shows which source is missing. The system continues to grow: connecting a new vendor source does not affect the existing ones.

This case study is based on a real system, anonymised. Time savings were not measured, so no figure is claimed for them.