Most industrial and commercial sites are already generating the data for a live energy picture. The meters are in. The building management system is logging. The half-hourly data is sitting with the supplier.
What is missing is not instrumentation. It is that nobody ever connected any of it to a screen someone looks at.
What you probably already have
Half-hourly data from your electricity supplier. Larger sites are metered this way as a matter of course, and the data is usually available on request or through a portal. It is often the single richest source on site and the least used.
A building management system. Controlling heating, ventilation and cooling, and logging as it goes. Most have some form of interface or export; many have never had anything read from them.
Sub-meters. Fitted during a refurbishment or a grant scheme, wired, powered, and read by nobody since commissioning.
Machine-level data. Modern plant frequently reports its own consumption, into a system that reports it to nobody.
The unglamorous truth of most first engagements is that the first month is not installing sensors. It is finding out what already exists and what it talks to.
What week one usually surfaces
Three things come up so consistently that they are worth naming in advance. What they are worth at your site depends entirely on your site — but they are what people find.
Baseload that makes no sense. The consumption when nothing is running. Every site has one; few know what theirs is. Once it is on a chart, the question "what is drawing that at 3am?" becomes answerable, and the answer is usually specific and fixable.
Weekend and shutdown load. The plant is closed and the line is still drawing. Compressors cycling against a leak, ventilation on a schedule nobody has revisited since a shift pattern changed years ago, heating in an unoccupied building.
Behaviour that changed quietly. A machine drawing more than it did last quarter, gradually enough that nobody noticed. That is often a maintenance signal before it is an energy one — which is why the alerting tends to earn its keep in ways the business case did not predict.
Notice that none of these are AI findings. They are visibility findings. The data was always there; nothing was looking at it.
Why annual reporting cannot show you any of this
An annual figure is one number describing a year you can no longer influence. It cannot show you a weekend, because a weekend does not survive the averaging.
This is the real cost of treating energy data as a compliance chore. The reporting obligation is what forces the data to be collected, and then it is collected in a form that answers only the obligation — annually, in aggregate, too late to act on.
The same data at half-hourly resolution, visible continuously, answers a different question: not "what did we use last year" but "what is happening now that shouldn't be."
What a first build actually involves
Establish what data exists and what will talk to what — the part that takes longest and is never in a proposal. Collect it on a schedule into one place. Put it on a screen someone will actually open, showing consumption against time, by site and by circuit where the metering allows it.
Then alerting, which is where it stops being a dashboard and starts being useful: a message when consumption departs from its own pattern. Not thresholds someone has to maintain — a comparison against how this circuit normally behaves at this hour on this day of the week.
And underneath it, a retained record, so the figures that end up in your annual disclosure trace back to source rather than to a spreadsheet.
What this is not
It is not control. This builds visibility and evidence, not a system that changes your setpoints. Optimising plant with a guaranteed saving is energy engineering with formal measurement and verification, and it is a different profession from ours — we build the data layer those engineers need, and we say so rather than blurring it.
And it is not a promise about your numbers. What visibility is worth at your site depends on your site. Anyone who tells you the figure before seeing your data is quoting someone else's building.
What we will commit to is the chain: the collection, the dashboard, the alerting, the record. That is what we build, and a Reality Check is a week and a fixed fee to map what your existing meters and systems could support before anyone buys anything.
This is the part we do — the crossing from a demo to a system that survives production.