Skip to content
Zora

Dashboards and steering

Custom dashboards: your numbers, computed by your method

A dashboard is only worth something if you can click a number and see what it is made of. Without that, it produces arguments about whether the number is right instead of decisions about what it shows.

Why dashboards go unused

Most dashboards fail for a reason that has nothing to do with technology: nobody trusts the numbers on them. A director who doubts an indicator does not use it to decide — they ask for an export, and the dashboard becomes scenery.

The distrust nearly always has the same cause. The number is computed on data extracted yesterday, aggregated by a rule the tool wrote rather than the company, with no link to what it is made of. The natural reflex — “where does this figure come from?” — finds no answer on the screen.

The other failure is the average. Total revenue, average margin, average occupancy tell you nothing you can act on: they hide precisely what you are looking for, namely which unit, which site, which contract is dragging the result down. An average is an answer to a question nobody asked.

So a useful dashboard is defined by two things before any consideration of styling: it computes by the company’s rule, and every number stays clickable down to the lines behind it.

What the work covers

Your indicators, your definitions

Occupancy, net margin, cost price are defined at your company before they are computed. The definition is written into the system, visible, and it holds for everyone.

Every number traces to its source

Each amount opens onto the lines behind it: the entries, the jobs, the bookings. That is what turns a viewing screen into a decision tool.

Analysis at the useful grain

Profitability unit by unit, performance site by site, margin job by job. The total is still there, but it is no longer the only level you can read at.

Data brought together

What comes from operations, from accounting, from third-party tools: reconciled on stable identifiers, with discrepancies flagged rather than smoothed over. A silent reconciliation is a wrong one.

The exception, not the state

What moved, what falls outside the norm, what is approaching a threshold — brought to attention without anyone having to open the screen. A dashboard that requires a daily visit receives none.

Exports where they are legitimate

A board, a banker, an accountant needs a file. The export comes from the same calculation as the screen, which avoids the classic meeting where two versions of the same figure are circulating.

The signs you need more than an export

None of these is fixed by plugging a visualisation tool into the same data: they are fixed by reworking the calculation.

  • The question “where does this figure come from?” has no answer on screen.
  • Two departments present two versions of the same indicator.
  • Monthly consolidation takes several days of rework.
  • Decisions are made on averages that hide the extreme cases.
  • A BI tool was installed, then abandoned for lack of trust.
  • The information arrives late enough that the decision is already made.

Frequently asked questions

How is this different from a BI tool?
A BI tool displays what it is given: if the calculation rule is wrong or the source data inconsistent, it produces a convincing chart from a wrong number. The work described here starts upstream — establishing the definition of each indicator, reconciling the sources, making the calculation replayable — and the display comes after. That is also why the number stays clickable down to its source, which a dashboard plugged into an export cannot offer.
Do we need an ERP before considering dashboards?
No. You need data, in some form — a management system, an accounting tool, spreadsheets kept seriously. The decisive question is not where the data sits but whether it can be reconciled: two sources with no common identifier cannot be consolidated, and that then becomes the first piece of work.
How current are the figures?
That is decided indicator by indicator. An operational measure can be computed on demand against live data; an indicator that depends on an accounting close is only correct after that close, and showing it earlier would mean presenting an estimate as a result. Either way, the screen states the date it is drawn to.
Who can see which figures?
Permissions are defined with you, and it is often a sensitive question: margin per job need not be visible to the same people as activity volume. The system enforces the rule rather than relying on separate files, which is the usual and unreliable way of handling confidentiality.
What if our calculation method changes?
A definition changing is a normal event. The rule is edited in one place, with its effective date, and history stays readable under the rule that applied at the time. Recomputing the whole past under the new rule would produce figures that no longer match what was decided or communicated.

Related work

The engagements in the portfolio where this work is visible.

Let’s talk about your project

Describe what you are trying to build. We answer with an honest first read on feasibility and scope — even when the answer is that you do not need us.

Tell us about your situation