The product

What Metrica does with your data, step by step — and what it does not do

Metrica is a management intelligence and operational control platform. It reads what your existing systems already produce, puts one set of definitions on top of it, and carries the result through to a named person doing something about it — then checks whether that worked. This page is the long version of that sentence.

How it works

Metrica has one shape, whatever the department or the industry. Four steps, and the fourth feeds the first — that loop is the product, not the reports it produces.

1 — Consolidate

What your systems already produce is mapped once into a consistent set of records, so a field means the same thing in every view. Nothing is re-keyed, and nothing is written back to the source system.

2 — Define

Each management measure gets a written definition: the formula, the fields it reads, the period it covers, and whether the data behind it was complete. A number without those four is an opinion.

3 — Act

A written rule decides what is outside limits. That becomes an exception with a severity and a trace to the rows behind it, and an exception can become an action with one named owner and a due date — visible, and countable when it is not done.

4 — Verify

The same measure is taken again after the action. If it did not move, the exception returns and so does the question. Management control is the loop closing, not the report being sent.

Everything below is detail on one of those four steps, and the page ends with what the model deliberately refuses to do.

Getting the data in

How a source is reached is designed per deployment, not fixed by the product. No two customers run the same systems, so the method follows what yours can actually do: a scheduled export, a file dropped in a folder, a connection agreed during implementation — whatever produces the data the definitions need, reliably. What is standard is the discipline around it.

Metrica reads; it does not write back to your source systems. Whatever the route, an import is visible before it changes anything: you see what will be accepted, what will be rejected, and why.

The import path in the application today is a short, visible sequence:

1. Upload a CSV file, within a size and row limit.
2. Map the columns: Metrica suggests a mapping, and you confirm or correct it. Nothing is guessed silently.
3. Preview what will be accepted and what will be rejected, row by row, with the reason for each rejection.
4. Import, with duplicate detection: a row that looks like an existing record is shown to you before anything is written, and nothing is overwritten silently.
5. Take the error report: a downloadable CSV of every row that could not be imported, with the reason, so it can be fixed at the source.

Every row that cannot be read is rejected with a reason and handed back to you as a file, so the fixing happens at the source rather than inside the tool.

What goes into a definition

The reason a management report and a spreadsheet disagree is almost never the arithmetic — it is that the two were counting different things under the same word. So an indicator in Metrica is not a query someone wrote once. It is a definition with four parts:

That last part is the one most systems get wrong, and it is the rule this product refuses to break: missing data is never shown as zero. A gap and a zero are different claims about the world, and confusing them is how a manager ends up making a confident decision on nothing.

This is not a promise for later — it is how the working application behaves today. Its overview says, in its own words, that cost analytics are not active yet because no expense records exist, “so Metrica reports nothing rather than zero.”

How an exception is raised

A dashboard that shows everything shows nothing: if forty numbers are green and three are not, the three are the work, and they are the three nobody finds at 7 a.m. Metrica is built the other way round — the screen leads with what is outside limits.

An exception is raised by a written rule over the data: a promised date passed without dispatch, a backlog growing for three periods running, a cost line moving beyond its usual band. Each exception carries a severity, the rule that produced it, and the records behind it — so “why am I looking at this?” always has an answer, and an exception can be dismissed with a reason rather than deleted quietly.

How an action works

An alert that ends at an alert is a notification, not management. Any exception can be turned into an action: a sentence saying what will be done, one named owner, a due date, and a status. The register shows what is open, what is overdue and what is closed, with the history of who changed what — because the failure mode of corrective action is not that nobody knew, it is that nobody owned it.

How verification works

The loop closes by measuring the same indicator again after the action. If the backlog fell, the action worked; if it did not, the exception returns and so does the question. Without this step an action register slowly becomes a list of things people said they would do.

One order, followed through

An illustration of how the four layers connect — deliberately a plain case, not a screenshot of the running system.

What it does not do

A product is also defined by its refusals. These are deliberate.

Security and hosting are described on the Security & Trust page. What the application stores, and for how long, is in the Privacy Policy.

How this usually starts

Metrica is introduced one management need at a time, not as a platform rollout. A typical engagement has four stages: a discovery of what is actually measured and where the data lives; a business audit that says what is worth controlling first; an implementation of that slice; and continuous improvement as the next need appears.

That order matters. Starting with the software means discovering in month three that the export everyone trusted was wrong since January.

Questions

If you are not sure whether this fits, describe the management question you cannot currently answer and we will tell you whether Metrica is the right tool — including when it is not. meavida@metricahq.com.