Analytics is where the platform's numbers are defined, shaped and published. Not a reporting tool bolted onto a database: a governed layer over records that already exist.
The chain
Everything here follows the same three steps, and understanding them makes the six surfaces obvious.
A measure is a definition. What "revenue" means, precisely: which records, which field, which arithmetic. Defined once.
A widget is a measure given a shape. The same revenue measure as a number, a trend, a breakdown by owner, a cross-tab against region.
A board is a set of widgets somebody looks at.
The important property is the first step. Two people asking about revenue get the same number, because there is one definition rather than two spreadsheets.
The six surfaces
| Surface | What it is for |
|---|---|
| Measures | Defining what a number means |
| Studio | Turning a measure into a widget |
| Library | Everything already built |
| Boards | Assembling widgets into something people watch |
| Reports | Tabular output, scheduled and exportable |
| Explore | Templates and the vocabulary, for learning |
It knows your schema
The studio does not ask you to write SQL or to know table names. It reads the platform's own catalog of what exists, which has two consequences worth knowing.
Your custom fields are there. Anything your organisation added to a record is available as a dimension or a measure without anybody adding support for it.
It only offers what actually works. A field's type decides which arithmetic applies, and how many distinct values a dimension has decides which charts fit. You cannot sum a status, and a dimension with forty values is not offered as a pie.
That second point is a deliberate constraint rather than a limitation. Most bad charts are made by tools that let you build them.
Five ways to shape a question
Beyond a plain widget, there are five authoring engines, and each answers a different shape of question.
Pivot for a cross-tab: this by that.
Funnel for movement: how records travel through stages, and how long they take.
Analysis for one measure broken down.
Correlation for whether two measures move together.
Affinity for what goes with what.
They share one treatment: scope on the left, the live result on the right, and a bar that describes what you are looking at in words. Each saves as a real widget that can go on a board like any other.
What makes this different
The usual arrangement is a BI tool pointed at a copy of your database. That copy is behind, the definitions live in the tool rather than in the business, and the numbers drift from what the operational screens say.
Here the analytics read the same records the product runs on. There is no copy, no sync window, and no second definition of a customer.
The practical test is one people run constantly: does the number on the dashboard match the number on the screen where the work happens. Here it has to, because it is the same row.
Who does what
Somebody governs the measures. Usually one person per area. This is the part that matters most and needs the least ongoing work.
Anybody builds widgets, from the studio, without asking for help.
Somebody owns each board. Boards without an owner become graveyards.
The failure this prevents
Almost every organisation ends up with the same problem: three answers to one question, from three spreadsheets, and a meeting spent reconciling them rather than deciding anything.
The cause is never arithmetic. It is that nobody wrote down what the word meant, so three people each decided reasonably and differently.
A governed measure is the fix, and it is worth understanding that this is the whole point of the surface. Charts are the visible part; the definition underneath is the part that stops the meeting.
Where to start
Read Measures next if you are setting this up. Everything downstream is only as good as the definitions, and it is the one part that is genuinely worth getting right first.
Read The Studio if the measures already exist and you want a chart.
Read Reading a Widget if you are looking at a number you do not trust. That piece covers drilling into the rows behind it and finding out why it is empty.
A first hour
Open Explore and read the vocabulary. Fifteen minutes, and the rest of the surface stops being opaque.
Open Measures and read what already exists. You will find most of what you need is defined.
Build one widget for a number you can verify by hand. Trust comes from checking once, not from being told the engine is correct.
Put three widgets on one board for a question you actually have.
The vocabulary
Four words the rest of the surface uses constantly and rarely defines.
Measure. A governed definition of a number.
Dimension. Something to break a measure down by: owner, stage, month, region.
Scope. Which records count, and over what period. Two separate ideas that get conflated: filters decide which records, the window decides when.
Cardinality. How many distinct values a dimension has. It sounds academic and it decides which charts you are offered, so it is worth knowing the word when the studio narrows your options.
What it costs
Nothing per chart, per board or per query. Analytics reads records you already have, so there is no metering on looking at them.
The AI describe bar spends credits, because it is a language model writing a sentence. Everything else on the surface is free however often it runs, including scheduled deliveries.
What it will not do
It will not fix your data. A measure over incomplete records produces a confident wrong number, and the studio has no way to know.
It will not decide what matters. Nothing here suggests which measures your business should care about.
It will not replace the operational screens. A pipeline is worked in Sales, not on a chart of the pipeline.
Did this answer your question?
No, ask a person