Skip to content

Data & Intelligence

Analytics platform: new questions, the same definitions

Reports answer the questions you knew to ask. The rest end up as exports and private spreadsheets, each with its own idea of “revenue”. An analytics platform gives people a place to explore, built on the same governed model as the reports: pick a defined metric, slice it by anything in the model, and get an answer that matches next Monday's report. We build it on the same definitions as your reports, so the answers match.
Talk through your dataSee three kinds of question ↓
Metrics · governedCompleted jobsBooked jobsRevenue (invoiced)Average ticketDimensionsLocationLead sourceServiceMonthBooked jobsby Lead sourceRiverside · last 90 daysSearch ads34Referral22Web form19Calls15Marketplace9same definition as the weekly reportrevenue = quotes acceptednot the governed Revenue (invoiced) · saved as a personal draft

Every new question becomes a new spreadsheet

“Which lead sources book the most jobs in Riverside?” isn't in the weekly report, so someone exports the data and builds a sheet. The next person builds another, with revenue counted a little differently. A month later there are three answers and a meeting about which one is right.

An analytics platform puts exploration on the same model the reports use. The definitions are already agreed, so a new question starts from the agreed numbers, and anything new goes through review before others rely on it.

Today, and with an analytics platform

Today

  1. 01New questions mean exports and private spreadsheets.
  2. 02Each sheet defines revenue, jobs or customers its own way.
  3. 03Answers can't be checked against the report.
  4. 04The person who built the sheet is the only one who understands it.

Every question gets answered, and none of the answers agree.

With an analytics platform

  1. 01People explore governed metrics by any dimension in the model.
  2. 02Answers match the reports, because they use the same definitions.
  3. 03New metrics are drafted, reviewed, then shared.
  4. 04Useful questions become saved views or dashboards.

Questions get faster answers, and every answer agrees with the others.

Three kinds of question

  • Search adslast quarter 29this quarter 34
  • Referrallast quarter 24this quarter 22
  • Web formlast quarter 13this quarter 19
  • Callslast quarter 16this quarter 15
  • Marketplacelast quarter 11this quarter 9

Booked jobs in Riverside by lead source. The same metric as the report, broken down in a way the report doesn't.

Four layers under every chart

The chart is the top layer. The three underneath decide whether it can be trusted.

  1. Events

    What's recorded

    Site, app and system events captured to a plan.

  2. Model

    Facts at a stated grain

    Jobs, invoices and customers, modelled once, with history.

  3. Metrics

    Defined once

    Revenue, completed jobs and the rest, each with one definition and an owner.

  4. Exploration

    Where people ask

    Charts, saved questions and dashboards, all built from the metrics.

Who decides what a number means

Define a shared metricThe metric's ownerProposed, reviewed, then published to everyone
Explore and save a questionAnyone with accessUses governed metrics; saved views can be shared
Draft a new calculationAnyoneVisible only to its author until reviewed
Publish a dashboardNamed editorsFrom governed metrics only

The parts it's built from

Each part is built and tested on its own, so you can start with one and add the rest.

  1. Instrument

    Analytics Instrumentation

    The events your sites and apps record.

  2. Model

    Data Transformation · Data Modeling

    Facts and dimensions, defined once.

  3. Define

    Business Intelligence

    • the model
    • agreed definitions
    • owners

    an undefined metric → a personal draft until reviewed

  4. Explore

    Dashboarding

    Charts, saved questions and dashboards.

Underneath every stepData Quality & ValidationEvery load checked before anyone explores it.

What changes from one business to the next

MetricsWhich numbers are governed first, and who owns each
AccessWho sees which locations, teams or financial data
ExplorationSaved questions and dashboards for each team
SourcesWhich systems feed the model, and how often

A common starting point is PostgreSQL for the model, dbt for transformations and metric definitions, Metabase for exploration and dashboards, and PostHog for web and product events. These are starting candidates, not requirements.

Is an analytics platform the right system?

Good fit

  • People ask questions the reports don't answer, often.
  • Private spreadsheets give different answers to the same question.
  • Someone will own the metric definitions.

Avoid when

  • A few fixed reports answer everything you ask.
  • The underlying data isn't settled yet. Start with the pipeline and definitions.

How we'd approach it

Most of the early work is defining the metrics, who may see what, and how far people should be able to explore.

  1. 01

    Understand

    As much as the problem needs

    We map how the work runs today: who does what, which systems hold the records, and where it breaks.

  2. 02

    Design

    Decided before it's built

    Ownership, rules, integrations and exceptions, agreed with you and written down with acceptance criteria.

  3. 03

    Build

    Connected to what you use

    We build the scoped system, connect it to the tools you already run, and test it on the cases that matter.

Related systems

The capabilities underneath

Questions about analytics platforms

Can't we just buy a BI tool?

The tool is the easy part. Without a model and agreed definitions underneath, it gives people a faster way to build disagreeing charts.

Who can create new metrics?

Anyone can draft one. It's visible only to them until the metric's owner reviews it and publishes it to everyone.

Will it replace our reports?

No. Reports keep their schedule; the platform answers the questions in between, from the same definitions.

How do people learn to use it?

Each team starts with saved questions built for them, and explores from there. Most never need to write a query.

Answer the questions your reports don't

Tell us which questions end up in private spreadsheets today. We'll show you how they'd be answered from one governed model.

Start a project