Skip to content
Fona

Service

Data and BI

We bring together the information scattered across systems and spreadsheets, historise it properly and turn it into dashboards you can defend in a board meeting.

Let's talkSee case studies
Abstract illustration of an ascending bar chart with a trend line

The problem

Two reports for the same month with two different numbers

Every department has its spreadsheet, every spreadsheet has its own definition of "active customer", and month-end turns into an argument about who is right instead of what to do.

Signs this is you

  • The monthly report is assembled by hand and takes three days
  • Nobody knows which figure is correct when two reports disagree
  • No history: when a value changes, what it said before is gone
  • Business queries run straight against the production database
  • Every new question from management requires development work
  • Decisions get made on gut feel because the data arrives too late

What we do

What we do

Data warehouse

One place where ERP, CRM, web and spreadsheet data converge, with scheduled loads, quality checks and traceability back to the origin of every figure.

Proper historisation

Modelled with slowly changing dimensions (SCD type 2) so you can answer "how did this look in March" without relying on somebody having saved a copy.

Dashboards

Reports per area with the agreed metrics and their written definition alongside, so the conversation stops being about the calculation and starts being about the business.

Metric dictionary

A catalogue where every indicator has an owner, a formula and a source. It is the least glamorous part of the project and the one that prevents ninety per cent of the arguments.

SCD 2
versioned history, not scattered snapshots
1 source
of truth for the whole organisation
10
business questions as the starting point

How we do it

How we do it

  1. Questions before tables

    We start from the ten questions management needs answered every month. The data model is designed to answer them, not the other way round.

    DeliverableQuestion and metric catalogue

  2. Model and first load

    Model design, first sources connected and historical load, validated against your current reports so every difference can be explained.

    DeliverableWarehouse with priority sources

  3. Dashboards

    We publish the first reports and refine them with the people who will use them daily, until they genuinely replace the spreadsheet.

    DeliverableDashboards in use

  4. Automation and expansion

    Scheduled loads, data quality alerts and new sources in waves, prioritised by the return each one brings.

    DeliverableAutomated loads and alerts

Technologies

  • PostgreSQL
  • MySQL
  • SQL
  • ETL
  • Laravel
  • Python
  • Dimensional modelling
  • Dashboards

FAQ

What people ask us

Do we need an expensive BI tool?

Almost never. Most of the value sits in the data model, not the visualisation layer. We start with what you already have or with an open-source option, and a licence only comes up when there is a need that justifies it.

Can you work on our existing database?

We can read from it, but not analyse directly against production: heavy queries and daily operations do not coexist well. That is why we separate the warehouse, which also lets us historise and combine sources.

How long before we see something?

The first genuinely useful dashboard usually lands between week six and week ten, depending on how many sources there are and the state of the source data. Cleansing is always the part that surprises people.

What if our data is messy?

It is. Everyone’s is. Part of the work is measuring how messy, deciding what gets fixed at source and what gets fixed on load, and surfacing the quality level so nobody decides on a doubtful figure without knowing it.

Next step

Half an hour well spent

Walk us through the problem on a short call. You leave with a first read on how we'd approach it and what it would involve — no commitment, no sales deck.

Let's talkHow we work