Data warehousing

A warehouse where each number your teams report on has one agreed definition and one place it comes from.

Where the effort goes

A warehouse project reaches into how numbers are defined well before it reaches the storage. These are the areas the work usually involves.

Agreed definitions
Writing down what each important number means, so that two teams reading the same figure read the same thing.
One source per number
Deciding which system a number comes from, so the same metric is not calculated two ways.
The model underneath
Structuring the warehouse so the agreed definitions are built into how a number can be read at all.
Loading and refresh
Getting data in from the source systems and keeping it current on a schedule people can rely on.
Reconciliation
Comparing the figures in the warehouse against the systems they came from, down to the individual record where that matters.
Access and control
Deciding who can read and change which numbers, and keeping the raw data under your own control.
Handover
Leaving your team able to operate the warehouse and to add the next number on their own.

The cost of one agreed number

The phrase buyers reach for is single source of truth, and the usual version of it is a technical project that ends with dashboards. That part can go smoothly and still leave two departments quoting different revenue figures to the same board. The disagreement was about what revenue means and who decides when two systems give different answers, and a new database does not settle either question.

The slow work is agreeing the definitions, because it is a conversation with the people who own the numbers. Each figure that decisions rest on gets one written meaning and one source system, settled by the people who will have to live with it, and the warehouse is then built so the number can only be read that way. Most of the budget and the calendar go on this, and it is often the first thing dropped when a project runs late.

How to start

The brief

Say which numbers your teams argue over and which systems they come from today. Name the person who can settle a definition when two sources disagree.

Starting small

An engagement usually opens by defining and loading one of your real, contested numbers, so you can judge the approach before the full warehouse is built.

Who does the work

Amos Weiskopf

Amos Weiskopf has been building data pipelines and the software around them for more than twenty years, so the person who agrees a definition with you is also the one who builds it.

More from Data engineering

Questions buyers ask

What is data warehousing?

It is bringing the numbers your business reports on into one place, with each figure given a single agreed meaning and a single source system, so that teams stop working from different versions of the same metric.

How much does a data warehouse cost?

It turns mostly on how many numbers have to be defined and how many source systems feed them. The figure is agreed after your written brief has arrived.

How is a single source of truth actually achieved?

By agreeing what each important number means and which system it comes from, then building the warehouse so those definitions govern how the number can be read. The storage is the easier half of the job.

Will our data stay with us?

Yes. The warehouse runs in your own cloud account or on your own servers, and your raw data stays with you.

How do we know the warehouse figures are right?

Every figure in the warehouse is checked against the system it came from. For the numbers that carry the most weight, the check is made on individual records as well as on totals.

Can our own team run it afterwards?

Yes, and the build is planned for that. The warehouse is set up around the skills your people already have, and the handover leaves them able to run it and to add the next number themselves.

Engagements by written brief.

Work begins from a documented problem statement, not a sales call.

Write a brief.

It goes straight to Amos.

Or email amos@tevelis.com, or book a meeting.