Approach
Five steps take each data domain from discovery to scale
Every step ends in something you can hold. If a step has no output, it did not happen.
| Step | What happens | Output |
|---|---|---|
| 1 Discover | Profile the sources and agree what a customer is. We profile each source system and work with the business until there is one agreed definition of the entity in question. Most later disputes start here if this step is skipped. | Data profile |
| 2 Design | Set match rules, ownership and quality thresholds. We set the rules that decide when two records are the same, who owns the result, and what quality level counts as acceptable. | Target model |
| 3 Build | Configure the golden record and the pipelines. We configure the golden record and the pipelines that feed it, and ship a release that runs on your real data. | Working release |
| 4 Prove | Test with business users and tune the rules. Business users test the results against records they know, and we tune the rules until they sign off. | Signed-off results |
| 5 Scale | Repeat the pattern for the next domain. We write down what worked and repeat the pattern for the next data domain, which starts again at the first step. | Domain playbook |
Which domain comes first?
Tell us where your data hurts most. That is usually the right place to begin.
Start a conversation