A written answer before a platform or a model

Know whether your data can carry what you want to build.

We assess one proposed use case against its sources, definitions, owners and access, then put the dependencies and recommendation in writing.

Readiness is a decision, not a score.

A percentage cannot tell you whether the source is owned, whether two teams mean the same thing by a figure, or what an assistant would be permitted to read. The assessment follows one proposed use case through those questions and puts the answer, the gaps and the next decision in writing.

Each figure it needs is traced to a source and an owner, and the recommendation is written down.

Most assessments start with one of these.

We follow each figure the idea needs back to the system it comes from and the person who owns it.
We write down where the two definitions differ, and the business decides which one it will stand behind before either feeds a model.
We check who can read each source today, because an assistant usually sees whatever the person asking it can see.
We list what the build depends on and who owns each item, so a date can be set once those are settled.

We follow each figure the idea needs back to the system it comes from and the person who owns it.

We write down where the two definitions differ, and the business decides which one it will stand behind before either feeds a model.

We check who can read each source today, because an assistant usually sees whatever the person asking it can see.

We list what the build depends on and who owns each item, so a date can be set once those are settled.

Which use case would you want a written answer on before you spend money building it?

Talk it through

An assessment of the work beneath the idea.

Built to be checked

One use case gives the review a boundary and a decision it can actually answer.

Set the use case
We agree what is being assessed and what a useful result would need to show.
Trace the sources
We record the data domains, their owners, the definitions and where lineage stops.
Check access
We review who may read the information and what an assistant or report would inherit.
Write the decision
The recommendation names dependencies and risks, including when the answer is not yet.

What we need to reach a useful view.

Messy data is a reason to assess it. The review needs a specific question and people who can explain the sources.

One proposed use case or report to assess, rather than an open-ended AI programme.
Access to the people who own the relevant systems and business definitions.
A decision-maker who can confirm which meaning the business will stand behind.

If it is not there

If a source, owner or definition cannot be confirmed, the gap becomes a named finding and dependency in the written assessment.

What you have at the end.

The assessment document: It fixes the use case and scope, records who signed off the view, and gives the recommendation.
A map of what the answer rests on: Sources, owners, definitions, lineage and access are named where they can be established.
A dependency and risk register: Each dependency has an owner; each risk says what it affects and when it needs review.
A decision you can use: The written recommendation can say proceed, resolve a dependency first, or not yet.

The assessment is a document you keep.

It sets out the scope, the sources and owners behind the use case, where definitions or access need work, and our recommendation, which can be “not yet”.

See the Data Platform & Analytics work

Where this page sits.

Our work sits in three layers, and each one depends on the one beneath it. The page you are reading is marked.

You are hereData Readiness Assessment

AI

Data

Data Platform & Analytics
You are here Data Readiness Assessment

Foundations

Start with the question the data has to answer.

Tell us what you want to build or report, and we can scope the assessment around that one decision before anybody commits to a platform. The Fortnight looks at data readiness as one of its areas across your whole estate, and this assessment goes further into one proposed use case.