Data engineering · Health check and strategy
Two weeks to know what you actually have
A short, fixed-scope assessment of the data estate: where it lives, who owns it, how it moves and how far it can be trusted. The output is a costed plan, and sometimes the news that it is in better shape than feared.
Where this starts
Nobody holds the whole picture
In most organisations, the knowledge of where data lives, what it means and which copy is right is spread across a handful of long-serving people, none of whom holds the whole map. Every data initiative begins by rediscovering it at some expense, then loses it again when the project closes. The result is a series of programmes that each start from nothing.
What we are usually asked
- Which system is actually the source of truth for customers, products or revenue
- Why two reports built from the same data disagree
- Whether our data can support the reporting or AI work the board wants
- What it would cost to fix, and in what order
- Whether a new platform is needed, or the current one is simply under-used
- Who is accountable for which data, today, in practice
Our position
If the estate is in better shape than you feared, we would rather tell you in two weeks than sell you two quarters.
What the work covers
Mapped, scored and sequenced
Fixed scope and a fixed fee, because an assessment that grows to fit the available budget has stopped being an assessment.
Source and flow mapping
Every system that creates or holds data the business relies on, how data moves between them, and where it is copied, transformed or quietly re-keyed along the way.
Source of truth
For each core entity, whether customer, product, order or employee, which system is authoritative, which hold copies, and exactly where they currently disagree.
Quality sampling
Completeness, validity, duplication and freshness measured on real samples rather than estimated in a workshop, so every finding arrives with a number attached.
Ownership in practice
Who actually maintains each dataset today, as opposed to who is named on an organisation chart, and the datasets nobody maintains at all.
Readiness against intent
The reporting, automation and AI the business is planning, tested against what the data can support now, so ambitions and foundations are compared honestly.
A sequenced plan
Remediation ordered by value and dependency, with effort ranges, separating quick wins from structural work so the order of the programme can be defended.
How the map stays current
An inventory that keeps itself true
A data map drawn by hand is out of date within a quarter. These keep it accurate without a second assessment.
- New sources
- Tables and sources detected as they appear on the platforms we mapped, so the inventory grows with the estate instead of being redrawn every year.
- Profiling
- The quality measures from the assessment rerun on a schedule, so improvement or decline shows as a trend rather than a fresh surprise.
- Lineage
- Flows captured from the systems that actually move data, rather than drawn once in a diagram and left to diverge from reality.
- Owners
- Owners recorded against datasets where the work happens, with gaps flagged openly rather than filled in by assumption.
- Scorecard
- The scores the plan was built on reported periodically, so progress is measured against the starting point rather than asserted in a steering meeting.
What you are left holding
- A map of sources, flows and copies across the estate
- A named source of truth for each core entity
- Quality findings with numbers attached
- A costed plan sequenced by value and dependency
- A baseline scorecard to measure progress against
Tell us the question your data cannot answer
That is usually the right place to start the map, and within days it shows whether the problem is the data or the reporting.