Every systems implementation I have ever been part of has a data problem. Every single one. The nature of the problem varies — master data that hasn't been cleaned in a decade, transactional data spread across six legacy systems with no agreed source of truth, organisational data that nobody owns and everyone uses — but the underlying dynamic is the same.
Nobody wants the data.
What "Owning Data" Actually Means
When I say nobody wants the data, I mean something specific. Technically, organisations often have data warehouses, data lakes, analytics teams. What they don't have is someone in Finance who genuinely owns the Chart of Accounts and is accountable for its quality. Or someone in Supply Chain who owns the Material Master and treats it as a living asset to be maintained, not a database table to be queried.
Data ownership is a business responsibility, not a technology responsibility. IT can build the systems. IT can establish the standards. IT can create the tools for maintenance. But the data itself — its accuracy, its relevance, its completeness — belongs to the business. And the business, in my experience, would usually rather pretend otherwise.
The Consequence of Ungoverned Data
What happens when nobody owns the data? Several things, all of them bad:
- Reports nobody trusts. I have been in organisations where three different versions of the same metric — revenue, headcount, inventory — existed simultaneously in three different systems, each with a different number. Decisions get made on gut feel because nobody trusts the data.
- Implementations that go wrong. A new system is only as good as the data you load into it. Garbage in, garbage out is a cliché because it's true. I've seen SAP implementations where the data migration took twice as long as planned and was the primary driver of go-live delays.
- Analytics that mislead. Machine learning and advanced analytics are completely dependent on data quality. An organisation that hasn't invested in data governance is not ready to get value from AI. Full stop.
The Gift Nobody Wants to Receive
Here's the metaphor I use: data stewardship is like inheriting a large, important, valuable, and completely disordered archive. The person who takes it on has a thankless job — sorting, cataloguing, discarding, standardising — and they'll be doing it while the business continues to generate new data at pace. The archive never stops growing.
Nobody volunteers for this. In most organisations, data stewardship is either undefined (so it belongs to everyone and therefore no-one), or it's buried in IT as a technical function, or it's handed to a junior analyst with no authority to enforce standards.
The result is that data quality degrades steadily until there's a crisis — usually a large implementation that forces the issue — and then there's a frantic, expensive scramble to clean up years of neglect in a short window.
What Good Data Governance Looks Like
I'm not going to describe a complex governance framework here. Most organisations don't need a framework — they need accountability. Specifically:
- Named business owners for each critical data domain, with explicit accountability for quality
- Agreed data standards — what does a "complete" customer record look like? What's the standard for vendor master data?
- A regular cadence of review — not a one-time cleanse, but an ongoing process of monitoring and correction
- IT as an enabler of that process, not the owner of it
In twenty years of running technology organisations, I have never seen a company suffer from having too much data discipline. I have seen countless companies suffer from having too little.
The Right Time to Start
The right time to address data governance is before you need it — before the implementation, before the migration, before the analytics programme. The second best time is now.
If you have a major system implementation on your roadmap, put data governance at the top of the pre-project work. Not the bottom. The investment you make before the project starts will pay back many times over in reduced migration risk, faster go-lives, and systems that people actually trust to run their business.
