The Layers Don't Lie
Same problem. Same cycle. Different answer.
There’s a moment in every long project when you realize you’ve stopped building what you set out to build.
We started helping companies clean their MRO data. Duplicate part numbers. Inconsistent descriptions. The usual mess that accumulates when procurement systems grow organically across decades and acquisitions.
We were good at it. We built tools. We refined processes. We delivered clean data.
And then we watched it get dirty again.
Over the past few months, I’ve written about the hidden costs of duplicate parts, why data cleansing must happen before governance, and what it takes to prepare for EAM migration. Those articles addressed specific problems we see repeatedly.
This one steps back to share what living inside this problem has taught us — and why those lessons led us to build Ark.
Lesson 1: Cleaning Is Not Solving
Year three, a major refiner came back to us. Same problem. Different parts. Same root cause.
We’d cleaned 100,000 records. Eighteen months later, 15,000 new duplicates had crept in. Not because anyone was careless. Because the system allowed it. Because nothing prevented it.
That’s when we understood something uncomfortable: data cleaning is a service. Data prevention is a platform.
We’d been selling the service. The problem demanded the platform.
Lesson 2: MDM Platforms Solve the Wrong Problem
After the third or fourth cleanup cycle, clients would ask: “Should we invest in a Master Data Management platform?”
They’d seen the demos. SAP MDG. Informatica. Stibo. Impressive platforms with impressive price tags. Surely that’s the answer?
Here’s what we learned watching these implementations: MDM platforms are designed for IT, not operations.
They assume governance is a policy problem — define the rules, configure the workflows, train the users, enforce compliance. In theory, elegant. In practice, a mismatch.
The reliability engineer creating a part request at 2 AM doesn’t have time for your approval workflow. The contractor extending a turnaround doesn’t know your taxonomy. The planner juggling three systems doesn’t care about your data stewardship hierarchy.
MDM platforms enforce rules after data exists. They catch violations in reviews, generate reports, create tickets. By then, the damage is done. The duplicate is in the ERP. The purchase order is issued. The part is on a shelf next to its identical twin with a different number.
We’ve seen million-dollar MDM implementations become expensive shelfware — not because the software was bad, but because it was solving for compliance when the problem was friction.
Governance that fights the workflow will lose. Every time.
Lesson 3: Prevention Requires Presence
You can’t prevent bad data with a one-time project. You can’t prevent it with training. You can’t prevent it with governance documents.
You prevent it by being present at the moment of creation.
When someone enters a new part, that’s the moment. Not after. Not in a quarterly review. Right then.
Does this part already exist? Is the description following standards? Is the classification correct?
If you’re not asking these questions at the point of entry, you’re not preventing — you’re just documenting the mess for the next cleanup.
Lesson 4: Complexity Is the Enemy
We tried building comprehensive solutions ourselves. Platforms with every feature. Workflows for every scenario. Rules for every exception.
They failed. Not because they were wrong, but because they were too much.
The only systems that work are the ones that make the right thing the easy thing.
Not the enforced thing. Not the mandatory thing. The easy thing.
If finding the existing part is faster than creating a new one, people will find it. If standardizing the description takes one click instead of ten, people will standardize. If prevention is frictionless, prevention happens.
Why We Built Ark
These four lessons — cleaning isn’t solving, MDM platforms solve the wrong problem, prevention requires presence, complexity is the enemy — converged into a single principle: prevent at the source, with zero friction.
Ark isn’t a data cleaning tool. We’ve built plenty of those.
Ark isn’t an MDM platform. We’ve seen where those end up.
Ark is a lightweight layer that sits at the point of data creation. It checks, suggests, standardizes — before the record exists. When you search for a part that might exist under three different names, Ark finds them all. When you try to create a duplicate, Ark shows you the original. When your description drifts from standards, Ark corrects it.
No six-month implementation. No dedicated data stewards. No approval workflows that people route around.
Prevention, not remediation. Continuous, not periodic. Simple, not complex.
What We're Still Learning
We’re still learning. Every organization has its own definition of “good enough.” Legacy systems have gravitational pull. Change management moves slower than software development.
But we’ve learned the problem is solvable — not with bigger platforms or stricter policies, but with a fundamental shift in approach.
If you’re living this problem, I’d be curious how you’re approaching it. What’s worked? What hasn’t? Have you tried the MDM route — and if so, where did it land?
This is the first in a new series exploring what we’ve learned building platforms that prevent problems rather than clean up after them. More to come.
Next in the series: "The Math That Maintenance Managers Don't See" — tracing the real cost of a single duplicate, from entry to downtime.
About the Author
Author Bio: Raghu Vishwanath is Managing Partner at Bluemind Solutions and serves as CTO at KeyZane, a financial inclusion platform live in Central and West Africa. Over 30+ years across software engineering and technical leadership, he has watched the terms of specialization change — and learned that the only sustainable expertise is the willingness to build it again.