Your data will never by perfectly ready. Here is what to do instead

Insight  |  July 22, 2026

At a glance

  • Mateo Canarte-Toro, Head of Sales at COSOL Americas, draws on seven years of cross-industry experience to challenge the myth of data readiness
  • Poor data quality costs organizations an average of $12.9M per year - and 20-30% of MRO spare parts inventory routinely goes excess or obsolete before anyone acts.
  • A crawl-walk-run framework - assess, prioritize ruthlessly, then govern at the point of entry - replaces endless cleanup cycles with measurable near-term progress.
  • Sustainable data quality requires executive ownership, data stewards, and validation rules that stop bad records before they enter the system.

Article by Mateo Cañarte-Toro,
Head of Sales | COSOL Americas

For seven years I have sat across the table from maintenance managers, materials managers, and supply chain VPs in energy, utilities, transport, oil and gas, you name it. Different companies, different continents, different problems. And almost every time, the conversation lands on the same sentence:
We’d love to move forward, but our data isn’t ready yet...

I have heard it so many times I stopped being surprised. What I never stopped doing was asking one question back: “ready by what definition?” Nobody ever has a clean answer, and that is the whole problem. “Ready” is a moving target, and chasing it is how a two-year cleanup becomes a four-year one while the business falls further behind than where it started.

So let me say the quiet part out loud: waiting for perfect data is not a neutral decision. It is a financial one, and the meter is running the entire time.

The cost of waiting

20-30%
Of MRO Inventory in a typical plant is excess or obsolete

Gartner puts the average cost of poor data quality at $12.9 million a year. Read that again. That number does not politely wait for your cleanup project to wrap up before it kicks in. It is draining working capital right now, while your remediation sits on pause.

In MRO spare parts alone, I regularly see 20 to 30% of the inventory on a shelf go excess or obsolete before anyone acts. Holding costs pile up. Reorder points nobody has touched since go-live keep firing off purchase orders. Safety stock levels set five years ago keep tying up cash you could put to far better use.

I have seen both sides. The organizations that move fix what matters most, unlock the working capital, and let the cleanup run in parallel. The ones that freeze everything for a three-year master data project come out the other side worse off, because data does not sit still and wait to be tidied. It ages while you stall.

The lesson, every single time: the best operators do not wait for perfect. They iterate.

Master Data is a 10,000-piece puzzle

Odd analogy for a supply chain article, I know…. but stay with me.
"You don't start with a random piece from the middle!"

I have friends who tackle 10,000-piece jigsaw puzzles for fun. Watch a good one and you will notice they never start by grabbing a random piece from the middle and building outward. That is how you lose an afternoon and your sanity. No scale, no orientation, no idea where anything goes.

They start with the edges.

Build the frame first, because the frame tells you how big the problem actually is. Then they attack the high-visibility clusters, the bright red lighthouse, the one patch with an obvious color, while the frame holds it all together. Then they protect what is built so it does not collapse before the picture comes together.

Spare parts master data is that 10,000 piece puzzle. And most organizations are elbow-deep in the middle of the pile, wondering why nothing fits. So here is the frame: crawl, walk, run.

Image
Crawl

Find the edges

Image
Walk

Build Section Clusters

running icon
Run

Lock the Picture

Image

Crawl: Find your edges

"You simply cannot fix what you cannot see"

“We have a data problem” is not actionable. No leader can fund, prioritize, or champion a sentence that vague. So before you fix anything, size the thing.

Across our assessments at COSOL, six dimensions define the shape of almost every spare parts data problem:

Duplication
Same part, multiple records, because free-text entry was never governed. The most common one I see.
Standardization
No consistent naming convention, which makes the duplication nearly impossible to clean up.
Completeness
Missing attributes, no unit of measure, no commodity code, so downstream analysis cannot be trusted.
Integrity
Broken relationships, parts with no asset link and no active bin location.
Redundancy
Inventory that has sat on the books three to five years without a single transaction.
Accuracy
The inheritor of all the others, wrong values because nobody updated the parameters as the asset environment changed.
COSOL's RPConnect - Data Quality Dashboard Example
Above: COSOL's RPConnect - Data Quality Dashboard Example
"The best assessments don't just list the problems - they show you where the density is"
The point of the assessment is not a to-do list. It is a prioritization map, showing you which problems carry weight, which can wait, and which you can ignore entirely given where the business is headed. The good assessments show density, not just presence, because that is what turns “we have a data problem” into “these three areas are costing us the most, and here is what fixing them first looks like.”

Walk: prioritize ruthlessly, and never delete

Now pick your battles. This is where ISO 55000’s principle of proportionality earns its keep.
Ask yourself: "If this attribute were wrong....would it cost us money or cause a failure?"

You do not govern a $3 gasket the way you govern a $300,000 motor. One you replace on your lunch break. The other takes the plant down. Different risk, different consequence, different financial weight. Apply the same heavy rigor to every line in the catalog and you bury the people entering data in friction, and friction kills compliance before governance ever takes hold. Put the mandatory fields where decisions actually get made. Everything else is drag.

And when you consolidate duplicates, do not delete them. Merge them. BRG 1000 and Bearing 1000 are the same part, but each carries its own purchase history, lead times, and consumption pattern. Keep one, bin the other, and every future optimization decision runs on half the picture. Move the full history into the master record before you archive. Lose nothing.

IBM Maximo Inventory Optimization - Managing Duplicates via History Consolidation
Above: IBM Maximo Inventory Optimization - Managing Duplicates via History Consolidation

Run: stop bad data at the door

None of this holds without the right governance underneath it. The structures that survive share the same backbone: data stewards on day-to-day enforcement, data owners accountable in their own domain, a cross-functional council for organization-wide calls, and an executive steering committee with a real mandate.

"Top management shall demonstrate leadership and commitment with respect to the asset management system by ensuring that the asset management policy and objectives are established."

- ISO550001 Clause 1

Naviam DataStudio - In-GUI Data Validation
Above: Naviam DataStudio - In-GUI Data Validation
That last one is not optional. Without executive ownership, governance dies, and I have watched it die in as little as three months. ISO 55000 says top management must lead it, not delegate it, and the organizations that take that seriously are the ones still running two years later.
"Governance is not a committee meeting. It is a system of enforced standards at the point of entry."
Then govern at the point of entry. Once a bad record is in the system, it is already making decisions for you. Duplicate detection, validation rules, and rejection workflows that fire before a record is saved are what stop the mess from spreading, instead of you chasing it forever.

Four things to take with you

  1. Assess before you act. Know the full shape of the problem first.

  2. Prioritize ruthlessly, and match your effort to the risk and consequence of each item.

  3. Preserve consumption history through every merge and migration. It is the foundation of every smart inventory decision you will make.

  4. Govern at the point of entry. Clean data comes from keeping bad data out, not hunting it down later.

Perfect data is a destination that does not exist.
Good enough to start is a decision you can make this afternoon. And that decision is the line between the organizations recovering working capital and building real confidence in their data, and the ones still waiting at a starting line that never arrives.
Image
Profile picture of Mateo Canarte Toro

About Mateo Cañarte-Toro

Head of Sales | COSOL Americas
Mateo Cañarte-Toro is Head of Sales at COSOL Americas, with over seven years of experience in supply chain and asset management across energy, utilities, transport, and oil and gas. COSOL is a global asset management partner helping organizations in asset-intensive industries run their assets reliably, every day.

About COSOL

COSOL is built on one belief: in asset-centric industries, reliability is everything. We’re a trusted, data-led asset management partner for organisations around the world who can’t afford to fail. And known for our deep expertise, dependable delivery, and ability to keep critical assets performing at their best. 

The company recently celebrated 25 years in business, are Australian-owned and operated, and  recognised as reliable partners by their clients across the globe.