In digital transformation, it is not those who have implemented the most expensive AI system who win. It is those who know what data they already have, where they don’t have it, what they are missing, and where the contradictions lie. This is not a story about artificial intelligence. It’s a story about the mess you keep in your closet and pretend you don’t see.
At first glance, everything seems fine:
- CRM is working,
- SAP isn’t crashing,
- reports are flying.
But when you ask two people from two different teams about the same indicator, you get two different answers. Because one takes data from Germany and the other from Italy. Because one filters by quarter and the other by month. Because one has data from last week and the other from last year.
That’s how it is when things are nice,
but it can also be different: one has a different definition of the indicator than the other, one has incomplete data, and the other has contaminated data in a completely different structure. All true. All false.
Meanwhile, what you do with data is like a restaurant kitchen. The customer only sees the plate with the food. They don’t see the burnt pots, they don’t think about whether the chef cut the peppers with the same knife as the chicken.
But if the kitchen is dirty, someone will eventually get food poisoning.
In data, this poisoning takes the form of bad decisions, delays, inconsistencies and endless analyses.
That’s why this text won’t be about Big Data.
It will be about Big Mess.
Smooth slides, messy data
Many organisations declare that “data is the new oil”, that “we are data-driven”, that “analytics is our priority”. Give me a break.
In the daily work of an analyst, project manager or operational ninja, data often looks like Excel spreadsheets glued together with tape.
Columns without labels, dates in different formats, text fields containing everything from phone numbers to customer complaints.
And above all: a lack of analytical thinking and a lack of doubt.
An organisation may have millions of records, but none of this information will pass the test: “Can a decision be made based on this?” Because decisions are not born out of noise. They are born out of clarity. From data that is consistent, complete, up-to-date and… understandable.
We have plenty of data… just not the right kind
“Can a decision be made based on this?” - this simple question turns into a corporate nightmare.
It turns out that no one has a map of the data. Data is everywhere and nowhere. It’s in transaction systems, CRMs, SharePoints, Excel files, emails, Slack messages, and Jira tickets.
And if you’re lucky, you’re drowning in a datalake, crushed by containers from data warehouses (of course, containing other data and incomplete data).
On the other hand, we have data that the organisation doesn’t have yet, but desperately needs. For example: “Why do customers abandon their shopping carts?”, “Which application features are used the least?”, “Which salesperson frustrates customers the most, even though they have the best sales results?”
(I’m simplifying with rather obvious examples, but if your company really lacks such indicators, stop reading this text and report them for implementation as soon as possible).
Ops vs Chaos: The front line with data
It’s not the AI Team, it’s not marketing, it’s not the innovation department. It is the Operations team that should be asking the question: “What data do we have? And can we trust it?”
The Ops team should build data maps. Create source registers. Mark which data is “dirty”, which is “incomplete” and which is “unclear without context”. And which data we collect unnecessarily.
Different country, same… different shit (and customer definition)
In international companies, data can be like folk tales: roughly the same, but with completely different endings. Data standardisation is not just a technical procedure. It is a fundamental conversation about what “truth” means to us. Data is not numbers. Data is decisions.
Copy-Paste Enterprises
Duplicating data is not just a mistake. It is a systemic habit. Instead of going to the source, we copy data from the last report. Because it’s faster. Because it works. Because “we’ve always done it this way.” And then we end up with five versions of the truth – each in a different format, each accompanied by the famous “don’t touch it, it works.”
Organisation as a data gardener
We need Data Governance.
But not in the form of a 200-page policy that no one will read. Only as a real, living organisational practice that, like watering plants, can be boring and repetitive, but is essential if we don’t want everything to wither away.
The Ops team can be the one to start
Not because it has the most authority. But because it has a panoramic view. Ops sees how data flows through the organisation, where it gets stuck, where there is no owner, and where there is no validation. Ops can be the first to create a data registry, establish critical points, propose a steward system, and initiate the process of rebuilding trust in information.
But the key is not for Ops to always do everything. The key is the moment when the Ops team can say: “It’s working – now it’s yours.”
When governance is in place, the role of Ops becomes to pass the baton: to product teams who take responsibility for metrics, to analysts who document sources, to business owners who approve definitions. It’s like handing over a garden – already planted and tidied up – to those who will continue to nurture it.
A mature Ops team doesn’t keep work to itself. It knows when to step aside and make room for others. But before it does, it tidies up like no one else can.
Toolkit for data with a fever
- Feature Store – a single source of truth for data attributes
- Data Catalog – inventory of data assets
- Data Contracts – establishing rules between teams
- Governance Light – rules without overregulation
Best practice, or how not to go crazy (or bankrupt)
Well-functioning organisations do not have the most data. They have the most clarity. They know where the data comes from, who is responsible for it, and what decisions can be made based on it.
Data as a bridge, not a wall
It is not data that divides us. It is a lack of understanding of what data is. Every organisation has the potential to build bridges between teams, countries and systems.
But first, we must abandon the fiction that data will sort itself out. It will not. Not without people. Not without a strategy. Not without groundwork.
Data is not tables. It is decisions, emotions, reputation. It is the trust that an organisation builds in itself.
If you see the same metric in three versions every day, you don’t have a systemic reporting problem. You have a problem with reality.
Digital transformation does not begin with the implementation of AI, but with the question: “Do we even know what we know?” Can we describe our reality in such a way that another team, in another country, in another system, sees the same thing we do?
If the answer is not yet, then we are on the right track.
Because awareness of the mess is the first step towards order. And Ops, as always, is on the front line. Not to patch holes. But to build a structure that won’t fall apart when the next fad for “something with data” comes along.