Utbenat: Data Trust

Introduction

Mitra:

A very warm welcome to Syntra Utbenat. This is the podcast where, in ten minutes, we take the words and concepts being thrown around in data and AI Sweden and break down what they actually mean together with Sebastian Mildgrim.

Sebastian:

Imagine you're in your car. You're heading to an important meeting at an address you've never been to before. You type the address into your GPS. You drive for a bit, and then it says "Turn right in 200 meters." You look over, and to your right is a dead end. Straight into a concrete wall. What do you do? Do you turn? Of course not. You trust your eyes. Not the screen. But what happens the next time the GPS says "Turn left"? Even if the road looks perfectly fine, you'll hesitate. You'll slow down. You'll double-check the street signs. Because your trust is broken. And trust is not something you repair with a software update.

Today we're going to talk about that GPS. But we're not going to talk about maps – we're going to talk about why trust in the numbers in your weekly reports is often lower than it should be. We're going to talk about Data Trust – or in plain terms: trust. This is a concept that often gets buried in technical jargon like governance and lineage. But it really comes down to one thing: Do you dare to turn right when the data tells you to? Or do you trust your gut more?

What happens when no one trusts the company's data?

Sebastian:

I think you'll recognize this situation. It's Monday morning. The leadership team gathers. Peter, the marketing director, throws up a sleek graph showing the sales forecast for the current quarter – up 10%. Everyone should be celebrating. But then Josefin, the head of logistics, raises her hand and says: "Hold on. In my system, where I can see what we've actually shipped from the warehouse, we're down 5% so far."

What happens next? Do you discuss how to grow sales? No. You spend the next 45 minutes arguing about whose Excel, whose data, is correct. That's the cost of low trust. Decision-making grinds to a halt. It's like driving a car and stopping at every intersection to get out and ask for directions – even though you have a GPS.

What is the difference between Data Quality and Data Trust?

Sebastian:

What's interesting here is that Peter in marketing might actually be right. His data might be technically correct but it doesn't matter. Because Josefin doesn't trust it. This is where we need to untangle two concepts: data quality and data trust. Quality is technical – is the number right? Trust is emotional – do I believe the number is right? You can have the world's best data quality, but if no one knows where the number comes from or how it was produced, very few people will dare to act on it.

So what happens when trust breaks down? Do people stop working? No, they just hold their cards closer to their chest. When Josefin doesn't trust the shared report, she starts building her own truth. She puts time into building an Excel that captures her reality. Suddenly there isn't just one truth in the company – you have an entire Excel society. Every department head is sitting on their own little micro-truth, on their own computer.

And this barely shows as long as each department handles its own corner. But the moment you try to produce a report that spans departments — like one showing how long a manufactured item sits in the warehouse before it's sold – it becomes impossible. You can't do it without data you can trust.

Why is low Data Trust especially dangerous when you introduce AI?

Sebastian:

Often the root cause is that the data is orphaned. When a number shows up in a report with no sender, with no one having signed off on it, it's like eating cold food from a completely anonymous hatch in the wall. You don't know who cooked it, or when – so of course you're not going to eat it. We lack ownership, and the when, where, and how of our data.

This is a question that is especially urgent right now – because you, like everyone else, want to start using AI. You want to plug in a co-pilot. You want to be able to ask an AI chat: how is the business doing? If you have low trust today, when you're looking at graphs that humans have produced, what do you think happens when a black box – an AI – spits out an answer?

Think of AI as an extremely polite but conflict-averse diplomat. It reads Peter's cheerful report about record-breaking sales, and it reads Josefin's angry report about warehouse inventory errors. A human would spot the conflict and say: wait a moment, something doesn't add up. But AI is trained to be helpful. It wants to make everyone happy. So what does it do? It tries to reconcile. Not out of incompetence or stupidity – but because it assumes that the data it has been given is correct. It has no reason to distrust it.

So it stitches together a narrative where both are right. It might say that sales are looking strong but that we're facing certain logistical challenges. That sounds professional. It sounds reasonable. But it could be a dangerously misleading conclusion – if the truth was that Peter's forecast was wishful thinking built on bad data. By assuming all data was correct, the AI has hidden the critical warning signal: that the warehouse hasn't shipped what the sales forecast implies. If you act on that diplomatic summary, the GPS has led you astray. That's why trust in data is the foundation. Without trust, all you get are diplomatic lies.

How do you build data trust in practice?

Sebastian:

So how do we increase trust? How do we get Josefin and Peter to stop arguing about numbers and start talking business? Many people – often in IT – want to solve this with large and expensive systems. We'll build a data warehouse. If we scan all our data sources and build end-to-end lineage for our data, trust will follow. Forget it. Trust is not built by systems. Trust is built by transparency. By control and accountability.

Imagine you're buying a house. Maybe an older villa. You go to the viewing. It's freshly painted. It's home-staged. Fresh cut flowers and scented candles. It feels right. The surfaces are perfect. But do you sign the purchase contract on the spot? Absolutely not, because you want to see the inspection report. Not because you think the seller is trying to deceive you on purpose. But because once the ink is dry on the contract, you own the house. You're the one who has to live with the consequences if there turns out to be dry rot in the basement or moisture damage in the attic. It's exactly the same with your reports.

Three key questions that build data trust

Sebastian:

Here's what you can do on Monday: Don't try to fix all your data, you can't. Pick one number. One single KPI that matters to your business. Maybe gross margin. Then create a simple inspection protocol for that number – a label that answers three key questions:

1. Who is the owner? Who is responsible for this number?
2. Where does the information come from? Which system – but more importantly, which process generated the data?
3. When does the data come from? Is it from this morning, last week, or even older?

Put that label next to the number in the report. If Josefin puts her name on it – I, Josefin, vouch that this number is correct – then Peter and the leadership team will probably dare to sign the contract. Because in the end, it comes down to accountability. When you do this, you go from guessing to knowing.

Summary

Sebastian:

And when you then feed certified numbers to an AI, you can start trusting the answers too. Trust in data isn't about having flawless systems. It's about the confidence of knowing what you're buying. Start with one number, assign an owner, create an inspection protocol. Do that, and you'll stop arguing about who's right and start steering the business instead.

Thank you for listening to today's episode of Syntra Utbenat. Want to talk more about today's topic, or just talk data? Reach out to us. We're on LinkedIn. Or contact us through our website, fiwe.se. Until next time – take care out there.