About the episode
Många företag bränner miljoner på att "städa" data som ingen någonsin tittar på, samtidigt som det affärskritiska drunknar i mängden. Det finns ett utbrett missförstånd i datavärlden: att god datakvalitet innebär att alla fält är korrekt ifyllda, alla poster kompletta och alla mätare gröna. Men verkligheten är en annan.
I det här avsnittet av Utbenat bryter Sebastian ned vad datakvalitet faktiskt innebär och framför allt vad det inte innebär. Svaret handlar om affärssyfte, inte perfektion. Och det kan vara skillnaden mellan att investera klokt i er data och att betala dyra pengar för att damma vinden som ingen någonsin ser.
In this episode, you’ll learn:
Voices in this episode
Listen to the episode
Transcript of the episode
Introduction
Mitra:
Welcome to Syntra Utbenat. This is the podcast where in ten minutes we break down the terms and concepts that are used across data and AI and explain what they actually mean together with Sebastian Mildgrim.
What is data quality, really?
Sebastian:
Imagine you are hosting a big dinner tonight. Guests arrive in two hours. Your home is a bit messy. What do you do? You tidy up the hallway, clean the table and make sure what people see looks good. You do what is needed to create a clean and welcoming impression.
But imagine you delay the dinner by three weeks to clean the attic. You go through old moving boxes with clothes your kids have outgrown hidden behind holiday decorations. Guests have to wait but when they arrive even the attic is spotless. It sounds unreasonable. Yet this is exactly how many companies treat their data.
Today we will talk about those boxes in the attic. We will talk about data quality.
The problem: time spent on the wrong data
Sebastian:
Companies can spend millions cleaning information no one will ever use while business critical data gets lost in the noise. We have been told that data is the new gold. That sounds good but it has also created pressure.
Johan in IT buys expensive tools that scan databases and find issues: “40,000 customer records are missing a valid postal code.” Johan panics. Projects start. Consultants are called in. Data needs to be cleaned. But no one asks the most important question: Which customers are these?
If they bought once in 2012 and never returned it does not matter if the postal code is wrong. Yet we fix it anyway because the metric should turn green.
The myth of perfect data
Sebastian:
We need to remove a myth. Data quality is not absolute.
There is no such thing as perfect data. There is only data that is good enough for its purpose.
Same data, different requirements
Sebastian:
Take Josefin in logistics. For her the delivery address must be correct. If it is wrong the truck goes to the wrong place. That costs money. Here quality must be close to 100 percent.
Now take Peter in marketing. He wants to understand where customers are located. If a street is slightly wrong it does not matter. He only needs the city.
Same data. Different purposes. Different requirements. If Peter has to wait for perfect address data the business slows down.
Treating all data the same
Sebastian:
Another issue is that all data is treated as equally valuable. In reality most companies sit on large volumes of data and only a small part creates value.
Log files, old transactions and fields that were created years ago and never used. When rules say all fields must be filled we force the organisation to maintain data that has no value.
We pay to store and improve data that creates no return. It is like renting a warehouse full of old furniture and paying someone to clean it every week.
The measurement trap
Sebastian:
"What gets measured gets done". That is true and dangerous when we measure the wrong things.
If a system shows that 5,000 products are missing a colour description teams may spend hours filling in values. The metric reaches 100 percent and everyone is satisfied. But the products may be discontinued or colour may not matter for the customer. Time was spent but no value was created.
That is not quality. That is wasted effort.
How to think differently
Sebastian:
Stop treating data quality as an IT project with a goal of zero errors. The goal is not perfection. The goal is business value.
Think about packing for a trip. You do not prepare everything you own. You focus on what you need based on your plan.
If you go hiking you check your shoes. If you go to dinner you prepare something suitable to wear. Everything else stays behind. Data should be treated the same way.
Data quality linked to initiatives
Sebastian:
Your "travel plan" is your business initiatives.
- If you work with personalisation, customer data is critical.
- If you optimise logistics, weight and volume are critical.
No data point has value on its own. It becomes valuable in context.
How to start
Sebastian:
- Review what you measure
- Identify which data quality metrics you track
- For each issue ask:
- Who uses this data?
- What happens if we do not fix it?
If the answer is “we do not know but it looks better in the system” stop measuring it.
If the answer is “invoicing stops” it is critical.
Conclusion
Sebastian:
Data quality is not about cleaning everything. It is about improving what matters. Good data quality means the data serves its purpose.
- Stop measuring everything
- Stop cleaning the attic
- Focus on what creates value
Closing
Sebastian:
Thank you for listening to Syntra Utbenat. If you want to discuss the topic or talk about data, reach out to us. You can find us on LinkedIn or via fiwe.se. We will talk soon.
