What’s your favorite wrong question?
Most data projects are still approved on trust instead of evidence. Someone says the system was tested, the invoice gets signed, and nobody asks to see what actually ran, against which data, and with what outcome. This article looks at why that question is so often missing, why that silence creates risk at board level, and why documented test results should be treated as a standard project deliverable, not a technical extra.
Old habits die hard, but your Data Testing habits shouldn’t
Many data teams still rely on testing habits they inherited years ago, not because they are effective today, but because building something better used to be too difficult. That excuse no longer holds. As data environments grow more complex, outdated checks quietly pass risk, cost, and accountability back to the business.
Why nobody tests their data
Ask almost any data leader whether data quality matters and the answer usually comes quickly. Of course it does. Bad data leads to bad decisions, broken reports, failed AI initiatives, and loss of trust across the business, and the list could easily go on… And yet, in practice, very few organizations actively test their data in a systematic or ongoing way.
This is rarely because teams are careless or unaware of the risks. Much more often, it’s because data fails in ways that are easy to overlook, hard to assign to a single owner, and difficult to define with absolute certainty. That’s why it’s worth pausing for a moment and asking why data testing so often never really gets started.
Použití CAT pro testování datových agentů
V posledních měsících je jedním z častých scénářů využití AI tvorba agenta, který odpovídá na otázky uživatele na základě vašich dat. Může jít o SQL datový zdroj nebo například Power BI sémantický model. V tomto článku se nebudu zabývat tím, jak takového agenta postavit. Chci se zaměřit na jeden důležitý aspekt datového agenta – spolehlivost.
Why AI’s future depends on data you can trust, and how CAT makes it real
In 2026, the AI discussion has matured. The question is no longer „should we use AI?, most organizations already are“. The real question has quietly shifted to something much harder: why does AI work well in some companies, and fail completely in others?
Using CAT for Testing of Data Agents
We built a data agent for one customer on top of Power BI model. However, one of the unpleasant aspects of using LLM for creating queries (in our case it was a DAX) is it can produce different DAX for same user’s question. If you repeat same question three times, it’s quite probable you will get 3 more or less different DAX queries. There are methods to increase consistency in Agent implementation however you cannot eliminate creativity of LLM model.
Regression Testing: DAX User Defined Functions
User Defined Functions is a new feature in PowerBI currently in public preview. There is a lot of buzz in the air regarding this because it opens new scenarios with modularity and reusability of the code.
The Hero’s Path in Data
After years of projects and migrations, I’ve learned that the loudest tools rarely solve the hardest problems. We keep hearing about AI, automation, and “next-gen” platforms, but when someone asks, “Is the data behind this actually correct?” — the room often goes silent.
That silence exposes the real issue: not lack of technology, but lack of verification.
Because trust isn’t a by-product. It’s a decision — earned by testing what we believe to be true.
Data Testing in the Age of AI: What’s New and What Stays the Same
AI is changing how we work with data every day. It can write code, generate reports, and analyze patterns faster than ever. That’s exciting – but it also means small mistakes can spread more quickly. Some things, however, never change: if your data is broken, your results will be broken too. Garbage in, garbage out still applies.
You might be managing BI dashboards, feeding data into AI models, or handling big data pipelines. And you’ve probably felt that sinking feeling when a small data issue throws off a report or model. AI won’t magically fix bad data – it just makes the problem more obvious. So how do we handle this new reality? Let’s look at what’s changed, what hasn’t, and how modern tools like CAT help.
When engineers don’t test: Key lessons from our Data Quality meetup
At our recent meetup “How to measure Data Quality in engineering” we went deeper: technically, practically, and without fluff. This time, we focused on the role of data engineers, developers, and teams at the beginning of the data lifecycle. And there was plenty to discuss.
The discussion revealed a few recurring mistakes that unnecessarily increase risk, lower trust, and introduce chaos into companies. Here’s a summary of the most important takeaways.
