By Danette McGilvray, Granite Falls Consulting, Inc.
“Why are you talking with me about data quality? The big deal right now is artificial intelligence!”
Let’s be honest, AI is sexy. AI is the hottest buzzword. AI is like the fun hot tub or the sleek swimming pool. The less sexy (but essential) part of the hot tub and the swimming pool is that they require clean water. Which hot tub would you use?
The (unspoken) assumption is that clean water will be available when and where you want it. Think of what goes into ensuring clean water. You must have a source of water. Infrastructure such as pipes, treatment plants, and water heaters move the water from the source, test and treat it to meet cleanliness standards, and store it until ready to be used. No one would ever mistake the water for the pipes.
Yet when it comes to data, how often is there an assumption that just having the right technology (pipes, treatment plants, water heaters) will ensure high-quality data? Of course, it is important to have the right tools and technology. But they are not the end of the story. We must have people with knowledge to create, sustain, and manage the data. Those who use the data must understand how to apply it correctly. We must have good processes in place and training for those who touch the data in any way throughout its life.
AI is one use of an organisation’s data. Yes, an important use. If you haven’t heard the real warnings and real experiences about AI failures due to data quality, you have not been paying attention. Often the output from one AI use is input to another. If the data is bad, then it continues with an exponentially negative impact to our organizations and those we serve.
AI is more fun to discuss than data quality. After all who wants to talk about the plumbing when you can discuss the hot tub? When leadership demands “I must have AI – that is what I care about” they must be reminded that, like the hot tub and the swimming pool, there are basics that must be in place for AI to work properly.
AI (and every other use) needs high-quality data. Take the following sniff test, an informal reality check, to see what you think about the quality of your organization’s data.
- Can you find the data – get to it and access it?
- Is the data available when you need it – is it timely and not late?
- Does the data include everything you need and nothing is missing?
- Is the data secure – is it safe from unauthorized access and manipulation?
- When you get the data can you understand it – can you interpret it?
- Is the data correct – is it an accurate reflection of what is happening or what did happen in the real world?
- Is the data free from bias – does it accurately represent the population of interest?
- Can you trust the data and use it with confidence?
If you answered no to any of the above questions, you have a data quality problem. If successful AI is your goal, you MUST pay attention to the quality of your data. That includes having knowledgeable, trained, and motivated people in place. Too many do not know how to get started to solve their problems, but help is available from a proven, practical approach called Ten Steps to Quality Data and Trusted Information™. Learn how to create, manage, and sustain high-quality data, including root cause analysis and preventing data quality problems. Show the business impact of your data. Get insights into communications and the human factor of your data quality work. Bring your real issues and leave with practical solutions and an approach to data quality you can use for years to come.
Danette McGilvray will be teaching Ten Steps to Quality Data 10 – 12 Feb 2026 online. To find out more and to book the course visit https://irmuk.co.uk/ten-steps-to-quality-data/
