Stop saying "Garbage In, Garbage Out", no one cares

We recently ran a small poll on LinkedIn asking people what are the main blockers for GenAI for analytics - looks like there are many - but still, no one in the business cares

Last week I ran a poll on LinkedIn - “Before rolling out GenAI for analytics, what’s the biggest blocker you’d fix first?”. I gave people four options, and they were largely split between them:

• Data quality

• Metric definitions

• Documentation/context

• Governance & security

And some even said “all four”.

Which makes sense - all of those are things we should get better at before we unleash AI on our data.

But something I realized recently - is that no one (outside the data team) cares. They want AI connected to their data, now. ChatGPT, Claude and the others have set a high bar for human-data interaction and your colleagues want it in their workplace today.

But Yoni, these are real problems!

True, they are. However, these days, we’re all expected to run faster. Faster than ever before. AI has accelerated everything around us - job transitions, workflows, people’s expectations. In many places, it hasn’t made any tangible impact yet, but people are expecting it to.

And now, the interest rate in the US is going down. This will accelerate things even further. Historically, interest rates were raised to cool the economy, and lowered to heat it up. If you thought it was hot thus far, you’re in for a treat!

Your first step is to embrace this mantra:

We don’t have enough time to get our data to a perfect state. We don’t have time to solve all the quality and governance issues. We need to start delivering.

The way to do this, without having it all blow up in our faces, is to be selective.

You already have some data that is fairly clean, in a data mart or gold layer. Start there. Focus on making that data accessible to business users through AI. Show them value and potential, and then use it to get more resources so you can clean up the rest of your data step by step.

Don’t try to boil the ocean. Take a step at a time and get your first use case out and into the hands of your business stakeholders before the end of the year.