Data has so much promise, but organizations struggle to use it effectively, especially with the increasing complexity of managing large projects and the pressure of incorporating AI. To help us understand where organizations go right—and wrong—with their data strategy, Simon Asplen-Taylor, the founder and CEO of DataTick and author of “Data and Analytics Strategy for Business”, joined The AI Forecast. Simon has spent years counseling organizations on getting their data right. Simon shared the common missteps organizations experience with their data – with all roads leading back to data quality and a wild west of data collection across a business.
Here are some highlights from Paul and Simon’s conversation.
You can’t value data if you don’t understand it
Paul: Is data really still a hard sell for some CEOs?
Simon: Yeah, it is. A CEO six months ago said to me, “People talk about data, but I don't get it.” This ultimately comes down to everyone having a different view of data. If you're a regulator for the Bank of England, you think of data as being one thing. If you're in marketing, you think of data as another thing. Technologists are historically subservient to those in the business who will come to them with certain requirements to build a data set around.
The problem here is oftentimes people don’t know what to ask for. Technologists end up explaining the weaknesses they’ve seen, and you don’t really maximize the full power of data. We need to get people to the place where they aren’t giving their requirements from the jump. Instead, allowing the space for technologists to proactively explain the value of data and what it can do. When that happens, it’s much easier for the CEO or another leader to “see” the value of data.
Data leakage is actually a data literacy problem
Paul: I want to touch data quality. I remember one of my first projects that I worked on in the supermarket business and they were trying to establish master data management. They just wanted a single view of all their products, what they bought and sold to their customers. It seemed to me they were constantly chasing their tail with that. How much of the problem do you think in enterprise is data quality? And how much of that is born out of the division of labor and business silos where someone owns this chunk of the data and someone else owns another chunk?
Simon: Yeah, big question. Data quality is not a homogenous problem. It starts with what I call data leakage. For example, you start off at the beginning of a process and you may have a salesperson talking to a customer and then only talk about products relevant to them. But your organization may sell three or four products, and that salesperson won’t ask about other interests that may be relevant to those additional products. That’s data leakage because there's an opportunity to do some upsell and some cross-sell. You could sell more things to them, but you haven’t caught that information. That's data leakage.
Someone else could have used that data downstream. Suddenly you have a problem later on where someone has to go back to the customer. That means more interventions picking up on data that's being missed out – the quality isn't good enough. Data quality problems and that leakage is caused by real world problems and by peoples’ incentivization to do certain things. You could argue that it comes down to a data literacy problem – making sure people understand that doing this level of data collection is part of your job. You are going to help the organization grow more because if you do this, we can sell more downstream. Ultimately, it’s about business incentivization and business process and people being able to say, “I am going to do the right thing for the organization, not just for myself.”
+++
Don’t forget to tune in to Spotify or Apple Podcasts to listen to future episodes of The AI Forecast: Data and AI in the Cloud Era.
