Wayfair is well-versed in the business of BI, according to Nathan Kollett, senior manager of the ecommerce company's web analytics team. This is evident in the analytics team's regular BI-based reprioritization process and its close proximity to the business unit. At the recent Real Business Intelligence conference in Cambridge, Mass., SearchCIO staff sat down with Kollett to discuss Wayfair's self-service BI strategy and its place in organizational structure, as well as the hype vs reality around self-service BI.
You're part of the analytics group at Wayfair. What is your team's relationship with IT?
Nathan Kollett: Within the analytics team, we're actually tied a lot more closely to the business. We think of the data as being owned by the business and we work with IT to make sure that the technologies we're using are available, well-integrated and performing for us.
You referred to IT as a bottleneck to self-service BI. What creates that bottleneck and what role does the IT department play in self-service BI strategy?
Kollett: I think in traditional BI models, the IT group can be the bottleneck because you think of the IT group as being the ones that, as I mentioned, own the technology as well as the reporting and the insights that are coming out of that. But in a self-service model, that report generation is distributed to the business. That allows IT to focus on what they do a great job at, which is maintaining the technology and making sure that it's integrated and available.
What does your team's reprioritization process look like and how often does it happen?
Kollett: Within the BI team we are agile. That means both Scrum and Kanban, depending on the team that you're talking about within BI. As often as once a week we'll meet with stakeholders and look at the Kanban board and just make sure that we have the right priorities in place for what we're working on and with the business questions that we're answering. With the Scrum teams, it works pretty similarly, just within two-week sprints. Every two weeks we are reprioritizing.
What's real and what's hype when it comes to self-service BI?
Kollett: I think the dream of self-service BI can be real. What's overlooked is the discipline and the people that are involved in actually making that happen. A lot of my talk was focused on making sure that you have the right people aligned in the right places. Just thinking about IT's role in a self-service BI strategy is a great example of that; making sure that you have the right people working on the right things. And then, once you've put everything in place, making sure that you have the right processes to be disciplined about maintaining that.
How is AI affecting BI?
Kollett: When you mention AI, I think many people imagine one thing: a generalized machine intelligence that replaces all human decision making. But actually, there are much narrower applications of AI that might have a bigger impact on BI. One example that I've come across in the world of BI is the idea of multi-armed bandits. The idea that you can have an algorithm that's testing different variations of an element and then automatically converges over time. You learn something about that with statistical significance as well.