Customer segmentation with HANA – Prototyping with an unexpected catch
Can HANA be too fast? Well, I thought that break-through performance would be welcome for all, but during a recent customer project I learned that, from some perspectives, HANA can just be too fast for some.
We were working with the IT team from a large retail bank who had been tasked with driving innovation into what had traditionally been a rather conservative organization. The organization’s core banking platform was declared out of scope, so we decided to focus our prototype on the one remaining high volume/high value area: customer management. Here increasing analytical speed would mean better insight and ultimately more effectively focused marketing campaigns.
To make things exciting we decided to race our HANA prototype against the bank’s current system, taking daily loads of production data to make sure that we had a level playing field for our analytics Grand Prix.
Crowds gathered as the race began and we saw the prototype return complex queries results between 250 and 2,000 times faster that the existing solution. Analysis that had taken two hours to crank out now took five seconds. Everyone cheered….well almost everyone.
Business users and managers cheered. They saw the potential to fine tune campaigns as they ran, driving up success rates and ROMI. HANA could enable them to do things that had been impossible in the past.
The IT guys cheered, happy to see that the business recognized that technical innovation could directly drive business innovation.
The Business Analysts were, well… quiet. Perhaps they were celebrating internally? We realized that these guys had just seen a week’s work for their team boiled down to a few minutes. They’d also seen the ease of use and fast return time for queries would allow business users in other departments to select customer data by themselves, without today’s necessity to ask the Business Analysts to do it. We presented a business solution, but created a potential political problem. HANA was just too fast for guys who saw their role as Masters of the Black Art of Analytics.
We’d aced the Process and Technology aspects of the prototype, but we’d under scoped the People aspect. The performance of HANA delivered a surprise to everyone, for the good and the bad.
Have you seen HANA being too fast?