Why Wave went to extremes for agile BI

Profile picture for user jreed By Jon Reed February 27, 2014

kensisense2
Two years ago, Wave had a serious BI issue on their hands. Their existing BI suite was simply not agile enough. It is said that crisis and opportunity are flip sides on the same coin - which is exactly how Hui (Ken) Zheng, BI Manager with Wave, approached Wave's BI dilemma. With more than 2,000,000 users of its online small business software, Wave is in aggressive growth mode; clunky BI is unacceptable. I recently talked with Ken Zheng about how his team looked these challenges in the eye - and moved ahead.

Jon: Ken, have you always been a BI guy?

Ken: My background is rooted in Computer Science and Technology Management. I've been with Wave for two years. Before joining Wave, I was a technology consultant with Deloitte Consulting on their information management service line. Information management and business intelligence is my trade.

Jon: What motivated you to move from Deloitte to Wave?

Ken: At Deloitte, the BI solutions we provided to clients were comprehensive, but they were not agile. Getting these complex BI solutions up and running requires a great deal of time and effort. I joined Wave because I wanted the opportunity to approach BI in a more practical and hands-on way. I wanted to find the right BI formula for companies that move fast and make decisions fast.

Jon: This is not unlike Wave's difference with QuickBooks, right? At Wave, you want to give smaller companies an option that is not as cumbersome.

Ken: Exactly. Wave is all about understanding the needs of small business owners, entrepreneurs, consultants, freelancers and so on: small businesses instead of 'small-and-medium' businesses. By keeping this focus, we make Wave easier, because the tools make more sense, and there’s no avalanche of complex features that the small business owner doesn’t need. You can use Wave without being an accountant, and without being tech-savvy.

Jon: Take us back to your BI crisis. You were struggling with a large BI suite?

Ken: Yes, we were trying to work with BI software from a major BI vendor. But it was just too difficult to implement, and way too hard to use. We actually implemented the solution, and tried to use it for four months. It was just too complicated and too slow.

Jon: I'm sure that was a very stressful time for you and your team. What did you do about it?

Ken: We opened up a new BI vendor evaluation, and looked at almost all the major vendors, such as SAP BusinessObjects, Tableau, Birst, MicroStrategy, IBM Cognos, Oracle Business Intelligence, TIBCO Spotfire, Oracle Business Intelligence, QlikView, Pentaho, Microsoft BI suite, GoodData, DataHero, Domo and SiSense Prism.

Jon: Basically you looked at them all. :)

Ken: Yes, in the end, we had a final four of SAP BusinessObjects, Tableau, Birst, and SiSense. And we chose SiSense.

Jon: And how did you make the final decision?

kensisense3
Ken: I built a pilot project for each vendor solution. I simulated the most complicated scenario at Wave to see how each application could solve that problem. I worked that scenario to see which one would be the best fit for us. We're a growing, 60 person company, but as an online business that empowers each team for decision making, our analytics needs are intensive. We have the data usage needs of a bigger company; we needed a solution to match our intensive requirements.

Jon: Tableau is the BI solution du jour, so readers will want to know why you didn't go in that direction.

Ken: A lot of people ask me, 'Why didn't you choose Tableau?' Tableau is great; it has beautiful visualizations, and it is great for data discovery. But to me, Tableau is more like Excel 2.0. I believe it's still far away from being a cross-organization BI solution, because it lacks the unified data modeling capabilities that SiSense has. Managing unified data model in a centralized manner is critical for any enterprise-level BI application.

Jon: What other factors affected your BI decision?

Ken: SiSense had some critical advantages. Their solution is very easy to set up and build reports on top of. Plus, SiSense can handle a lot of data without any tuning to the database, and that is very important. As an online web company, we have a lot of data, and we don't have time to fine tune our database for every single report. Startups like Wave have to move very fast. In a big company, the manager might be able to wait a week for a new report, but our management cannot wait for even two days for a report.

Jon: But it's not just about the speed, right?

Ken: That's true. At Wave, we have a slogan that we want to give every 'Waver' the insights and the data. That means we are a very transparent company. Our BI team should deal with all the teams inside of Wave, and help them think about how to use the numbers and analytics to improve their operations. So we need technical solutions that work for this kind of transparent culture and, hopefully, empowering work environment. Performance is also critical for us. Our BI solution must allow us to push out reports without any performance delays or bottlenecks.

Jon: Tell us about the implementation.

Ken: After we did the SiSense pilot evaluation and compared it with all candidates, the winner was clear - we decided to purchase it. From there, we did the enterprise-level implementation. 90 percent of the implementation was done in one month - we went live in April 2012. Before SiSense, our report development time was two to three days. After SiSense, the development time went down to two or three hours. Also, our report query time improved from one-to- ten minutes to two-to-thirty seconds. Currently, we are a two person team, and we're delivering ten times the BI output that we did on the old system.

Jon: But does the speed make a difference to the business?

Ken: Yes. A manager will not wait in front of a computer for more than one minute to get their numbers they are need, so a slow BI report is very, very frustrating.  By the way, if we wanted to improve that response time in our old BI system, we would have had to hire another DBA and put a lot of time and resources into database tuning.

Jon: So is the management team making more informed business decisions as a result?

Ken: Absolutely - they get a lot of insights now. We do revenue analysis, user activity analysis, and ad sales analysis. We analyze the acquisition channel; we also analyze monthly user activity and churn analysis, and we do A/B testing as well.

Jon: Sounds like you tackled some critical BI issues there.

Ken: We needed a more agile solution. For our market and the competition we are in, it's agile or die.

Image credit: images provided by Ken Zheng.

Disclosure: SiSense helped to arrange this interview; SiSense is not a diginomica partner.