As a product, think of Birst as a fabric that sits on top of all these silos, tapping into them and linking them all together to provide predictive insights.
Sitting down with Birst’s VP of product strategy, Southard Jones, in San Francisco this week, he said:
You want to solve a use case problem, like address the way that purchasing happens across a 78 country company that’s highly decentralised, across 29 ERPs. That’s not a desktop discovery tool. That’s an analytics solution.
It’s great to have a discovery tool that can give an end user a faster, beautiful way to answer questions on a spreadsheet. But to address and create value for the organisation and to deliver millions of dollar of value, we need a platform that’s going to address a larger problem. And IT and business are both going to team together to make that purchasing decision.
Jones believes that although enterprises are interested in the latest data technologies, such as Hadoop and AI, ultimately they’re more interested in fixing the fundamentals of their data structures. He argues - and I tend to agree with him - that companies seek relief from just getting some governance in place, and getting a better understanding of their entire data portfolio.
AI and Hadoop are nice to have, but ultimately they can’t do a great deal if you don’t even know what data sits where. Jones said:
Birst was built on delivering automated machine learning capabilities. So, under the covers, we had this already. What we are hearing is that companies want a pragmatic approach to enable those end users, who are operational in nature, to leverage the intelligence of machines and AI.
I wouldn’t say it’s the number one thing they’re asking for though. We talk about it, but the chief purchasing officer, for example, says ‘that’s great, that’s nice, I want to do that some day’.
'But let me at least understand how many purchases, by supplier, in all countries, across all ERPs, are happening on a daily basis. I will be ecstatic if I can get that’.
That’s a hard problem for me to solve. After you solve that problem, yeah let’s talk about this AI stuff. Let’s not get the cart before the horse.
Making Hadoop operationalA problem we regularly see with customers’ Hadoop systems, is that whilst they are incredibly scalable and quick, they’re also incredibly skills intensive. Enterprises hire data scientists to sit around the data lake, crunching data in the hope that they find those nuggets of gold to get a competitive advantage.
However, for the average employee, the idea of using a Hadoop cluster and introducing the insights into their day to day job is typically a bit of a pipe dream. Things revert back to Excel spreadsheets. Jones said:
When Hadoop first came out, there were a lot of science projects and people thought it was great because it was a cheap place to store a lot of data and they didn’t have to worry, so they just chucked it all in there. Now what? How do I get benefit around that? We are back to like the data feed world.
Hadoop is a great solution for a true data lake. But it’s a data lake, it’s not a data warehouse. You still have your data warehouse, it’s not going away. The more you look at it, the more you realise why it’s not going away and will sit side by side with your data lake. And there’s a bunch of data that may never make it into either one of those. So what do we need? We need this fabric on top that allows my users to have access to this.
How do you expose all that to end users that allow them to make daily decisions that make a big impact?
Embedding analytics throughout the organisation
Jones explained that Birst typically gains traction in the enterprise once it proves the use case at a departmental level. Once the returns begin to be seen within that line of business, other departments want in on the action.
However, I was keen to find out which companies Birst faces competition from in its sales discussions. Jones argues, which is understandable, that the biggest competition is convincing business and IT that they need yet another BI tool. He said:
Our largest challenge competitively is that who we sell to, they’ve all got at least two or three BI systems in there. And the first reaction from an IT person generally is: why can’t I do it with what I’ve got? Then the business person is saying, here is the challenge I’ve got, I haven’t been able to address it for years, here’s the value I can get from it. Our biggest challenge is the incumbents that are already there.
However, Jones adds, whilst people in the past have dismissed broken analytics platforms in favour of gut instinct, they now find that they need trusted, well governed data. Proof points are becoming increasingly necessary. He said:
The CIO of one of our large customers said something like everybody wants to do analytics, but getting everyone to do it in a way that makes sense is one of the hardest change management problems there is.
Everyone has their own perception of what analytics means. Everybody has their own way of doing it. But 10 years ago it was kind of a badge of honour to be like ‘I don’t need data, I’m a smarter person, I’ve been doing business for 30 years and this is the way it works’.
I think for the most part, most companies, that doesn’t fly anymore. People now want to see analytics to make smarter business decisions.
Cloud analytics was always going to be somewhat of a slow burner. However, I do think Birst is on to something with this idea of networked BI. Companies generally have such a poor understanding of their data, and yet are desperate to get the insights they need. I do think the there is resistance from buyers to add yet another platform into the mix when they’ve already made investments, but if Birst can prove the returns - which it claims it can - then this may be less of an issue.