The CEO and co-founder of Birst, Brad Peters, tells me what he learned at Siebel, why Birst is different and why being overly focused on data quality is 'bullshit'.
Brad Peters, Birst co-founder and CEO.
Is there any relation between Birst and the similarly-named Eclipse open-source reporting project, BIRT?
No, they were both named at about the same time, it's just a coincidence. We came up with ours as it is a simple name with BI [business intelligence] in it.
As a private firm you won't tell me revenue; how many staff do you have and are you profitable?
We have around 150 staff, and although we're not profitable as we're investing for growth, we could be profitable on a dime if we wanted to be.
Tell me about your background at Siebel and what you learned there.
I actually had two tours at Siebel. On my first tour I was responsible for the first on-demand version of Siebel, originally called Sales.com, and which later became Oracle On Demand. I had exposure to what it's like trying to start a new company in a large company. But I also learned that taking something that is on-premise and trying to port it to the cloud is not ideal, to say the least. It was interesting to watch salesforce.com come after us, too. But we had a lot of large customers with a lot of data and of course they wanted to analyse that.
We had gone and done a deal with Business Objects and that wasn't really suited as a broad-based analytics tool. A lot of our customers at that time were what we called 'red' customers, which meant they had the worst customer experience. It seemed there were application people and business intelligence people, and they didn't understand each other at all. We built Siebel Analytics and it soon became the biggest line in Siebel's business, giving customer and operational insight.
But it was still pretty painful for users to deploy, and customers made underwhelming progress. We were shipping too many pieces, not just from us but partners like Informatica. We sold them parts, not the complete car. We like to say the only thing integrated at Siebel is the price list, and the same is true of Oracle and Microsoft. Whereas if you look at salesforce.com it was a fully deployed application, and I believed that we should be able to do the same in analytics. It's much, much harder than you'd think but we started in 2006.
I know you can run analytics in the cloud. Are companies still nervous about having sensitive data in the cloud?
I think we're starting to see an inflection point. We're 1oo% cloud-architected but we are still able to 'appliantise' our system. You can run it on a pool of servers, deployed virtually on-premise. But I think it's shifting. The whole notion of security and safety is changing. People realise that it's safer to keep your money in a bank than hide it under your mattress. The biggest risk is your own employees. Some of our largest clients are financial services companies and that data is absolutely sensitive. People talk about regulations and security but really what you need is control. There's a perception that if you put data in someone else's cloud that you lose control. I agree with that in some cases, which is why we also offer an on-premise version.
It's funny though, you find that once companies have actually bought it, saying they'll run it on-premise, they realise it's much faster to deploy it in the cloud. They say they'll move it back on-premise later but it never happens. But generally with cloud I'd say it won't be all cloud or all on-premise, I think for most companies there will remain a hybrid.
You talk about agile analytics. What do you mean by that?
Well, all the other tools on the market assume that you have a database, populated with data. That's the classic way business intelligence and analytics have been delivered. If you look at something like Microstrategy, it's a tool; it has no clue how the data got into the database. But that's the hardest, messiest bit. We have the pieces you need to do that too. We integrate data management with the front end to auto-create the system. That means when it comes to the data piece one person can do what it would have taken five people to do. That means you can get up and running with Birst in three weeks instead of six months or a year. And if you make a change you can do that much faster too.
But when we talk about agile we're also talking about the process methodology, not just the software. It means getting much better and faster results.
Do you agree that mobile business intelligence is going to see particularly rapid adoption?
Absolutely. You get much better access and value from data when you have it at the point of interaction. It's more in context than traditional desktop stuff. Mobile data is often very purpose or role-specific, and is in the hands of people who would have never used PC-based reporting. It's broadening the audience for analytics. You can now bring data into the discussion that you are having live, and analytics are much more powerful in context. You may be able to answer a question before a meeting but if that leads to two or more questions that you can't answer, you're wasting valuable time. Mobile BI can give you the answers to those questions.
What does Birst offer that the reporting offered by the app vendors themselves, salesforce.com for instance, don't offer?
A ton of things. They provide analytics the way that we did at Siebel. Operational analytics have serious limitations. Number one, you're reporting against transactional systems. That slows them right down, so you can only do basic queries. You almost exclusively look at specific sets, or one object at a time. But with our analytics there is the ability to look at multiple objects, for example to look right across from lead to close. You'd never throw that at salesforce.com because it would simply time out.
You also want to do things like look across silos of applications, not just salesforce.com. There's your packaged applications but also home-grown apps, and so on. You need to be able to look across silos of information and ask computationally powerful questions. Salesforce is like an odometer - it tells you how far you have gone. But you also need a speedo to tell you how fast you are going.
What are the implications of this Big Data trend on the future of analytics?
First off I'd say Big Data is a misnomer. It's been around for a lot longer than two years. Teradata is a billion dollar business analyzing very large data sets, Netezza and so on. But there is something to Hadoop: it's about the ability to look at lots of unstructured data at a very low price point. You can stick some data in Hadoop and someone can play with it. Our customers are doing this already; we partnered with Amazon on Redshift [their data warehousing as a service offering] so we're supporting [Big Data] 'out there'.
The analysts talk about the three 'v's of Big Data [volume, variety and velocity] while some add another for Value. I've also heard some data quality vendors talk about a fifth 'v', veracity. They argue you need to be sure of your data quality before you start analysing it...
If all the data had to be perfect then 90% of data analytics projects wouldn't happen. Data quality can't always be up to 100% regulatory-type standards. That's just an excuse not to do it. 80% of a good answer today is way better than 100% of an answer in 12 months' time. A lot of that talk about data quality is bullshit. Companies need to be able to make meaningful decisions, now.