Tableau, along with rival Qlik, is virtually synonymous with data visualisation software. Its user conferences vie with evangelical church gatherings in passion. “We. All. Love. Data,” boomed the voice at its last user conference in New Orleans.
Away from such sonic drama, and as the year turned, Adam Selipsky, chief executive officer at Tableau, spoke to Computer Weekly about the company’s top-line strategy.
Selipsky joined Tableau in 2016 after spending 11 years as vice-president of marketing, sales and support at Amazon Web Services (AWS). He is a Harvard MBA and was a strategy consultant at Mercer.
What follows is an edited version of a slice of his interview with Computer Weekly.
Whenever I get a chance to talk to a CEO, which is fairly often but not that often, I like to ask what their top few strategic priorities are. I usually caveat that by saying that if they have more than two or three, that’s just too many.
Selipsky: I’m not sure I agree with that. There are often a number of things going on and doing them all well can be the difference between being good and being great. You don’t want to be distracted, but I think single-thread things lots of people can figure out; multi-threaded things not everyone can figure out.
But I’ll be happy to talk about two or three strategic issues.
What we’re seeing is that more and more customers want to deploy analytics, and deploy Tableau very broadly – whether they’re public sector or private sector, large organisations or smaller ones. And so, in order to do that, they need to be able to have more and more people working with data, doing analytics. We have to continue to evolve our analytics platform in many ways.
“We’re in the middle of a multi-year journey to get better and deeper enterprise capabilities”
Adam Selipsky, Tableau
So customers need ever broader and ever deeper analytics platforms. And so that is the first theme, a product theme of continuing to drive better ease of use, more intuitive functionality. Essentially, we need to have it be easy enough to do analytics that anybody is going to be able to do it in the future.
So, I think if you look ahead a few years, you’ll have many [large] organisations with tens of thousands of people doing analytics. Not all analysts are data scientists. You’re talking about accountants, graphic designers, lawyers, product managers, and what we really need to do is make analytics and make software accessible to those people.
That’s a big theme and then there’s a lot of technologies underneath, like natural language processing. But our product development is also about being more useful to sophisticated users – analysts at the top of the pyramid. And so we have broadened our platform to include data preparation. Otherwise, you have can have garbage in, garbage out.
Tableau has been very popular with end-users from its inception. But there is scepticism about how suitable your software is at the enterprise level. How do you feel about that?
Selipsky: Tableau has always been very popular, very beloved by individual analysts. But we are seeing more adoption by big organisations. And as we get deployed more and more deeply in big enterprises, we become a mission-critical application inside the enterprise.
For us, there is a company-wide effort to truly become more and more enterprise-ready. There are product elements to that where we have to not only build analytics functionality, but also security and governance and compliance features into our platform, or else IT and chief information security officers aren’t going to allow this to be deployed. We need to essentially help our customers market internally.
And so you have to do a lot better with things like total cost of ownership studies, return on investment studies, and so on – studies giving great examples of enterprise use cases and public enterprise references – customer evidence.
We’re in the middle of a multi-year journey to get better and deeper enterprise capabilities.
OK, but in terms of the narrative of Tableau, for quite a long time, the story was “land and expand” – a bottom-up adoption from the grassroots approach. But there is another way –- approaching the problem from the top down, starting with the needs of the enterprise and going down to the users. And starting from a perspective of having a sophisticated understanding of a customer’s top-level business strategy. Information builders, for instance, champion that approach and they stress data governance as an enterprise capability.
Selipsky: I think that’s often what people say when they are not particularly beloved by their users. I think that the answer is that to truly be effective, you need to do both. And it’s not a choice that we can make. If we want to be truly useful and really get ubiquity, we need to do both.
Tableau did become very well known for this “land and expand” approach – focusing on the end-user, getting into an organisation, having usage of Tableau grow virally because we’re so useful to end-users. At the same time, particularly within enterprises, I think you also need “discover and descend”, which is much more top-down.
I spend a lot of time with CIOs, CTOs and chief digital officers, VPs of architecture, heads of data strategy and so on, and that is very important. But typically, those people aren’t very interested in suppliers if nobody in the company is interested in them. I think it is incredibly powerful when you’ve got users pulling you into the company. And that’s what gets the attention of the senior leaders, saying “OK, we know we’ve got these pockets of usage happening and we know people want to use Tableau. We know you're useful to our business”.
And if, at the end of the day, if you aren’t useful to their business, then what else matters? “We know you’re useful to our business,” they say. “Let’s figure out how to get you in here. Let’s make sure that you’re secure, let’s make sure we can govern this usage in the right way. Let’s make sure we get the right economic terms”.
I think you really need that “land and expand” and that top-down “discover and descend” combination in order to truly get penetrated [into companies] these days.
Do you see at least the earlier generation of BI software – MicroStrategy, Cognos, BusinessObjects – as a threat?
Selipsky: I’d say the landscape around the old-guard BI companies has changed significantly even in the two-plus years that I’ve been at Tableau, starting in September 2016. I remember those first conversations with customers in my first couple of months at the company. And it was much more forward-looking, saying “hey, we’re thinking about moving away from some of those platforms”.
But the reality at that moment was that Tableau was still predominantly being used for new use cases. And now, for these new use cases, we’re going to deploy Tableau.
Fast forward two years, and you can really see migration happening. Now, there are still a lot of people deployed on those older technologies, but the movement has started. We are working with a lot of customers who either have or are in the middle of tearing out some of those traditional BI deployments.
I think there are going to be many more of them to come in the years ahead because those technologies were being deployed for decades, and they are complex deployments that are going to be hard to get off of because of the complexity. But what happens is that as companies are more and more deeply deployed with Tableau, they look at these multiple BI deployments and ask themselves: “Why do we continue to support multiple platforms?”
We are working with one big European bank that is going down from six or seven BI tools to essentially one and a bit, with Tableau being the one, and that’s quite a common scenario.
Typically, a few years back, I’d ask that sort of question either of a user of Qlik or yourselves and people would say “really big organisations have everything”. So it’s never a binary choice. They have a variety of BI and analytics tools, so “let a hundred flowers bloom”, to quote Mao. And from the supplier side, they have usually said “there is enough pie for all of us”. And only up to 15% of people have typically used BI software anyway, so there is another big percentage of any organisation that is there for the taking. But it sounds as though you’re saying that there is a trend towards decommissioning some of the earlier investments?
Selipsky: Look, I was just with a global 1000 company recently here in London which is already a customer, but is looking to dramatically expand its Tableau footprint and, at the same time, decommission one of those old-guard BI suppliers. It’s absolutely a trend that we see, this trend towards consolidation.
By the way, I’m not saying it’s binary. I’m not saying everybody’s going to go from many to only one. Some are going to go down from 20 down to two or three. But I think there is a trend towards consolidating the analytics platforms that companies are using.
It’s driven by a couple things, one of which is just intense cost pressure. The entire financial services industry is trying to take out cost. And there is the organisational cost of having to train people on multiple tools and platforms.
Read more about modern business intelligence developments
- MicroStrategy’s latest BI platform supports a ‘zero-click’ analytics future with help from new voice and vision capabilities. The aim: quicker decisions and the democratisation of data.
- Tableau unveils AI and NLP tools intended to improve users’ experience with its BI platform by adding smart capabilities to Tableau 2019, now out in a public beta.
- Self-service BI software comparison: Tableau vs Power BI, Qlik Sense.
There is also the cost of different people thinking about data differently because they are working with different platforms and not really speaking the same language as each other. I hear about this all the time from customers. They’ll have two people from their own organisation pitch up in a meeting with two different answers to the same questions, supposedly from the same data source.
But those data sources are being run through different platforms and end up getting corrupted or changed or amended. And so you no longer have a single source of truth in your own underlying data. And that’s a big problem that many enterprises are facing. That is why, I think, for all those reasons, you absolutely see a trend towards consolidation.
Where do you think we are, as organisations grappling with data? Do you think much will change radically, with respect to, for example, more machine learning being added to do more augmented analytics, and more natural language processing? Where are we?
Selipsky: I think it’s still very early in the development of analytics and of data capabilities in general. It feels like we are far along, but I think we’ll look back in five or 10 years and say: “Wow, you know, we had a long way to go back to 2018, 2019.” And all of us – meaning our customers, ourselves.
I think if you extrapolate out some number of years – I can’t tell you exactly how many years – in terms of how many people do analytics, how many people are making decisions based on data, and have data analytics as part of their daily and their weekly routine, it’ll be the same as today, people have word processing or spreadsheets or messaging applications.
I think analytics will be another one of those knowledge worker applications that is used ubiquitously, with tens of millions, probably hundreds of millions of knowledge workers around the world – in time, not immediately. And then you have to ask yourself: “Is just going to happen automatically?”
You know are there other things you need to enable that to happen. There is clearly the demand, there is clearly that trend with all the data being created in the world, to have very robust analytics capabilities atop that data. But how? It will be things like natural language.
So, just to take a metaphor, a few decades ago, lots of people were putting on their CVs that they were certified in Microsoft Word. It was a new technology. And I remember when I was in college, I had classmates who were putting certified Microsoft Word on their CVs. That just seems like such an odd concept today.
I think that, in the fullness of time, that is the kind of ubiquity and the kind of breadth of usage that analytics will see.