This BeyeNETWORK spotlight features Ron Powell's interview with George Mathew, President and COO of Alteryx. Ron and George discuss how Alteryx provides clear-box analytics, either in the cloud or on premise.
George, it has been a while since we last sat down to chat. Why don't you start by telling us what Alteryx has been up to since your 8.0 announcement?
It's great to have this opportunity to chat with the BeyeNETWORK. Since the 8.0 release, we have, of course, been getting ourselves focused for a big 2013. Our customers have really been giving us great feedback in terms of what we've accomplished by delivering the Alteryx Analytics Gallery
, which is directly deployed onto a public infrastructure and is now available for anyone to access so they can execute analytic applications in the cloud.
What's really exciting for us is to be able to take that capability and to give customers the choice of not only being able to have that available as a set of applications that you freely deploy into the cloud infrastructure, but also the ability to take and customize those applications very seamlessly and put them into a private studio. Those private studios can be deployed directly on the same infrastructure that you have your public Analytics Gallery that Alteryx has delivered or deployed directly onto private servers within your four walls. A lot of our focus right now is to get the Analytics Gallery and private studio offering more available to a broader set of our customer base as well as being able to introduce it to new prospects so they can share and consume analytic applications wherever they’d like to do it.
And it's really a great way to collaborate. For companies to see what others have developed within the Gallery is probably one of the biggest benefits. Is that what you're hearing already from your customers?
George Mathew: Absolutely, and it’s customers as well as partners in the ecosystem. Because if you look at even the fidelity of the applications that are in the Gallery today, we've noticed that the partner applications are the ones that are actually getting a fair amount of traction above and beyond the Alteryx powered applications that are in place.
The reason for that is that the partners are actually putting lot of their time, and attention, and energy into creating high-quality applications that will end up becoming lead generation sources. That ranges from the analytic consultants, the ISVs, as well as the marketing service providers who've all created great analytic applications.
What I also would like to ensure is that we continue to be extremely ecosystem friendly and have the Analytics Gallery continue to evolve. And to your point Ron, the idea is always going to be how you can see the best practices, collaborate on the best practices around an analytic use case that you're trying to solve, and very seamlessly recombine, mix things up, be able to change it, make it your own, and either share it back to the broader community or establish higher analytic IP that you might not want to share. That can be delivered internally to a set of constituents who want to be able to grow that base from what they started with in the Gallery. The more interactive development and collaboration are key aspects of how we see the Gallery continuing to evolve.
Now the term "analytics" has become directly associated with the actionable insights that users can glean to better understand current information and anticipate future trends. How do you foresee companies effectively implementing these predictive analytics processes to bring immediate value to business decision makers?
George Mathew: What we've seen in the past is that predictive analytics has been very helpful in terms of helping the decision maker craft the next best decision in his or her organization, largely by using various techniques from a predictive and statistical standpoint to use data to understand where the not too distant future is going to go. That being said, I think the biggest challenge in the predictive world is that the analytics that have been predictive in nature have largely been delivered in a very black-box fashion. You don't necessarily understand the math or what data came together to create the analytic output that gave you that predictive function.
As decision makers rely more and more on predictive analytics to make good data-based decisions, what ends up happening is that you need to be much more clear-box in terms of being able to describe the lineage, the source of the data, and how you've actually calculated, say for instance, the predictive function that you've put in front of that decision maker. This is where I think analytics really needs to shine in the coming year, largely to move away from these black boxed predictive capabilities and really get the process into the hands of the decision makers, not only the analysis, but how it was done, the fidelity of it, and the lineage of how things were actually delivered so that there's a trust in how the prediction works. Rather than trusting what someone else has done, you want to be able have a much more logical and explanatory approach to your analytics. This is where I think we're going to see a little bit more change in terms of black-box versus clear-box analytics coming to the enterprise.
I really think when you look at the insights and being able to determine how they were actually developed, you have to be able to see the actual work or process flow.
George Mathew: And when you look at the companies that are actually succeeding right now in the analytics market, it is the ones who actually provide a level of visual overview or visual capability into how the analytics are being built. Alteryx definitely has had some success in terms of bringing ingestion of data from various sources, running predictions against it, and creating packaged analytic applications like we do. And then, when you look at it even companies like Tableau, they've actually done extremely well by creating the experience for the visual analysis to occur around the data that's being brought in. Those are the use cases where people can actually trust the data and the analysis that's occurring because they have a way to visually inspect how that's effectively being delivered. Again, it goes back to the notion that we need more clear-box approaches to analytics.
Obviously, one of the keys for customers is to be able to access multiple sources of data to gain this business insight across a hybrid environment. How do you see this demand changing the role of BI platforms in 2013 with all of the data sources out there? George Mathew:
That's a really good question. One of the things I've seen this past year is actually the shift of interest away from just the notion of a BI platform. If you look at where the Gartner Quadrant for 2013 is going to emerge, they are actually highlighting it as a BI and analytics platform. The reason for that is that most BI and analytic systems today have to rely on an underlying ETL
and data warehousing process to be able to bring multiple sources of information into making better decisions.
I think the big difference between BI and analytics is the assumption that you don't necessarily need to have a structured data warehouse underlying an analytic platform. In the case of Alteryx and other solutions in the space, what we look at is a very simple and easy way to ingest those multiple sources of data, whether that has a formatted structured data warehouse underlying it, simply going against an Excel spreadsheet, going against operational data that might come from a NoSQL or semi-structured data source like MongoDB or Cassandra, or also be able to reach into a Hadoop distribution environment where there might even be petabyte scale information that's stored in an unstructured way.
The beauty of an analytic platform is that it doesn't worry about the source and the shape of the data as it is coming in. It's able to ingest anything into its environment to be able to create a meaningful view of the analysis surrounding it, help predictive analytics be built in a clear-box way, and then be able to package and deploy an analytic application. And I think this is where we're seeing a shift. Even when you see the market growth rates of companies now, the BI platforms are actually slowing down in terms of the people and the companies that are deploying BI, and the analytic platforms are now actually increasing in adoption. This is where I see an interesting dividing line between BI and analytics emerging, particularly this year and the coming years.
Analytics spans the BI sources as well as the operational sources. And you mentioned a lot of big data sources like Hadoop, MongoDB and Cassandra. How important is the traditional data warehouse if we now have all of these big data and unstructured sources?
George Mathew: I think this is where the fundamentals of the traditional data-warehousing world have broken down. Ultimately, most of the rationale around the data warehouse was the fact that you can actually bring all of your internal sources of data together in making good decisions by being able to package that up and report and analyze it. Now, with petabyte scale, unstructured information being inside of a Hadoop source, cloud data coming from places like Salesforce.com, certainly the growth of Teradata both from a traditional data warehousing standpoint but also some of their capabilities as a big data appliance, these are becoming more and more important to making better analytic decisions inside of an organization.
And then interestingly, social media data from Facebook, Twitter, Tumblr, LinkedIn, Foursquare and the like is becoming a much more impactful source of external content that needs to be very seamlessly analyzed inside your environment. In addition, there are a lot of great syndicated sources of content whether that be demographic insights, firmographic insights, or population statistics from the Census Bureau. These are sources of information that are actually indicative of customer behavior that you need to be able to very seamlessly bring in, particularly if there's a focus inside of an organization on customer analytics.
We see both the inside sources as well as the external sources like social media and syndicated content as being a natural path forward for companies that can actually deliver analytics to ingest both those internal and external sources seamlessly and blend them together to create a composite analytic view they can make decisions around. That's where a lot of our focus and attention here is at Alteryx.
George, there are many challenges that are holding customers back with regard to analytics, such as a limited number of data scientists. But your approach to analytics is really helping to create applications that reduce the need for data scientists and actually allow business users to handle that. What's your view on this going forward, and how it will be addressed in 2013?
George Mathew: Well, let me start with the perspective on the supply issue around data scientists. I think fundamentally the supply issue around data scientists being unavailable or that there are not enough of them trained inside the marketplace as it is today is a fundamental one that's going to pervasively impact us for at least the next half decade to decade or so. Fundamentally, we need more data scientists in the market. When you want to figure out how to approach this next half decade to decade with the fact that the supply issue is occurring, I think that there are two things that need to happen. One is you just need better tooling so that data scientists that are in their roles can actually do their jobs in a more automated fashion without having to hand roll a lot of code. When you look at the activities around blending data together to able to run predictive analysis seamlessly, we actually focus a lot of our attention on the user experience, particularly for the designing of analytic applications. Alteryx Designer really pays a lot of its attention toward how to blend data sources together without having to do it in a heavy programmatic fashion and to creating predictive analytics without having to do a tremendous amount of programming in a proprietary language like SAS. We use open source R, but we put a bunch of user interfaces on top of R so that you can actually have a parameterized approach to a predictive function inside of R versus having to do straight hard-core R programming.
I think that the approach to managing this supply issue is to get better tooling in the hands of folks that have responsibility for the creation of these analytic applications. That includes data scientists, but it also includes the data analysts and the data artisans that have some of these skillsets that they can acquire as the data science world becomes more pervasive. The tools need be able to represent those skillsets better, and that's where we see a real opportunity in the market – to help more people acquire these skillsets by giving them better tools, giving them better approaches to the techniques that are involved in data science, and not having to assume that we're going to be stuck because there is such a short supply of actual data scientists.
This is where a lot of our attention and a number of other companies' attentions are as analytic platforms emerge in this market. We want to bridge this skills gap that's in place by giving better tools and better experiences for a broader set of users – not only to consume apps, like the ones that we put into the Gallery, but also to be able to very seamlessly design things without being a hard-core MapReduce person or an R programmer.
That's great. Well, George, industry watchers have forecasted 2013 will be the year that big data moves to mass production. We've seen a lot of sandbox implementations within major enterprises. In fact, over 70% of the enterprises we follow have been experimenting and looking at Hadoop and other big data source sources. In your opinion, how will organizations be taking advantage of analytics in more unique ways?
George Mathew: I believe a lot of those initial experiments are becoming active production systems, and those production systems can be internal to the four walls of an organization, and it's just as easy to spin up compute cycles and capability from a processing standpoint in the cloud. So using Amazon Web Services, for instance, to be able to spin up a Hadoop cluster and then be able to deploy an analytic framework that you can take advantage of is just as straightforward to do inside your four walls as it is outside your four walls. That's becoming much more of an opportunity for people to productionize things much more quickly.
Now what I see in 2013 when it comes to this sort of mass production opportunity is the fact that you still need to get better tools in the hands of people who aren't data scientists who can actually still drive analytic insights. We really put a lot of our attention as we're going to 2013 into just getting help to organizations so that their data analysts can be doing the designing of these analytic apps without necessarily having an in-house data scientist.
The second thing is the fact that organizations will need to be able to take the data that they've actually sifted through, create insight around it and be able to share that more seamlessly. And the sharing concept around the Alteryx Analytics Gallery and the application focus that we put in the Analytics Gallery is one of the ways that people will be able to deliver a set of insights to a broader set of users inside of an organization.
And so where are the big challenges and opportunities going to occur? My belief is that customer analytics, in particular, are going to be a great opportunity to focus on in 2013. I say that because there's a tremendous amount of internal data that can be sifted as far as customers go – point-of-sale data, customer records that you have inside your CRM systems or perhaps hosted in a cloud source like Salesforce.com – and then actually combine that with behavioral characteristics or demographic profiles as well as buying patterns and spending behavior that you can actually get from demographic providers like Experian. If you're a B2B, you can get similar insights from a firmographic standpoint from great sources such as Dun & Bradstreet. Being able to have customer analytics, whether you're a B2B
company, and delivering more advantageous views of customers and their behavior across those internal and external sources is a real opportunity for big data and analytics in 2013. I'm suggesting that a lot of the applications that are going to go into mass production in 2013 are going to have a much more heavy bent toward customer analytics, in particular, because that’s a use case that is going to be much more seamlessly solved by blending that internal and external data sources together.
I couldn't agree with you more. Our research also shows a heavy focus on the customers' side especially with regard to social media where we can now combine both external and internal data sources to get even a more complete picture of the customers that we're working with.
Thank you so much for taking the time to provide this interesting information about Alteryx and the outlook for analytics in 2013.
Recent articles by Ron Powell