So many companies fall prey to the idea that “If you build it, they will come” that it is refreshing to encounter a CEO who accepts the brutal reality of business, that no one cares, that nobody wants your product unless they are systematically convinced, that creating a successful company goes far beyond building a technology product but actually involves building a business.
Mike Tuchen, CEO of Talend, is one such CEO. In a couple of recent conversations we covered a lot of ground about what it means to build a business and how to avoid the many fallacies that are alive and well in the technology business.
Talend is an open source data integration platform that competes with companies like Informatica, IBM, Tibco, Abinitio, Pentaho, and MuleSoft and many others. Data integration is a hot topic these days because the need to move data around with precision, transform it, and track where everything is, has grown as the use of data has grown.
Data integration was a big business before the big data craze and has become a bigger business since. The way I explain this is that the world has transformed from an architecture based on a single repository, a data warehouse, to one that now includes a big data repository, often called a data lake, special purpose databases for graph processing, in-memory analytics, mobile back ends, and so forth. This transformation means that every company will have to get good at creating and managing data supply chains. (See “Why Building A Distributed Data Supply Chain Is More Important Than Big Data” for more on this concept).
The growth in the size of the total addressable market for data integration companies has spurred significant M&A activity. Pentaho, an open source-based company, was bought by Hitachi Data Systems in Feburary, 2015. Informatica, a closed-source vendor, was taken private by Permira Advisers in April, 2015. Talend is likely on the way to an IPO at some point in the next year or so.
Talend’s product strategy has been to keep up with trends in big data and to provide advanced capabilities for master data management, one of the key needs in a data supply chain. Talend’s business model is based on offering an open-source product that is enhanced by advanced functionality that requires a license. A individual developer can download and use the open source version for free and get a lot of value out of it. But if companies want a broader set of enterprise features, including collaboration, management, monitoring and support,they must purchase a subscription to their enterprise editions. This is theopen core model that was so-named by Andrew Lampitt in 2008.
When Does Open Source Help Build a Business?
Unlike many executives at open source-based companies, who don’t really have a clear story about how giving away software will end up creating a good business, Tuchen has a sophisticated analysis of when open source works and when it doesn’t and a clear vision of Talend’s recipe for success. (Some of his beliefs align with my analysis of the principles driving Red Hat’s success set forth in this story “Will There Ever Be Another Red Hat?”)
Tuchen’s theory of open source goes like this:
Principle 1: Open source works best as a disruptor to existing markets. Tuchen believes that open source succeeds most naturally as a disruptor to existing markets. The open source model gives a company an advantage in terms of R&D spend, because a significant amount of the product development and extensions are created by the community of users. Open source also provides a natural marketing model in which users can download the product and use it free of charge. If users like what they find, they can then upgrade to an enterprise subscription. James Dixon, CTO of Pentaho, hasargued for years that the cost of marketing and selling closed-source commercial software means that most vendors don’t make any money from the initial sale. But this model doesn’t work if there isn’t already a population of people interested in downloading and using the software. If you are creating a market, the open source model doesn’t make sense because each sale will be a hard fought process and you need to be rewarded with a high price.
"It’s really hard to create a new market. People should recognize it as a very, very high-risk endeavor,” said Tuchen “To create a new market you have to find early adopter customers that want to try something out that has never existed in the world before, and you’re going after unique problems. You’re also dealing with low volume leverage. People don’t come swarming en masse to download and try out products they have never heard of before. Creating a new market is an evangelistic, low volume, high touch, high price point kind of sale, usually with a lot of services attached to it. It makes more sense to have a proprietary model and charge high price points. It’s the exact opposite of open source."
Principle 2: Open source companies need a large total addressable market. Tuchen acknowledges right up front that Talend will make less money per sale than a proprietary product like Informatica. A lower price for a solid collection of well-supported capabilities is a large part of why open source can be disruptive. But unlike other open source companies, Talend acknowledges that this strategy only makes sense if you have a large total addressable market (TAM) already. The lower price can support a business only if there are a lot of people already looking for a solution. In other words, the open source model doesn’t work well to create a market for commercial software.
“You have to be in a big market for the open source model to make sense. If you go after a medium sized market, a billion dollar market’s just not big enough to try to ‘disrupt’ with open source. If you have a big enough TAM, you can gear yourself at a significantly lower price point than the established competition, the legacy players, and you still have plenty of TAM,” said Tuchen. "Then you design your distribution model to fit the economics of the price points you’re targeting. Once you have that right, you’re just back to plain old company building."
Principle 3: The open source model is not a substitute for business building. The principle of investing is really simple: buy low, sell high. But making the right decisions of what to buy and when to sell is not so easy. The principles of open source are simple as well: charge a lower price for a solid product developed with the help of a community. Use open source to reduce R&D and marketing expenses. What is not simple is deciding what product to create, what R&D to perform to support what unique features, how to position and market your product to attract buyers, and hiring a building a world class team. Remember, open source, especially commercial open source, doesn’t mean no marketing, it just means a different kind of marketing. Performing all these activities effectively inside the constraints of a budget so that you make money is what Tuchen means by business building. This process is much the same for an open source and commercial software company, even if the constraints differ in some ways.
"Lots of companies—whether they’re proprietary or open source—are raising crazy amounts of money. It allows them to be incredibly sloppy in execution,” said Tuchen. “This doesn’t help anyone in the long term. I sometimes think our European roots have made it easier for Talend to keep its eye on the ball and remember that we are always playing with real money."
Examples of Companies Going Against Tuchen’s Principles.
So, Tuchen’s story for Talend, the one he will tell if he takes the company public, will be clear and straightforward. Talend is a aiming to disrupt a large established marketing with a lower-cost open source product and some expertly executed business building. Here are some of my views on companies that are doing things contrary to Tuchen's ideas for some interesting reasons.
Contrast Talend's story to Hortonworks, which I have written about at great length (See “Can Hortonworks Dominate the Hadoop Market?”). Hortonworks went public at the end of the year, the first major open source IPO since Red Hat went public in 1999. The company is offering support subscriptions for 100% open source Apache Hadoop. While Hadoop addresses the needs of a large existing market for data management, data warehousing, and business intelligence, it does so in a new way. Talend’s offering is far more a substitute for Informatica than Hadoop is a substitute for Teradata or other data warehouse vendors. In fact, in most large companies Hadoop and the data warehouse works together. So, Hortonworks along with Hadoop-based companies like Cloudera and MapR and other large market participants are paying the cost of creating the market. That’s all well and good if you are getting full price, but part of the open source model is that your product must be cheaper. If it isn’t, why bother making it open source? Hortonworks's answer is that they expect the total addressable market to be huge and that they will gain the bulk of it because of their open source model. In addition, one of the key claims for value for Hadoop is that it helps companies gain access to a trove of insights contained big data. So although Hadoop is indeed cheaper in terms of cost per gigabyte, the story isn’t only about cost, but also about finding new types of value. Based on their financials, my view is that Hortonworks will likely have to raise more money before they become profitable, although the company says they will achieve breakeven based on adjusted EBITDA in early 2017 without having to raise new moeny In addition in the Hadoop market, the R&D advantage is muted because the Hadoop open source companies fund the bulk of the development. (See “Can Hadoop Survive Its Weird Beginning?”). If you think this story seems a bit hopeful, you are not alone. Hortonworks has a huge short interest that is betting something will go wrong (See “What Will Hortonworks Q1 Earnings Tell Us?”). Recent jumps in the price of Hortonworks stock may indicate the shorts are getting worried. The next report of short interest will be interesting.
Aerospike is another example of an open source company that is violating Tuchen’s principles. Aerospike is a high performance database company that has some large clients in the AdTech industry. Aerospike is a database designed to deliver incredible fast response based on a combination of advanced software and expert use of SSD technology. In June 2014, at the same time Aerospikedisclosed a $20 million round of series C funding, the company also announced it was open sourcing its software. Given that Aerospike is attempting to create a new market and that will be built at first through a modest number of high-priced sales, it is not clear what advantage open source confers. Aerospike is a product that is aimed at the high performance end of the market. While the advanced technology companies may be able to do a free trial on their own using open source, it is hard to imagine a groundswell of people trying and then buying Aerospike. In other words, when someone uses Aerospike, he or she will have to be sold. If so, why bother making it open source?
Peter Goldmacher, Aerospike VP of Strategy and Market Development, agrees with the spirit of Tuchen’s principles, but says that selling to developers demands an open source model. By allowing developers to download an open source version, Aerospike is getting a lot more interest and awareness. "When we made Aerospike open source our close rate percentage went down because for a number of prospects, unsupported open source was good enough. However, the number of deals we got into went way up because developers could look at our code and realized that our product was enterprise class, and they needed incremental functionality and support for mission critical deployments ” said Goldmacher. “Before we were open source, every deal we were in was hand to hand combat, and it wasn’t working. We were not able to scale. Once we went open source, our community exploded.” Goldmacher said a key part of driving sales is offering proprietary category capabilities, such as cross-data center replication, that are needed by the higher performance customers and the complement the open core. He also points out that even the most advanced companies won’t run software in production without support. So, when developers are controlling the decisions, being open source may be a must, even if it goes against some of Tuchen’s ideas.
With Talend, Cloudera, and MapR coming to the IPO market at some point in the next year or two, along with several others, we will get a clear view of the value of Tuchen’s principles about how to create a business based on open source. I suspect that the fundamental learning will be to de-couple the idea of open source from the detailed, painstaking work of building a business. If that happens, then Tuchen will have made a major contribution to the software industry.
Follow Dan Woods on Twitter:
Dan Woods is on a mission to help people find the technology they need to succeed. Users of technology should visit CITO Research, a publication where early adopters find technology that matters. Vendors should visit Evolved Media for advice about how to find the right buyers. See list of Dan's clients on this page.
