tecosystems

Infochimps and the Future of Data Marketplaces

Share via Twitter Share via Facebook Share via Linkedin Share via Reddit

Much of the focus of Infochimps’ recent announcement that they had productized their internal Big Data platform focused on the “Big Data-Heroku” angle, which is understandable. While the array of open source assets for attacking data problems at scale is large, standing up a cluster can be intimidating for practitioners new to the field. Besides the unfamiliar querying mechanisms like MapReduce, there are a lot of moving parts involved. For those willing – to borrow Mårten Mickos’ metaphor – to trade their time to save money, the experience of building Big Data stacks from the ground up using tools like Chef and Puppet can be enormously rewarding. But as commercial open source vendors have proven satisfactorily over the past decade or so, there’s a substantial number of businesses who prefer to trade money to save time. Hence the attraction and interest in Big Data as a Platform.

But the wider significance of the Infochimps “pivot” may be that one of the original would be data marketplace vendors found the opportunity sufficiently wanting to reposition itself as a more traditional software play, productizing their big data experience. This is the clearest indication yet that data marketplaces may be the latest “Application Service Provider” cycle, as in right idea, wrong time. ASPs, remember, pioneered what later came to be called Software-as-a-Service, selling to a market who was not yet prepared to either software or consume it strictly through a browser. Most could not adapt, and either dried up and blew away or were acquired for their assets, their people or both. A decade later, and SaaS is not only an accepted application delivery model, it is the dominant one in many contexts. The ASPs mistake wasn’t product, it was timing.

Certainly the friction towards the marketing and sale of data as an asset is, at present, high. In comparing current market perceptions of data to enterprise acceptance of open source a decade ago, we’ve argued that mundane issues like licensing make data marketplaces at present largely inefficient, which in turn acts as a drag on adoption in a vicious cycle.

As much sense as data marketplaces make in a world that’s increasingly oriented towards evidence based decision making, it may be that the market just isn’t ready for marketplaces whose primary product is data. That their time will come is not, for me, a question. What’s not yet obvious is whether the time is now, or not yet. The Infochimps news is evidence of the latter, but the pluarl of anecdote is not data as they say. We’ll be watching the various data marketplace vendors, then, for signs of progress or regression. The cost of the latter may be high – much higher than is anticipated, in fact. But that’s a post for another time.

In the meantime, give Infochimps credit for making the necessary adjustments in terms of their approach and positioning. History is littered with the carcasses of startups that couldn’t adapt to being ahead of their time; with their new product, Infochimps is far less likely to be one those.

4 comments

  1. I agree with your basic point but “timing” should IMO be taken to include the coming together of related and necessary pieces. For example, in the case of ASPs, I’d argue that the evolution of technologies like RESTful interfaces was one of the things that made more recent SaaS take off whereas ASPs mostly didn’t. 

  2. Nice post. Let me put side-by-side a different take on this:
    Edd Dumbill, On Oreilly Radar (
    http://radar.oreilly.com/2012/03/data-markets-survey.html), put the argument this way: “A principle of big data is that it’s often easier to move your computation to the data, rather than the reverse. Because of this, we’re seeing the increasing integration between cloud computing facilities and data markets: Microsoft’s data market is tied to its Azure cloud, and Infochimps offers hosted compute facilities.”
    While the discussion here also makes sense, I personally tend to agree more with Edd.

  3. I see the data platform serving to the data marketplace as Amazon’s Kindle does to their eBook business. Amazon currently sells more e- than physical-books, yet countless prior efforts to build a winning eBook business had failed. Amazon won by making it transformatively easier to enjoy digital reading, which in turn powered their large and durable content market.

    We had multiple conversations with companies who saw the value in diverse adjacent data — who knew that their own data lacked the necessary explanatory variables — but who basically said “we want that data, but don’t have your stack; let’s talk more once our internal capabilities improve”. Once we’ve helped a company regain mastery over a terabyte-scale data flow of customer analytics, the value of say augmenting geolocations with implied demographics is clear.

    The enterprise data marketplace that wins is the one that brings “the rest of the world into your data warehouse”, but I’d always thought the enterprise connector would be a late-game piece. Instead, we’re betting on it as the conduit that unlocks demand for data by the millions of rows.

  4. […] RedMonk co-founder and analyst Stephan O’Grady suggests this pivot on Infochimps’ part may be due to the concept of the data marketplace being before its time. O’Grady has written about how licensing concerns is holding back data markets, by making it harder for companies to both sell and consume data. […]

Leave a Reply

Your email address will not be published. Required fields are marked *