FEATURE30 May 2012

A Google for numbers


DataMarket CEO Hjalmar Gislason wants to make all the world’s data available at the click of a button, he tells Brian Tarran.


You can’t fault that for a statement of intent – an Icelandic start-up whose website currently attracts 30,000-40,000 unique visitors each month and has 30,000 searchable datasets to its name, set on doing for numerical data what Google did for text: that is, putting all the information you might ever need just a keyword-search away.

Five years is a long time, and DataMarket still has a long way to go. The information currently in its database is drawn primarily from open sources, from places like Eurostat, the International Monetary Fund, the United Nations and the World Bank, making it a good resource for investigating macroeconomic trends, but limited when it comes to understanding data at a micro level – individual business and consumer markets – which is where Gislason sees the greatest opportunity.

DataMarket was designed to “scratch” one very persistent and irritating itch, Gislason says: the difficulty in both (a) finding and (b) using the right information to support decision-making. Locating the data you need is only half the battle, Gislason says. It is often kept in so many different formats – spreadsheets, PDFs, proprietary databases, PowerPoint presentations – that extracting the information into a usable and comparable format is a struggle in itself.

DataMarket’s aim, therefore, is to take all the world’s data and store it in a normalised format and make it retrievable – and viewable – at the click of a button. Gislason demonstrates this by searching for and pulling in data on carbon dioxide levels recorded in Mauna Loa and overlaying it with information on global temperatures (below): data from two different sources, accessed and charted in less than five minutes, where it might have taken a person as much as one-and-a-half hours to achieve the same thing manually.

In another example, we search for “oil production libya”. DataMarket produces 33 matches. Gislason selects one and a line chart is displayed, alongside which is listed all the other countries for which DataMarket holds information on oil production. Check the box for a country’s data to appear in graph form. Select them all and switch to a bar chart view to see the long-tail of world oil production (below). Each chart is accompanied by buttons for social sharing, short URLs and embedding as well as exporting as PDFs, jpegs or as raw data in a variety of formats.

But oil production and carbon dioxide data does not help address the business questions Gislason and DataMarket set out to answer. For that they need access to the wealth of proprietary data held by information companies – including market research agencies. This data would be searchable in the same way as the public data, but DataMarket will charge users to access it. The Economist Intelligence Unit is already on board but to win other converts, Gislason is pursuing a different approach.

He’s recently relocated from Iceland to Cambridge, Massachusetts, to set up a sales office to license the DataMarket platform for private use. Think of it as an off-the-shelf version of the client data portals some research firms have invested in. DataMarket acts as a private search engine and visualisation tool for a company’s own data, including both time-series information and survey results – with which users can run cross-tabs for any two variables, chart them and export them for debriefs.

Out of these licensing arrangements, Gislason hopes to attract additional data partners for the public-facing DataMarket. It has five platform licensees so far, of which two are research companies – Yankee Group and Lux Research. Meanwhile, in it’s home country, DataMarket has long worked with the research group Capacent.

It shouldn’t be too much of an effort for Gislason to win other converts in research, where there remains an unmet need for simple-to-use visual tools for accessing and interrogating a wealth of consumer, market and business data.

The licensing model “can be a profitable business”, he says – but success here won’t distract him from his long-term goal. It’ll cost a lot of money and carry a lot of risk, but building a Google for numbers is an opportunity he can’t ignore.