Phil Cotter’s comment on last week’s post here really got me going. Now that I know that suicide bombers max their credit cards before setting off to do the deed I somehow feel a gathering sympathy for the security services. So the starting point is 5 million up-to-the-limit cards? We need to funnel cash into predictive analytics urgently if anything we do is to show better results than airport security (to begin from a very low measure indeed). So I began to look for guidelines in the use and development of predictive analytics, thinking that while we wait for terrorist solutions we might at least get a better handle on marketing. I am surprized and impressed by how much good thinking there is available, so in the spirit of a series of blogs last year (Big Data: Six of the Best) here are some starting points on innovative analytics players who all have resonance for those of us who work in publishing, information and media markets. And a warning: the specialized media in these fields all seem to have lists of favoured start-ups enttitled “50 Best players in Data Analytics”, so I am guilty of scratching lightly at the start-up surface here.

In the same spirit of self-denial that drives me to abstain from a love of eating croissants for breakfast, I have also decided to stop using the expression “B** D***”. I am so depressed by publishers asking what it means, and then finding that, because of “definition creep” or “meaning drift”, I have defined it differently from everyone else, including my own last attempted definition, that I am going to cease the usage until the term dies a natural, or gets limited to one sphere of activity. So Data Analytics is my new string bag, and Predictive Analytics is the first field of relevant activity to be placed inside it. Or do I mean Predictive behaviour analytics?

I was very impressed by analysts studying our use of electricity ( Since the work throws up some lessons which we should bear in mind as we push predictive analytics into advertising and marketing. The thought that it was easier to influence human populations through peer pressure and an appeal to altruism, as against offers of “two for one”, cash bonuses and discounts is clearly true, yet our behaviour in marketing and advertising demonstrates that we behave as if the opposite was the case. The emphasis on knowing the industry context – all analytics are contextualised – and the thought that, even today, we tend to try to make the analysis work on insufficient data, are both notions that ring true for me. We need as well to develop some scientific rigour around this type of work, using good scientific method to develop and disprove working hypotheses. Discerning the signal from the noise, like “never stop improving”, are vital, as well as being hard to do. I ended this investigation thinking that even as the science was young, the attitudes of users as customers were even more immature. If we are to get good results we have to school ourselves to ask the right questions – and know which of our expectations are least likely to be met.

Which brings me to the people we should be asking. Amongst the sites and companies that I looked at, many were devoted from differing angles to marketing and advertising. But many took such differing approaches that you could imagine using several in different but aligned contexts. Take a look for example at DataSift ( It now claims some 70% accuracy (this is a high number) in sentiment tracking, creating an effective toolset for interpreting social data. Here is the answer to those many publishers in the last year who have asked me “what is social media data for, once you have harvested it?” Yet this is completely different from something like SumAll (, which is a marketeers toolset for data visualization, enabling users to detct and dsiplay the patterns that analysis creates in the data. Then again, marketing people will find MapR ( fascinating, as a set of tools to support pricing decisions and develop customer experience analytics. Over at Rocket Fuel Inc ( you can see artificial intelligence being applied to digital advertising. As a great believer in sponsorship, I found their Sponsorship Booster modelling impressive. This player in predictive modelling has venture capital support from a range of players, from Summit to Nokia.

When the data is flowing in real time, different analytical tools are called for, and MemSQL ( has customers as diverse as Zynga, and Credit Suisse and Morgan Stanley to prove it. Zoomdata ( is a wonderful contextualization environment allowing users to connect data, stream it, visualize it and give end-user access to it – on the fly. This is technology which really could have a transformative effect on the way that you interface your content to end users, and you can demo it on the Data Palette on the site. And finally, do you have enough of the right data? Or does some government office somewhere have data that could immensely improve your results? Check it on Enigma (, the self-styled “Google of Public Data”, a discovery tool which could change radically product offerings throughout the industry. Perhaps it is significent that the New York Times is an investor here.

So, for the publisher who has built the platform and integrated search, and perhaps begun to develop some custom tools, there is a very heartening message in all of this. A prolific tool set industry is growing up around you at enormous pace, and if these seven culled from the data industry long lists are anything to judge by, the move from commoditized data increasingly free on the network to higher levels of value add which preserve customer retention and enhance brand are well within our grasp.


Name (required)

Email (required)


Speak your mind


1 Comment so far

  1. Seven Starters in Data Analytics « P U B L I S H I N G on June 14, 2013 08:29

    […] more: […]