Big data? Big deal! #ESOCong #MRX

… Live blogging from the 2013 ESOMAR Congress in Istanbul Turkey. Any errors are my own, any comments or terrible jokes in [] are my own…


Why Big Data is a Small Idea…: And why you shouldn’t worry so muchStephen Needel, Advanced Simulations, USA

  • Are companies like Google, IBM, and SAS stealing work away from market research companies? Why are clients buying into the big data hype?
  • These companies talk to marking, finance, info tech, and operations groups and completely bypass the market research group, this will kill market research
  • Big Data Vision – everything is connected, broadcast media, internet, shopping behaviours, medical history, all that stuff is connected
  • Combine it in a magical way, wave your wand, and patterns are revealed that are unseen by the naked eye, the computer can see these things
  • Big data relies on determinism (we can understand and manipulate data), data acquisition and merging, available paradigm (way of thinking to guide our use of big data), competency (must believe we can DO something once we understand the data)
  • Science is deterministic – If X then Y – we can explain behaviour: assumes behavioural consistency and homogeneity of domain
  • But shoppers are not consistent, we buy different cereals, we might buy a different one because it’s on sale
  • Shoppers are irrational, they seek variety, and are subject to moods
  • MR fails at building predictive models, and more data won’t help
  • We can’t know all that causal factors that go into making a purchase decision, assumes all outcomes are known
  • big data is an all-knowing power  –  that is SOOOO false
  • Search can be and is monitored, but we must abandon all privacy options and assumes everyone of our devices is connected to the internet including TV and radio and newspapers, your loyalty card, your credit card purchases, – dream on. And none of these data are merged, even if you think it should be easy to do
  • Big data doesn’t just recognize patterns on its own. You need to apply a paradigm to big data. You can’t just throw everything in a correlation and see what happens. It doesn’t create it’s own statistical model. And any model restricts the nature of the outcome.
  • even if you have great data, correlation does not equal causality, more data does not mean representative, action is not dictated by results, data crunching companies don’t foten have the expertise to interpret
  • It doesn’t matter if you don’t know what questions to ask, or what models to use, or if you don’t have all the data, or don’t know how to mash the data
  • Talking about big data gets us thinking in a positive way
  • Most of us have syndicated data or panel data and lots of questions reside in this data – cross purchasing, copurchasing
  • Remember you know your business better than anyone else, develop your theory of how products behave, your theory should lead you to useful questions, areas for new ideas

A New MR Mix for the New Age Information Ecosystem: Proposing big change in content mix, not processRadhecka Roy, TNS, Singapore, Sunita Venkataraman, Intel Corporation, Singapore

  • The consumer has changed dramatically. We use to think that consumers start by thinking. Consumers behave impulsively, and move from task to task without thinking.
  • business cycles are getting faster and shorter, decisions are often real-time and we immediately move on
  • market research started as independent sources of data, but now there is an explosion of information all around us
  • We have more information, it is better quality
  • 1) Directional is good enough – time is the cost of research in the new world
  • 2) from concepts to experiments
  • In the old days, farmers grew data for your reports and you had all the recipes. Now, chefs select the right ingredients and decide what to make of it.
  • embrace the change and dive in
  • start small
  • build structures to start integrating big data decision making
  • use the humanizing expertise of your research partners that none of the other players can match
  • Strike out at least one historical research project from your annual budget, mandate only short surveys, find 1 question every day to answer without primary research


%d bloggers like this: