Big Data and the Value of Social by Jeffrey Hunter #ACEI_CO #InvestigAction2013 #MRX

… Live blogging from the Colombian Association for Marketing and Public Opinion Research in Bogota, Colombia, any errors are my own, any comments in [] are my own…

ACEI Bogoto Colombia

Para leer esto en español, por favor, copie pegarlo en Google. Una mala traducción es mejor que ninguna traducción.

  • [i feel bad for Jeffrey’s friend steve who is unhappy that he has to learn new methodologie 🙂 ]
  • Steve doesn’t know why big data is such a big deal and why everyone thinks about it differently
  • There are many ways of informing marketing decisions and there are many more to come; the psychic burden for client side researchers is greater and the competitive set for agency side researchers is broader
  • now we have to deal with companies we’ve never heard of before
  • characteristic of marketing information – new data sources, new users, unstructured data, really big data
  • what can we do with the big data we’ve had all along but never touched
  • social data is often free or very low cost
  • competitive set used to be people like nielsen, gfk, ipsos but now also ibm, hp, and facebook are in there
  • case study: new product launch
    • two approaches from traditional survey to web listening data
    • the two approaches yielded identical trend results for awareness (volume of buzz) and trial
    • sentiment analysis was not the same for two agencies
    • information from germany is more reliable than from US because germans are precise about their spelling [tells me the coding company needs to improve their misspelling algorithms]
    • this project was well priced and effective
  • case study #2: pricing and price promotion for electronics
    • what is the price elasticity for the new products that arrive every year, how does the new product affect the price elasticity of the old products
    • premium vs economy, this year vs last years model, each had a pricing policy
    • trad surveys require a prior knowledge even of competitor products, and there’s a lot of overhead with 40+ markets
    • new models did indeed affect price elasticity of old models, but didn’t know which models or by how much
    • near time data was actually better than real time data, near time data gives consumers time to assess the information and then share their opinion
  • case study #3: share metrics and imperfect information
    • wall street, banks, and upper management always want to know about share
    • data is difficult to capture, or it’s not even captured because companies don’t want to tell other people how much they’re selling
    • used POS data from willing companies, also used a household purchasing panel
    • large brands were most accurate in their share estimations but small brands were so poor they were useless
    • some data sources are redundant, can we figure out which ones provide no value
  • old world focused more on social sciences, new world focuses more on empirical outcomes
  • the value comes from integrating multiple data sources
  • as big data moves forward, we will be better able to decide what the good and bad (redundant) data sources are

%d bloggers like this: