Connected homes, political polls, and quicksand oh my! #CRC2015 #MRX 


Live blogged at #CRC2015 in St. Louis. Any errors or bad jokes in the notes are my own.

Leveraging methodologies and optimizing your product: How NRG developed a connected home solution

  • help consumers better understand how to use and control energy when we don’t really think about it unless it doesn’t work
  • connect/smart homes can help you control anything from an app on your smart phone, lights on and off, check on kids, see what time people come home
  • people will always want a $10 000 ferrari. stop asking about that. no one wants a $200 000 kia either. stop asking that.
  • what is the most efficient product based on costs and preferences
  • Learn about the efficient fontier – optimizing preferences and costs
  • what features do people really like and which are undervalued
  • must narrow down the features using max diff first
  • people really wanted to be able to confirm that they had closed and locked the door
  • price obviously had to be included
  • there is no right answer but you get data to make decisions [totally agree. statistics never give you th right answer. they give you something to ponder]

Political polling 2016: what pollsters and corporate researchers can learn from each other

  • “This election finally prove that most market research is probably twaddle”
  • research is used to find the idioms that people like and these words show up in speeches
  • RDD has been the most prefered data collection mode, and in some cases still are
  • compare a phone and online survey side by side, n=500 for both, 21 questions around 5 minutes
  • housekeeping variables generally matched
  • no major differences between the two groups for many variables on voting, immigration, economics
  • study that appends voter data from registered voters
  • questions were about favorability and support, e.g., i don’t really like them but i’m voting for them
  • in single select questions, trump is favoured.
  • tending data is more important than single point estimates
  • brand liking is not always a good predictor of buying behaviora
  • change in question wording can yield substantially different data
  • online data can provide a reliable supplement, if not replacement, for phone surveys [in other words, there is NO perfect data collction method. be SMART in your data collection and interpretation]

Tiptoeing through innovation quicksand: methods to die for and methods that might kill you

  • Gartner hype cycle for methods
  • 100 responses so far, convenience sample – researchers, supply side, corporate; skewed quant
  • asked about 34 techniques, no forced answers
  • 3 people said they were fired for using a technique – virtual store research, prediction marketets, A/B testing, emotion detection [i wouldn’t want to work there anyways!]
  • if you did an online survey in 1996, it might have had a big impact on your career
  • people said they were rewarded, promoted, raise for using a new technique; helped companies shift directions
  • only 64% had used online surveys, half had used focus groups
  • never use again  -mail surveys, facial recogniton
  • Career builders, Career opportunities, Career investments, Career challenges
  • Microsegmentation – can identify micromarketing action to take, but complex to implement
  • Customer journey mapping – great holistic view of customer experience, too complex or externally focused
  • Uplift modeling – trying to action individual, strong ROI, 
  • social media analytics – unprompted items of concern, difficult to decipher, not-reliable data
  • mobile intercept surveys – on the spot real data, hard to get participation
  • neuromarketing – higher price, opinions vs actions
  • microsurveys under ten questions – quick,better for niche audience, low barrier of entry
  • facial recognition – might feel it doesn’t give new information
%d bloggers like this: