This document introduces prediction markets as a marketing tool that can harness the wisdom of crowds to predict outcomes. It explains that prediction markets apply the theory that people are better at predicting group behavior than their own. The tool aggregates responses from consumers who are rewarded for accurately predicting others' responses, rather than just completing a survey. This provides more nuanced insights into consumers' holistic thinking compared to traditional quantitative methods. The document outlines how the tool works and is tailored for different marketing projects. It also provides examples of prediction markets' accuracy compared to polls and notes their widespread use by major companies.
Bob Bernstein and Peter Gould of Sunbelt Dimensional, Inc., a solution provider at the marcus evans PharmaMarketing Summit 2012, on designing more effective pharmaceutical product promotions.
Interview with: Bob Bernstein, President, & Peter Gould, Sales and Design Manager, Sunbelt Dimensional, Inc.
Bob Bernstein and Peter Gould of Sunbelt Dimensional, Inc., a solution provider at the marcus evans PharmaMarketing Summit 2012, on designing more effective pharmaceutical product promotions.
Interview with: Bob Bernstein, President, & Peter Gould, Sales and Design Manager, Sunbelt Dimensional, Inc.
Bernard cools - DM, the good, the bad or the ugly for branding?bpost
DM is ook een efficiënte tool om een sterk merk te creëren. Een uitgebreid overzicht van de efficiëntie van direct mail, beginnende bij nieuwe betacijfers i.m.v. de herinneringsgraad.
Presentation given at the 2012 IASA Annual Conference on the use of analytics in the insurance industry. Case study examples provided with extrract of recent survey.
BBDO Connect: What a difference DM makesBBDO Belgium
In this digital age, direct mail and e-mail have to reinvent themselves to earn their spot in the media mix. How can intelligent use of customer data, compelling creative and innovative production techniques give your acquisition and loyalty programmes a ROI boost? Fifteen cases with proven results – both B2B and B2C – will be your source of inspiration for response-driven 1-to-1.
The Social Side of Behavioural EconomicsDavid Perrott
Understanding how deeply hardwired our brains are to be social gives us a better understand of how we make judgments and decisions, creating the right foundation for new forms of communication and design.
Bernard cools - DM, the good, the bad or the ugly for branding?bpost
DM is ook een efficiënte tool om een sterk merk te creëren. Een uitgebreid overzicht van de efficiëntie van direct mail, beginnende bij nieuwe betacijfers i.m.v. de herinneringsgraad.
Presentation given at the 2012 IASA Annual Conference on the use of analytics in the insurance industry. Case study examples provided with extrract of recent survey.
BBDO Connect: What a difference DM makesBBDO Belgium
In this digital age, direct mail and e-mail have to reinvent themselves to earn their spot in the media mix. How can intelligent use of customer data, compelling creative and innovative production techniques give your acquisition and loyalty programmes a ROI boost? Fifteen cases with proven results – both B2B and B2C – will be your source of inspiration for response-driven 1-to-1.
The Social Side of Behavioural EconomicsDavid Perrott
Understanding how deeply hardwired our brains are to be social gives us a better understand of how we make judgments and decisions, creating the right foundation for new forms of communication and design.
2. Briefly
Prediction Markets are a means of aggregating the inherent wisdom of the
crowd in order to predict an outcome
Prediction Market theory applied to consumer research is a powerful, technique that
enables marketers to evaluate ideas quickly, effectively and reliably
proteanprediction Collective Wisdom Engine is a simplified, streamlined
research methodology based on prediction market theory
Page 2
3. A marketing tool for today’s changed world
Enables marketers to harness consumer opinions once or at multiple points during the
project development
Talks to consumers at the speed they are used to
Allows marketers to benefit from consumers’ marketing savvy
Gives consumers a respite from research complexity by offering them simple ways to
make their opinions known and express them in their own language
Page 3
4. The underlying premise
2 Simple yet profound tenets
• Ask what they think other people would do, not what they personally would do
• Reward them for getting it right
1 Complex and profound tenet
• Apply prediction market algorithm to weight the responses
Page 4
5. Simply what you need to know
Will it work?
Why?
Why not?
Page 5
6. Three core questions If we asked 100 people like yourself whether this
1. campaign would make them want to buy
BRAND X, how many would say “YES!”
Now, how much of your $ would you bet
that your answer above is right? You only
2. have $100 to bet between the two
options?
You estimated that more people
would be convinced to buy
3. [Product} by watching „AD D‟ ….
Why did you think this?
Page 6
7. Plus: tailored to the needs of every project
Questions
Demographic and behavioral screening questions
Pre-exposure brand and competitive awareness and preference
Post exposure preference
Full range of diagnostic testing
Sample
Customer lists, hand raisers, brand enthusiasts
On-line panel
Any other source
Page 7
8. Prediction market difference
Traditional Quant Study proteanprediction
Focused on what people
think
other people would say, not
Subjective personal their narrow personal biases
Remove Bias
opinions
People are better at predicting the
behaviour of others than their own
behaviour
Respondents rewarded Respondents are rewarded for
for completing the
Engage respondents thinking about the question
survey, not honesty or
accuracy and being right!
Deconstructs evaluation Encourage irrational
Based on respondents’ into rational component of forming
holistic thinking bites, eliminating judgments and making
“irrational” thinking
decisions
Page 8
9. Prediction market difference
A more nuanced outcome
Looking at the results of this
actual study, the deeper nuance
Comparison ProteanPrediction of ProteanPrediction can be
vs. Average Responses clearly seen.
29.2%
Using the average value for each
Statement D: 33.4% statement (Red) would have lead
29% to a conclusion that Statement D
was far and away the best idea.
26.8%
Market Result Using the percentage of the
Statement C: 24.0%
sample that selected each
20%
statement as their “favorite,”
(Green) dampens “D” ‘s lead, but
16.7% changes the picture for the
Average
Statement B: 14.5% number two position – “A” is now
13% equal to “C”
14.8% In the ProteanPrediction (Blue)
Statement E: 16.9% Average of result, the difference between the
13% "favourite " lead and second closes
concepts significantly, indicating that the
12.5% market place has very nearly as
Statement A: 13.9% much “heart” for “C” as they do for
21% “D”
0.0% 15.0% 30.0%
Page 9
10. Visual analysis of open ended question
Comments about Statement A
Total sample
Page 10
12. Widely used
• Iowa Electronic Markets: political predictions more accurate than the most accurate
polls at least 75% of the time
• Hollywood Markets: Predict box office receipts
• Used by: Google, Hewlett Packard, Wrigley (Global); Kraft; GE; Microsoft;
Intercontinental Hotels Group; GM’ etc.
Page 12
13. Accurate predictions
1. In a study of prediction markets versus traditional research, professors at
Northwestern University found that prediction markets were closer to the outcome of
US Presidential elections 74% of the time compared to traditional polling (using 964
polls covering the period 1988 to 2004)
http://www.northwestern.edu/ipr/news/predictionmarkets.pdf
2. GM Car sales, November 2008
Forecast Actual Variance %
Chevy Light Chevy Light Chevy Light
Cars trucks Cars trucks Cars trucks
GM forecast
150,000 250,000 95,000 152,000 36.7 39.2
(beginning of month)
Edmonds estimate of 152,000
125,000 215,000 95,000 24.0 29.3
sales (end of month)
prediction markets (7
97,000 160,000 95,000 152,000 2.1 5.0
days into the month)
prediction markets
107,000 180,000 95,000 152,000 11.2 15.6
(end of month)
http://www.crowdclarity.com/learnmore.htm
Page 13
14. The science
These two academic papers give an interesting overview into some of the
academic thinking behind the theory of prediction markets.
Additional Links
http://www.youtube.com/watch?v=keVL0PkCpaQ&eurl=http%3A%2F%2Fwww%2Econsensuspoint%2Ecom%2Fpredicti
on%2Dmarkets%2Dblog%2F&feature=player_embedded
This link connects you to a video of the CEO of Best Buy talking about their use of Prediction Markets. Given their recent
business collapse, I am not sure they are necessarily the best example
http://www.hsx.com/
This is the link to the Hollywood Stock Exchange, which is probably the most famous prediction market site – it has
become an extremely important tool for movie producers to judge the potential of their future movies before they make
them.
http://www.pbs.org/wgbh/nova/sciencenow/0301/04.html
This is the most fun of all of them – PBS video that makes it all clear.
Page 14
15. For more information
call Laurence Bernstein
416 967-3337 x 101
bernstein@proteanstrategies.com
www.proteanstrategies.com
Editor's Notes
Prediction Markets are a means of aggregating the inherent wisdom of the crowd in order to predict an outcomePrediction Market theory applied to consumer research is a powerful, technique that enables marketers to evaluate multiple ideas quickly, effectively and reliablyproteanprediction Collective Wisdom Engine is a simplified, streamlined tool based on prediction market theory
allows people to participate in marketing decisionstalks to them at the speed they are used toallows people to apply their marketing savvygives them a respite from complexity by offering them simple ways to make their opinions knowncredits them being consumer-kings
two simple yet profound tenetsask what they think other people would do, not what they would doreward them for getting it rightone complex and profound tenetapply prediction market algorithm to weight the responses
Great at giving friends advice, lousy at taking our own adviceOne of the reasons purchase intent measures are so lousy at predicting purchase intent In some companies over 80% of new products brought to market fail Overall, more than 90% of products brought to market will not be there is 3 years