Consumer behaviour can be unpredictable, even to themselves. Measuring consumer volition can often be expensive, time-consuming and end in futile.
When using surveys in performing a market research, the widely recognized phenomenon is that customers notoriously overstate their purchase intent in their responses.
Curiously, asking the same group of people to predict the purchasing behaviour of others, rather than themselves, yields more realistic result. This concept forms the basis for prediction markets, which are becoming more popular in evaluation of success of a concept, product, or political candidate.
What are prediction markets?
Prediction markets are virtual stock markets that are run for “the primary purpose of aggregating information so that market prices forecast future events”.
They originate in the public domain (political polling, box-office returns), but they have also been applied by major corporations (Hewlett-Packard, Motorola, Intel, Microsoft, Google, and Ford to name a few) to access employee expertise to forecast answers to their business questions (sales, product features, release timing).
Prediction markets are designed similarly to the stock market or TAB Election betting. Participants are given points, virtual money or actual funds to invest in their responses. When the market closes, each “bet” is assigned a numeric value representing the probability of prediction success.
This mechanism has a surprisingly high level of accuracy. During 2012 USA Presidential Elections prediction markets the Iowa Electronics Markets and the Intrade Market are both predicted Obama’s win well before the actual election. And Hewlett-Packard’s and Intel’s employee prediction markets were overall more accurate (up to 20% more accurate) than official company forecasts.
Get a truer picture
The theory behind the prediction market is that if people have to “put their money where their mouths are”, they invest more thought into their responses. It provides a genuine insight into future market events and their value. Consensus Point, maker of the Huunu prediction market platform, compared surveys and prediction markets with the following result:
In addition, participant diversity of prediction markets can be a valuable asset for concept testing, pricing, new product development, test promotional campaign, by tapping into a wider pool of knowledge and experience.
“Wisdom of crowds” or herd of sheep?
A major criticism of the prediction markets is that the accuracy and true results seem to depend on how independently the estimates are made. If participants’ predictions are generated independently, the results can be pretty accurate. On the other hand, slight social influence has a potential to taint the result.
In the times of social media, social influence is almost impossible to avoid. So are the prediction markets always incorrect then? Certain risk in using employees for the prediction market audience lies in potential ‘groupthink’. However, using consumers as participants for a corporate prediction market should eliminate that risk. If participants are easily swayed in “betting’ on the same outcome, they would also be swayed into exhibiting similar purchasing preferences. Social influence has always been marketer’s friend and the cornerstone of branding.
Why not bet on it?
Given all the accuracy and added benefit of getting the customers more excited about the business, why haven’t prediction markets replaced surveys and focus groups?
Well, the complex calculations that require specific software programs in comparison to a lot easier surveys, might be one of the reasons.
There is also an argument that not all the consumers who provide business feedback are driven by materialistic incentives.
It might be so. However, people who do not bet are not necessarily more objective or truthful because of the lack of tangible interest. At least when people have bet incentive, it is clearly in their interests to be right, and that could lead to more accurate and unbiased information about the market.
Posted by Katerina King, ID: 214229162 (WordPress name: kkin108)