Prediction Market is a customisable general-purpose tool for aggregating market predictions on social and technological issues. Instead of asking to review a product, advertising material, or a general statement as a consumer, respondents are asked to take a world view and bet on answers to questions about the future of a product, a market trend, or other people’s reaction to messages and advertising materials.
Example prediction questions you can pose to your audience:
For each question, you will see which prediction won. Based on the betting results, there will be either strong or weak prediction for one of the two side of the argument presented to respondents, or "no prediction" when bets are split approximately equally.
Consensus history chart shows how bet allocation evolved as data collection progressed. The final split of the bets is used for prediction.
|not something vital to people||67%|
|I don't think ear wax removal is a life altering invention||67%|
|just does not sound like a product that would improve peoples lives so much as a convienience||57%|
|Never had a need and can't see a need now.||43%|
|I HAD ONE AND IT DID NOT TAKE ALL THE WAX OUT OF YOUR EARS AND IT GAVE ME AN EARACHE AFTER I USED IT.||43%|
For each side of the argument, you will see respondents' rationale for making their prediction. Comments offered by respondents are judged by other respondents. Those rationales that are voted to be most convincing are used to persuade later respondents to predict a particular outcome. You will see top comments in your report.
|Inaccurate responses (<50%)|
|Low accuracy (50% to 65%)|
|Good accuracy (>65%)|
Calibration accuracy shows the percentage of bets placed on the correct answers in calibration questions. It shows how diligent, engaged, and knowledgeable study participants were. It serves as an indication for how reliable predictions would be.
Learn more about other parts of prediction market reports.
Prediction markets generally require smaller sample sizes than in traditional market research because respondents make predictions not just for themselves, but rather are asked to project their beliefs on the market as a whole.
For the same reason, prediction markets work well when you cannot directly access your target buyers, but can instead rely on others with sufficient knowledge of a topic. For example, you may be evaluating whether a new toy would be a hit with teenagers, but you cannot ask them directly due to legal restrictions in your country. With prediction markets, you can ask their parents and teachers to offer their judgement instead.
Conjoint.ly prediction markets are optimised for consumers and panel responses, not trained experts who routinely make predictions. Our approach allows you to use low-cost sample and your own customers as research participants. In order to familiarise respondents with the procedure we start by training them using calibration questions, i.e., questions with one correct and one incorrect option about general or industry-specific facts.
On Conjoint.ly, prediction markets are done through a gamified chatbot-like interface designed to keep respondents engaged. They never see complicated words such as "prediction market", instead simple familiar terms such as "bets" and "odds" are used. Here is what actual respondents liked about taking part in past research:
I like the gambling/game aspect. It makes it way more entertaining and interesting to do.
It was fun and made me think (like a marketing person)
The guessing, the seeing other people's responses, the odds, the betting :)
It is easy to set up a prediction market test on Conjoint.ly using the guided interface, but if you ever need to chat, please let us know. Further, you can refer to technical notes.