February 12-13, 2019
Monadic designs are the gold standard in concept testing. A newly developed smart algorithm uses the strength of monadic and adds to it intelligent sample distribution where promising concepts receive more sample.
Classical monadic designs require a lot of sample, especially if many concepts are being tested, making them cost-intense and time-consuming. Since we don’t know from the onset which concepts will resonate more and which less, sample is evenly distributed and thus not efficiently used. In reality, we are interested in learning as much as possible about the top concepts, not so much about the less promising ones.
We made it our task to overcome these obstacles and developed an intelligent algorithm for efficient concept testing which relies on principles of the Bayesian Bandit. During field our algorithm learns which concepts are more promising based on answers provided so far and distributes further respondents to more promising concepts adaptively and in real-time. Together with Dynata, Factworks investigated the effectiveness of the adaptive algorithm in an online study in the U.K..
The self-learning algorithm has the power to increase efficiency with concept and naming tests and is applicable to future research studies.
The benefits of using this approach include:
Driving insight into the hearts and minds of stakeholders is Coca-Cola Knowledge and Insights’ biggest challenge. Examining which communications create most engagement with internal audiences and providing guidelines, we will [...]
Wednesday | 11:30-12:00 | Room 1