Free sandwiches and not enough focus

The supermarket shelf is a competitive space and brands invest huge amounts of money to reduce the risk of a failed product. A common way of doing this is to use focus groups where a few members of the public give their thoughts on the product. This can be done using interviews, groups or online tools, and each method has their own strengths. However, the difference here is one of scale, not of approach, and I think the industry is ready for change.

Most designers across most disciplines have had their work out into focus groups. This is pretty normal in the design process and is done for a number of reasons ranging from actually trying to figure something out to going through the motions in an ass covering manoeuvre, and most points in between. While design process and tools have grown and adapted hugely over the years, the process for carrying out focus groups has not. Typically a few members of the public  get paid to look at the designs in a room while eating free sandwiches and being asked questions by a moderator. The design and brand team sit on the other side of a two way mirror and observe them, take notes and then integrate this into the design process. Depending on the quality of work and research this can be useful in varying degrees, and all of this is fine. BUT this process is expensive and time consuming, and so often is not a part of the design process but only verifies preconceptions.

BUT, with the growth of AI, there must be a more efficient way of doing this. Focus group testing in this way must be the only kind of research where we choose not to learn, and do it from scratch each time. It seems odd that we choose to invest time and money to learn something in huge depth but don’t record this except in the intuition and experience of the designer. With advances in AI, eye-tracking data, natural language processing and sales data, it seems that a set of models to bring an AI focus group quickly, iteratively and cheaply into the design process must exist, and if there isn’t, someone should make one.

We imagine designers won’t love this idea as their creativity becomes a slave to an algorithm, and people that run focus groups won’t totally love it either as they might think it doesn’t get the subtleties of human interactions, but these groups are so small they are never representative of the buying population, so what are we really doing with focus group testing anyway? The number of times products that tested well have gone to bomb in stores suggests that this approach is a bit broken already.

We think that an ‘augmented focus group’ could harness the best of creativity and researching with focus groups, into the design process. A data driven AI approach could help to really focus design efforts at an early stage to allow for more expression later on.

We are certain that a learning algorithm could unlock a faster and better way of designing. Can someone help?