Large Innovation Spaces

The Affinnova evolutionary algorithm allows you to optimize very large “innovation spaces” of billions of concepts to 3 or more top concepts. The typical concept for a new product, ad or design is composed of elements, each of which has potential variants. For instance, a potential concept with 10 possible names, 10 positioning statements, 10 varieties and 10 packaging possibilities would represent an “innovation space” of 10,000 potential concepts (10 names x 10 positions x 10 different varieties x 10 packages).

The greater the number of variants per element, the larger the innovation space. For another example, a concept with six elements produces an innovation space of 36 concepts at two variants per element; this increases to 279,936 concepts at seven variants per element.

Affinnova clients have evaluated innovation spaces ranging from just under 200 concepts to over 2 billion.

Identifying the Strength of a Concept

Such large innovation spaces pose challenges for conventional market research techniques. In mathematics the problem is called combinatorial optimization, and the typical solution is to use a search algorithm or metaheuristic. Where mathematics will typically require some fitness function to determine the relative strength of a combination – for instance, the shortest trip in the classic Traveling Salesman Problem (“find the shortest tour that visits each city once”) – Affinnova uses an interactive fitness function, in effect having respondents evaluate the fitness of a concept.

We present a set of concepts to consumers and have consumers choose from this set. Although superficially similar to a discrete-choice exercise, the underlying technology differs dramatically. The first participants see random concepts, with each element randomly assigned a corresponding variant. Their preferences of one concept over another then inform the evolutionary algorithm, which updates the choices shown in the future.

Evolutionary Algorithms

In an evolutionary algorithm, selected concepts breed with one another, creating offspring concepts with many of the same characteristics as their parents. For instance, if a child concept might take its name from one parent, its positioning statement from its other parent and its varieties might be a random mutation. In this way, the concepts shown to consumers rapidly evolve in a “survival of the fittest” process, quickly converging on the top concepts in the innovation space based on consumer preferences.

Instead of reacting to handpicked concepts, consumers choose from algorithmically assembled concepts based on the choices made by prior consumers. In essence, consumers are unknowingly collaborating with one another to rapidly converge on the top concepts within very large innovation spaces. And all this is made possible by our patented application of evolutionary algorithms.

To see our evolutionary algorithms in action, please schedule a demonstration with us.