Back to kaikkonendesign.fi

Perceptual Mapping

One thing I've seen in many startup pitch decks is the gap map, where competitors are plotted on two dimensions that represent important product attributes, such as price and quality.

These maps are understandably popular, because they're a great way to communicate how the company's offering differs from the competition. They give you credibility. However, the information in the map is most likely produced by the company itself, which is not a reliable source of customer perception information.

A person creating a gap map is likely to determine the dimensions based on his own judgement. They might not represent the dimensions important for intended target segments. While a company can produce a more objective gap map than customers can, objectivity is not what we're after. We want to know how customers perceive the market, whether their views are objective or not.

How can we utilize a gap map that is not affected by our own judgement? And how can we find the dimensions that customers perceive the market through? One option is to use multidimensional scaling. Let's go through an example where we use my perceptions of the automobile brands.

We'll use Fiat, Mercedes Benz, BMW, Kia and Volkswagen as players in the marketplace. First I need to get data on how dissimilar each brand is in relation to one another. Based on my perceptions, I filled this spreadsheet:

A value of 0 means that the products are very similar (distance of 0) and a value of 10 is the maximum dissimilarity.

Now that we have dissimilarity data, we run the values through multidimensional scaling. It maps the dissimilarities as distances. Running the values in R, I get this graph:

Now we have a visual representation of how dissimilar the brands are. From this representation, we can try to identify dimensions. From my experience with how these cars are priced, Kia would have the lowest price and Mercedes-Benz the highest. Looking at the rest, pretty much confirms it. So we can assign a price axis to the graph:

Therefore one of the attributes through which I perceive the automobile market is price.

I can't identify any more dimensions. In what way would Volkswagen be an opposite of Mercedes-Benz and Kia? I don't know.

You should note that this two-dimensional representation can't represent the dissimilarities perfectly–for this dataset where we have 5 variables, we would need a 5-dimensional representation. For a hint on how many dimensions are adequate, you can use eigenvalues.

To conclude, here's how the simple process goes:

  1. Survey people on how dissimilar each of your competitors are,
  2. Use multidimensional scaling to create a map of your competitors and
  3. Examine the resultant graph and see if any clear dimensions emerge.
If you liked this article, you should follow me on Twitter: @tommikaikkonen