Disclaimer

This article is designed to provide information regarding Perceptual Maps and how to interpret them through an example. Data is available through Google, and the brands have been de-identified.

What is a Perceptual Map?

Perceptual Mapping, sometimes referred to as brand positioning, is a market research technique that visualizes the competitive landscape for a particular industry. Respondents are asked to rate how much they associate attributes to different brands in the market. Examples of these attributes can be, “cheap”, “high-tech”, “modern”, “reliable”, “trendy”, “healthy”, “functional”, etc. We recommend a mix of positively- and negatively-connotated words or phrases, but balance between them is not as important as in a typical scale-based perception question.

How these attitudes are elicited can vary: some surveys ask respondents to choose the company or companies they most associate with a particular attribute. These responses are then counted by company to form a crosstab of attitudes by brand. A downside of this method is that it does not reliably capture strength of feeling per brand, and also cannot measure distance between brands if multiple companies are selected. Imagine a scenario where a respondent is lukewarm toward all brands, and chooses their top brand despite having no strong preference. This respondent would like identical to another who picked the same brand because they are a loyalist, even though their attitudes toward that top choice are vastly different.

We recommend a semantic differential scale with a relatively broad width – a 1-10 NPS scale works well – that respondents apply to each brand-attribute combination. For many brands and attributes, this can be quite taxing for respondents, so we always work to prune competing brands and attitudes to the most important and impactful for the particular survey. We strongly recommend the 1-10 scale as it provides more choices than a typical 5-point preference scale, and thereby can create more differentiation between brands, without suffering from cognitive overload and subsequent fatigue that a 50-point scale might cause. A caveat of this approach is that respondents need to be aware of each brand to provide accurate data to use in the Perceptual Map. In absence of full knowledge of all brands, we suggest blending the two approaches mentioned and studying their differences to provide a clearer picture of the role information plays in constructing the competitive landscape.

Methodology (aka we have the data, now what?)

Constructing a Perceptual Map utilizes principal component analysis (PCA), a dimension-reduction technique. Imagine a matrix question where we ask respondents to evaluate 5 brands on 5 attitude statements. For each respondent, we have 25 data points we need to wrangle! Aggregating these perceptions into a mean – or count if asking for a “most” association – by brand misses correlation between attitudes and how brands are positioning themselves among these perceptions. PCA can be thought of as creating a series of lines, or vectors for the mathematically-inclined, that best explains the variation in responses. Each subsequent line, termed a “dimension”, is uncorrelated with all previous created lines.1 For Perceptual Maps, the first two dimensions are used to be able to visualize the data.

Before performing the principal component analysis, data should be scaled within each attitude question. By this we mean that the data is transformed so that its mean is 0 and its standard deviation is 1. This is done by taking the given value of a response, subtracting the mean from it, and dividing the result by the standard deviation.2 As a result of this transformation, responses are now interpreted as standard deviations away from the mean – positive numbers reside above the mean, negative values below the mean. Scaling data like this prevents response items with higher variance from receiving more weights in constructing the dimensions of the PCA; it puts all variables “on the same playing field” for the analysis.

Once the data is scaled appropriately and the principal component analysis is performed, the Perceptual Map is created using a biplot – so named as it plots both the “individuals” (the brands) and the “variables” (the attitude statements) on the same graph.

Interpreting a Perceptual Map

The best way to learn how to interpret a Perceptual Map is to look at one with supervision, so we will display the first Perceptual Map and discuss the results and interpretation.

The first map comes from Google data, where 1,000 respondents where asked to measure 10 brands (brand A, brand B, …, brand J) from 1-10 (1 representing “Strong disagreement”, 10 being “Strong agreement”) on the following characteristics:

  • perform

  • trendy

  • latest

  • serious

  • leader

  • rebuy

  • value

  • bargain

  • fun

The figure below plots the results of the Perceptual Map of this data.

Broad Notes

  • Lines that are close to each other (pointing in the same direction) are closely associated with each other
    • “serious”, “leader”, and “perform” are highly related to each other
  • Lines that point in opposite directions are not associated with each other
    • “trendy” and “latest” are not associated with “value” and “bargain”
  • Similarly, brands that are close to each other are associated, i.e. they are competing with each other
    • Brand B and Brand C are closely associated, as are Brand A and Brand J
  • Brands that are far apart (occupy different quadrants), are not closely associated
    • Brand B and Brand C are not competing against Brand A or Brand J

Comparing Brands and Perceptions

The image below provides a guide on how rank brands on the “latest” perception. The “latest” vector has been bolded in black and extended into the southeast quadrant to aid in visualization. For each brand, a red line has been drawn from its point on the graph to the extended “latest” vector so that it is perpendicular to the vector.

Brands that intersect the “latest” vector in the northwest quadrant (the direction it is going in the original Map) are associated with that perception. Brands that intersect the vector further from the origin are more associated with that attribute. So we can rank brands on the “latest” perception as: H, D, I, C, B. Brand H is considered “latest” among all brands, while brand B is still associated with the attribute, but not as much as C, I, D, and H.

The purpose of extending the “latest” vector into the southeast quadrant (the opposite direction it is going in the original Map) is that brands that intersect here are negatively associated with that attribute: respondents disagree that a brand embodies that attribute. The further from the origin they intersect, the more respondents disagree that the brand is associated with that perception. Note that brand J intersects the extended line close to the origin (the middle of the chart), so respondents do not have particularly strong feelings about brand J and “latest”, only perhaps it is slightly negatively associated with that trait. Brands A and E are slightly more negatively viewed on this attribute, but are still fairly neutral. Brand G is most negatively associated with “latest” among all brands.

Using the above figure as a guide on how to read Perceptual Maps, we can now effectively rank each brand on each perception. Let’s rank brands on the “fun” attribute that extends into the southwest quadrant. Brand A is considered most “fun”, followed closely by Brand J. These are followed by Brand D, Brand H, Brand I and Brand E. Extending the “fun” line into the northeast quadrant shows that Brand G and Brand F are considered “not fun”, but intersect close to the origin so are are not perceived strongly on this attribute. Firm C and Firm B are considered most “not fun”, with B being the most “not fun” brand. A good sign that reinforces this interpretation is that Brand B is considered most “serious” by respondents, followed closely by Brand C. Firms J and A are considered most “not serious”.

Insights Gained from a Perceptual Map

Now that we know how to read a Perceptual Map, how can it be used to improve marketing strategies? First and foremost, Perceptual Maps can reveal attitudes and associations towards your business that you never knew existed. Imagine you are Brand A in the above scenario. You now know that you are considered the most “fun” brand among your competitors, but consumers strongly feel you are not a “leader” nor are you associated with “perform”. How can you adjust tactics to be associated with attitudes you desire? There are two insights you can use from the Perceptual Map to change course.

First, you can study what a brand closely associated with your desired attitude is doing to harness that feeling – let’s pretend it is “rebuy”. Based on the Perceptual Map, Brands F and G are not close competitors of yours (see Broad Notes above), but they embody that “rebuy” attribute you want consumers to associate with your brand. How are they marketing themselves? What can you implement from their strategies to capture this feeling? If you successfully market yourself, you’ll move closer to the y-axis (the vertical dashed 0 line) as more consumers associate your brand with the “rebuy” attitude.

A second advantage of the Perceptual Map is that you can use highly correlated associations to indirectly have consumers attribute your brand to a particular attitude. If we stick with the same desired “rebuy” association, we can instead target other associated attributes to implicitly associate our brand with the desired attitude. The “value” perception is the one most closely related to “rebuy” (although they are not as correlated as “value” and “bargain”), so you could target “value” instead and through its link to “rebuy” capture that feeling as well. Highlighting the “value” either you or your product offers, especially compared to your closest competitor Brand A, can again move you closer to the y-axis in the Perceptual Map, and indirectly have consumers associate you with “rebuy”.

Conclusion

The above article has detailed what Perceptual Maps are, how they are used to visualize brand positioning among many attributes, how to prepare data for Perceptual Maps, how principal component analysis is used for the analysis, how to interpret the Map, and lastly how to use the Perceptual Map to plan a future marketing strategy.

With this information, you and your business will be better prepared to plan a series of questions to elicit consumer attitudes, better understand (and possibly critique) the Perceptual Map output of others, and plan a strategy around the insights gained from the study.

At Maximum Insight, we pride ourselves on our quality of work, transparency regarding all facets of survey design and analysis, and quantitative rigor we bring to every project. If you would like to reach out to us regarding Perceptual Map survey design, how we can help your business unlock the most from its market research data, or want to know more about services we provide, you can email Dr. Max Mathias to learn more.

  1. For math folks with linear algebra, each dimension of the PCA is an eigenvector of the correlation matrix between response items. Eigenvectors are by definition orthogonal to each other.↩︎

  2. For a given variable \(x_i\) for respondent \(i\), the new scaled variable \(\tilde{x_i} = \frac{x_i - E(x)}{\sqrt{Var(x)}}\)↩︎