Sometimes when looking at the results from survey data, we see something that makes us say “huh?” or “that doesn’t look right”.  When the odd results persist after verifying the data were processed correctly (always a good practice), there is typically still a logical answer that can be uncovered after doing some digging.  Sometimes the answer lies with something that we will call “unusual statistical phenomena.”  This is part 1 of a series that will look at some of these interesting – or confounding – effects that do pop up now and then in real survey research data.

This time we will look at Simpson’s Paradox.  And we aren’t referring to the fact that Bart Simpson never seems to age while the rest of us do.  It is actually a phenomenon first described by the statistician Edward H. Simpson in 1951.

It’s easiest to understand this phenomenon through an example.  So, let’s say that we have two ads that have been on air, ad A and ad B.  In our tracking survey among adults 18 to 65, we will ask respondents if they recognize having seen each ad on air.  Earlier in the survey we ask Purchase Intent for the product which is featured in each of the two ads.  From these results, we will compare Top Box Purchase Intent among respondents who recognized each of the two ads.  The results in the table below show somewhat higher Top Box Purchase Intent for Ad A:

However, the client is also interested in seeing the results among each of two age groups: age 18 to 39 and age 40 to 65.  When we table those results, we find something that just doesn’t make sense.  Purchase Intent is slightly higher for Ad B among both age groups – a reversal from the overall results.  How can that be!

After verifying with data processing that the data are correct, we have our team dig into the data to figure out what is going on.  Finally, an explanation is found.

Ad B was aired heavily among programming targeted to a younger audience, while Ad A was primarily aired in general interest programming – which skews to a slightly older audience.  Hence Ad B had much higher recognition among the younger age group – and as a result, a much higher proportion of young people in the set of respondents among whom purchase intent was calculated.

The table of base sizes shown below reveals this imbalance. When combined with the younger age group’s more skeptical nature (and lower results) when it comes to Purchase Intent – especially in our category – the apparent anomaly is explained.

This is an example of Simpson’s Paradox.  It is a phenomenon in which individual subgroups all show the same trend in results, but the trend reverses when the subgroups are combined.  This occurs when there is a confounding variable that causes an imbalance in base sizes such as we saw above.  In our example, the confounding variable was the differing recognition levels for the ads among the two age groups.

Simpson’s paradox shows us the importance of knowing and understanding our data and keeping a watch out for the kind of confounding factors that could end up misleading us if we don’t account for them.

All marketing is built upon mental models (conceptual frameworks that allows humans to make sense of the world) that explain how marketing works and inspires and guides action.

Many mental models are not explicitly stated, and this leads to confusion when different members of the marketing team have different models.  In our experience, companies that outperform have mental models that everyone uses, is understood by the ‘C’ suite, and provide a link to financial outcomes.

Many models have been proposed to explain the marketing process.

‘Practical men who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct academic scribbler…’

This is certainly the case with market research when every decision is influenced by the supplier’s mental model (most suppliers do not acknowledge the true academic underpinnings of their systems), which may or may not be congruent with the corporation’s model.

Models that have been proposed:

• AIDA (1898) Elmo Lewis’ customer journey from the moment that a brand or product attracts consumer attention to point of action/purchase.
• The Purchase Funnel (1924) William Townsend’s adaptation of AIDA.
• DAGMAR sometimes called ACCA (1961) Russell Colley added an important pre-awareness stage to the funnel. (Awareness, Comprehension, Conviction, Action).
• Moment of Truth (1986) Jan Carlzon model, any time a customer encounters a brand, they have an opportunity to form an impression.
• ATR-N (1997) Ehrenberg, emphasis the importance of post-purchase experience and interactions (nudges).
• First and Second Moment of Truth (2005) AG Lafley built upon Carlzon’s moment of truth to distinguish between looking at the product (first moment of truth) and then using it (second moment of truth).
• The McKinsey Consumer Decision Journey (2009). McKinsey introduces an “active evaluation” stage and updates the model making it less linear introducing a loyalty loop.
• ZMOT (2011) Google extends Carlzon’s and Lafley’s model to include a zero moment of truth when you start to learn about a product or service for the first time.
• How Brands Grow (2012) Byron Sharp.  Sharp’s insight is that the real challenge of marketing is all about availability; Mental availability that bring the brand to mind in a buying situations and Physical availability through Presence, Prominence and Relevance enhanced by a consistent memorable set of distinctive brand assets.  The primary goal of all brand strategies should be growth through customer acquisition rather than loyalty which is a function of habit.

At MSW, we offer our own – Predictive Brand Growth Marketing Model™

Five components describe the marketing process:

1. Long Term Brand Equity
2. Brand Perceptions
3. Brand Relationships
4. Creative
5. Marketing Gap

1.  Long Term Brand Equity

The Marketing Accountability Standard Bureau (MASB) conducted an 18-month tracking study covering 120 brands in 12-product categories, to identify a suitable metric that explained sales.

The results of MASB’s evaluation were presented to the ANA, AMA and ARF and have been written about in The Economist, The International Financial Review, The Journal of Brand Management, and the CFO Magazine.

Preference proved to be a better fit to market share than any other market research metric.

So, what is Preference?  Preference is a behavioral observation.  We have known since 1968 that people are sometimes incapable of articulating what they feel and want.  Consequently, we offer respondents the opportunity to win prizes in defined categories (prior to any survey research priming), and this is a measure of the desire for a brand, i.e., the strength of long-term brand equity.

Preference which ONE brand do you want to win is an extreme form of consideration.  Our data proves that Preference more strongly correlates to sales that does Consideration (R² = 0.76 vs. 0.42).

This approach has been proven to be a more accurate way of measuring brand desire than direct questioning, and it works around the world without any cultural adaptation, something that is not true for alternative measurements.

Preference has been proven for FMCG/CPGs, Durables including Automotive and Apparel, Services (Entertainment, Finance, Telcom & Travel) Retailers (both on and offline) including Food Service.

 

Preference is primarily the result of two elements: Brand Perceptions and Brand Relationships.

2.  Brand Perceptions

All successful brands have a set of distinctive brand assets (sensory cues: color, logo, design, character, jingle, etc.) that aid memory encoding and act as signals to enhance availability.  Additionally, all brands also have a differentiated positioning, (a reason to be).  Take Snickers as an example, the brand has four distinctive brand assets: logo, color, the rhombus shape, and the product tear revealing the ingredients.

Additionally, the brand has a very differentiated positioning based around a single idea: Hunger Satisfaction.  This combination has made Snickers the consistent leader in a highly competitive category for many years, around the world.

Our experience shows that the most powerful brands have the shortest, simplest expression of the differentiated position.

Every category has a set of functional attributes and delivery on these is usually mandatory; these are table stakes that provide entry to the category.  However, it is VERY hard to build a sustainable differentiated positioning on functional attributes (every product innovation can be copied).  Brands need to think about the emotional need that they address, and how to establish emotional intimacy.

In 2004 MSW conducted a study (57 brands, 12 categories and 31,052 evaluations) to determine which perceptions contributed to brand preference in a step toward building what would become our RDE framework.

We identified seven attributes that did this: Relevance (for someone like me), Differentiation, Willingness to recommend to a friend, Consistent high quality, Trust, Worth paying more for and Leadership.  The first three provide 73% of the explication.

We formalized the measurement of RDE using an implicit technique, and the weighted share of this combined attribute resulted in a close fit to Preference with an R² of 0.86.

 

Implicit measurement is something that all of us use in everyday life, without thinking.

Our brains are constantly judging the veracity of what people tell us, and we do this by comparing the speed of response for any question to what we know about the person we are talking to.

Every human has his/her standard reaction time, and slower than normal answers suggest that people:

• Don’t want to criticize
• Are trying to be polite or politically correct
• Haven’t made up their minds yet

Reaction times are behavioral measurements of certainty when combined with a five-point agreement scale. The quicker people respond the more certain they are of their answer.

The science behind Reaction Times has been known for a long time.

 3. Brand Relationships

The second element driving Preference is the brand relationship.  MSW uses a segmentation model that places every individual into one of eight groups for each brand in the category.

We utilize a relationship decision tree to identify the strength of the brand relationship with customers.

1. The consumer is either aware of the brand or not aware
2. If aware, they have tried the brand or not
3. If tried, they are currently using or no longer using
4. If currently using, it is their preferred brand or one of several brands they use
5. If no longer using it, they have either moved onto a different brand or they reject using the brand
6. If aware and not tried, they either consider using in the future, reject the idea of ever trying the brand or are aware of the brand but have no feelings either way.

The most important relationship is Loyalty, but most consumers are not loyal to any brand.  Most buy from a selection of acceptable brands.  In this example most of the users of the leading brand Alpha are Repertoire users of the brand (i.e., it is one of several brands that they use), and 38% say it is their preferred brand.

Nearly 2/3rd of the people that use Beta and half the people that use Gamma also use Alpha, some of these use it as there preferred brand.

We are measuring attitudinal rather than behavioral loyalty.  Attitudinal Loyal does not mean solus usage, it means that the brand is the individual’s preferred brand and that it is used most frequently.  During most shopping trips (online or physical) a consumer is most likely to instinctively reach for their preferred brand (triggered by the brand’s distinctive assets) and not notice in-store promotions.

Only half of all consumers feel ‘loyal’ to any brand. The other half rotate around a set of acceptable brands driven by availability at the point of purchase, promotions, or a desire for variety.

Lapsed users have tried the brand in the past and but it has currently fallen out of their set of actively used brands.  Frequently, these people can be triggered to reengage with the brand and add it back into their set of currently used brands.  Maybe the brand is a “birthday cake” with a very long interpurchase cycle, that used needed the right occasion for the brand to be reconsidered.

Rejectors are different, these people have tried the brand in the past and now say that they do not consider using it again.  Maybe they had a bad brand experience that was not resolved successfully.

The Attracted are people that have never bought the brand but say that they would consider using it.  The barriers to purchase are normally availability where they shop/search, price (the brand is usually more but sometimes less than they usually pay in the category), and the potential risk involved in trying a new brand.  Brands can ensure that they are noticeable with good packaging design, shelf positioning and signage.  Brands should examine their price relative to the category to ensure that it projects the right image.  Risk can be mitigated with trial sizes or free samples.

Hostiles are people who reject the idea of the brand and will not even try.  Sometimes these people are truly wedded to another brand.

Finally, there are the Indifferent. These people are aware of the brand but neither reject nor consider using it.

The most important brand relationship metric is saliency i.e., Top of Mind Awareness (the first brand that comes to mind when you think about the category or the consumption moment).  Awareness is necessary and important, but Top of Mind is golden.  It reflects the strength of brand relationships that have developed over time making TOM a stronger predictor of sales than Aided Awareness (average R² = 0.70 vs. 0.44).

We see a strong feedback loop between purchase behavior and brand relationships.  The most recent brand experience influences Top of Mind awareness.  Repeated purchasing leads to people adopting the brand as their preferred brand in the category.  Top of mind is also influenced by packaging and in-store displays. Word of mouth (old fashioned recommendations, social media comments and reviews) also boost TOM and generally amplifies brand presence.  These comments predominately come from three groups: the Loyal, Rejectors and to a lesser extent, the Hostile.

Simply improving brand saliency can often lead to an increase in share.  We see this in the correlation between TOM awareness and sales, it is usually the second most accurate predictor of sales after brand preference.

4.  Creative

Creative is the fourth component of the MSW marketing model.  MSW believes in the power of creative to move brands, we have seen this in study after study.

Creative Messaging influences both brand relationships and brand perceptions.

1. Relationships: brand communications (i.e., advertising, product and company information, user service information, sponsorships, events, packaging, instore activity) boosts top of mind awareness, strengthens loyalty and reminds lapsed users of the brands presence (it can also remind them of why they used the brand). Great creative is important but without the appropriate share of voice it won’t achieve the desired results.  Brands that invest in Excessive SOV tend to grow.  Strong creative combined with ESOV achieves real growth.
2. Perceptions: advertising is very rarely working from a tabula rasa. People have preconceived ideas about the brand and/or the category and the creative message needs to work within their framework.  Creative messaging normally works to strengthen pre-existing ideas or to nudge current perceptions.

Rational product and pricing messages tend to lead to short-term sales spikes, without any real impact on brand perceptions.

It is the emotional connections rather than the functional attributes that shape long term brand equity that explains most sales.  Consequently, our communication research activity is based around measuring five goals:

1. Attention
2. Branding
3. Communication
4. Short term sales lift
5. Long term brand equity

5.  Market Gap

Sales can be disaggregated into two parts: the contribution of long-term brand equity and the short-term marketing gap (the final element of the MSW’s predictive growth model).

Most of the time people buy their preferred brand; however, there are situations where people buy a different brand and analysis shows that there are five reasons for this difference

1. Price – Their preferred brand is too expensive
2. Promotion – The brand they bought had a special promotional price
3. Brand Availability – They did not see their preferred brand
4. Product Availability – They wanted a specific product or size, and it wasn’t available.
5. Influencer – The shopper was influenced by another person

(Child, Other family, Friend, Celebrity, Recognized expert, etc.)

To provide guidance to the shopper marketing and promotional teams, it is essential to understand:

• Which retailers are generating an adverse marketing gap and why?

• Which brands are you gaining from, and which are you losing to?
• Are these gaps persistent or temporary?

Conclusion

The Predictive Brand Growth Model influences everything that we do as Insights providers.

We utilize common metrics from the Model along with an analysis system that enable us to provide an integrated marketing and promotions information system that link all marketing activities to financial outcomes.