US Brands lost half their customers last year – more misleading metrics

Yesterday, Ad Age reported a study showing that US packaged goods brands typically lost more than half of their loyal users last year.  Oh no!  The sky is falling…next year we’ll have no loyal customers left at all!

“I think what’s surprising is the magnitude of some of the effects,” said Eric Anderson, associate professor of marketing at the Kellogg School of Management at Northwestern, who reviewed the study.

Hmm yes surprising.  Let’s put our brains into gear here, are we to accept that all these brands, which are essentially stable in market-share, lost half of their most loyal customers? There may be a recession on but this is still nonsense.

The truth is that the analysts misunderstood their own results, because of ignorance of the law-like patterns in brand buying.

The brands haven’t lost most of their loyal customers, the results are simply due to normal random fluctuations in buying, i.e. sampling (in time) variation – something any analyst should be aware of.  Nothing real or unusual is going on here.

Read on if you’d like to know why…

 

The marketing consultants who did the study used their loyalty program ‘panel’ data.  They classified a consumer as a ‘brand loyalist’ if the brand represented 70% or more of their 2007 repertoire.  If that consumer did not also devote 70%+ of their category buying to that brand in 2008 they  classed them as lost (typically about one third were ‘lost’ completely, while the other 20% still bought the brand but it wasn’t 70% of their repertoire in 2008).

But from one time period to another the brand’s weight in a consumer’s repertoire fluctuates.  And this normal fluctuation is what this study mistook for customer defection.  These loyals aren’t gone, they’ll be back again next year or the next.

Effectively their analysis excluded most heavy category buyers because these households will have larger repertoires, and so it’s much more difficult for one brand to represent 70% of their buying.  Most buyers are light category buyers and these light buyers are more likely to appear 70%+ loyal.  In other words their analysis largely is a report on lots of buyers who bought the brand once out of 1 category purchase, or twice out of 2, or three out of 4 – purchases in the loyalty program stores.

Now, all buyers are subject to random fluctuations in their on-going, steady, purchase patterns.  Sometimes you buy 3 times a year, sometimes 4.  Even if you buy two brands equally it’s seldom ABAB, it’s patterns like ABBABBBAABABAAB.  This stochastic variation is normal and follows predictable patterns.  This variation means that lots of people who were classed as “loyals” in 2007 fall out of this classification in 2008 – when nothing real has changed in their buying behaviour, and nothing has happened to the brand’s market share.

 

PS The study was by Catalina Marketing and the CMO Council.  The CMO Council should have known better.  Catalina Marketing sell targeted marketing services based on using this loyalty program data – which is a bit odd because this fluctuation seriously undermines the capacity to target consumers based on their loyalty level.

PPS I’ll leave the last word on the Ad Age article to Professor Gerald Goodhardt (co-discoverer of the Dirichlet model):

“After ‘Some brands lost more than a third…… while others held on to more than 60%……’ I stopped reading!”

Advertisements

2 thoughts on “US Brands lost half their customers last year – more misleading metrics

  1. Pingback: US brands are not losing their loyal customers – even more misleading metrics « Marketing Science

Please comment on this article if you wish

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s