What's going on with small brands?

Like, why don't they seem to follow the "laws of brands"?

There's this lovely theory of shoppers, brands & markets called the NBD-Dirichlet. (I know, I know: lovely theory, terrible name.) It's simple & elegant & predicts a whole bunch about how markets are structured.

But it doesn't do so hot with small brands. 

Take the sparkling wine market in Italy, f'rinstance. Francesca Bassi of the University of Padova analyzed 2 years of sparkling wine purchase data for 5,100 households: market share, household penetration, purchase frequency, and share of requirements (aka share of wallet).

This chart shows how off the model was from the data. It did a better job with the larger brands. But the smaller the brands got, the more the model was off (in both directions).

So does this mean the model doesn't work? No. But it does mean at least one of 3 things:

1. There's lots of "small sample error": if the sample sizes were bigger on these little brands, they'd match the model more accurately.

2. One of the assumptions of the model doesn't apply for this market or these brands. Maybe there are subsegments, or maybe the market is volatile and not stationary. Or maybe people's preferences are changing. 

3. Something else is going on for small brands. Like, gravity matters a lot for most things on Earth — until they're really light & wispy & float, and then wind matters a ton. So maybe other forces are at play.