WARNING: Do Not Engage with Engagement

Decades of data analysis have revealed that purchase activity is fundamentally the same in all types of markets, categories and geographies.

So Harley-Davidson customers are just like Coca-Cola customers are just like your customers. A whole bunch purchase just a little of what you and everyone else is selling, and very few purchase a lot of what anyone is selling. That is, there are consistent and predictable patterns in buyer behavior, revealing hitherto unknown Laws of Marketing. These revelations are truly revolutionary – or perhaps rather evolutionary – in the world of marketing.

These patterns are represented by the NBD-Dirichlet model as developed by Andrew Ehrenberg, which is best understood by looking at a histogram of purchase activity. For example, the chart below represents the distribution of cigarette purchases in America in 2002. Note that the category and the time don’t much matter because the point is precisely this: to all intents and purposes this applies across all categories at all times and in all places.


The essential pattern is the large volume of light buyers to the left, with the distribution tailing off quickly into a long tail of increasingly heavy buyers. It has profound and diverse implications for marketing and media strategy (well beyond the scope of this article).

What’s interesting to me is how the Dirichlet applies to your brand’s social properties: it applies in reverse. If you look at a brand’s Facebook fanbase, for example, you will see a lot of people who are heavy users of the brand, and a relatively small number of light users of the brand2. Given heavy users of a brand are less valuable advertising targets (because their upside is limited, and they are more attuned to promotions & discounts) and the fact that your social media posts only reach a declining fraction of the fans of any given page, we can quickly see it’s a not-very-useful platform for growing your business.

The same principle may apply to ‘engagement,’ i.e. it may follow a reverse-NBD distribution. The hypothesis is: people who engage with a brand (view a video, share brand content, etc.) are more likely to be heavier users of the brand. They are thus less likely to be meaningfully influenced (i.e. to purchase more) by engaging.

Even if this hypothesis is false, it’s arguably a moot point anyway: only a tiny fraction of customers or potential customers are likely to ever ‘engage’ with the brand3, limiting the efficacy of ‘engagement’ as a tactic in the first place. So, as with owned social, “engagement” efforts are often a complete waste of time.

When it comes to earned social, things seem like they might be different. “Sure, I want free media!”, you say, “For, there must be tons of value to be had!” Alas, the truth is that meaningful earned media driven by social sharing is the vast exception to the rule4. (So the next time your creative agency starts talking about dropping a ‘culture bomb’ into the social-sphere, nod and smile, and start tiptoeing backwards out of the room).

The truth about social media is that it works well if you pay for it. Sorry everybody: that just makes it ‘media.’

The industry’s fascination with engagement and social are regrettable but familiar examples of marketers:

a) failing to maintain the right focus (i.e. selling product, acquiring customers, driving traffic)


b) jumping into fads and trends without asking common-sense questions first.

Underpinning it all is the fundamental crime of not paying attention to the data. One key thing the data is saying is that you are better served acquiring new customers all the time rather than chasing nebulous concepts like engagement. Let Oxford Road help you shout from the rooftops just how incredible you are, because not nearly enough people know about you. I can promise you that.



1 https://www.marketingscience.info/

2 https://bit.ly/2LGtJho

3 http://adage.com/article/digital/study-1-facebook-fans-engage-brands/232351/

4 https://www.bookdepository.com/Viral-Marketing-Karen-Nelson-Field/9780195527988


Leave a Reply

Your email address will not be published. Required fields are marked *