Loyalty and Satisfaction

Humans are one of the few species known to be both completely erratic in some behaviours yet predictably habitual in others. This is why customer satisfaction is so difficult to measure. We buy some categories predictably, but we buy others whenever and wherever we feel like it. So how can we identify patterns?

One of the most widely used metrics of customer satisfaction that is still kicking today is the Net Promoter Score (NPS), which claims to measure loyalty by asking a series of survey questions on the likelihood of recommending a product or service to others. And apparently, it’s king shit.

The NPS uses a 10-point Likert scale to assess future recommendations, with anything above a '9' indicating unwavering loyalty and anything below a '6' indicating indifference. Researchers are then encouraged to ask follow up questions to discover why, sparking a discussion in the process. This is the meat of the NPS, where the real insights lie.

This type of methodology would be fine if it weren’t for three basic truths:

1. You can’t quantify a discussion and pass it off as measurable data.
2. Humans are notoriously poor at predicting future behaviour.
3. The NPS doesn’t even measure what it’s supposed to.

Even on an intuitive level it seems to be inherently biased. Many of the results would reflect whether or not the category itself is a loyal one. The NPS may work for car brands, sure, but what about something as mundane and saturated as a washing powder brand?

“Would you recommend X brand of washing powder to friends or colleagues?”

Well I’ve used it, it’s done what it’s supposed to, and I’ve heard of it before, so yeah, why not? And there you have it; from those results it would appear that people are loyal to one specific type of washing powder. Only they aren’t.

There is also cause for a least a little suspicion when the creator of the NPS himself admits that the scale in no way measures customer satisfaction, which one would assume is very strongly correlated with loyalty.

Often, customers won’t purchase a brand again despite being satisfied, but nobody rebuys a brand they find unsatisfactory. The fact is, people switch. If Brand A satisfies, but Brand B is cheaper and presumably also satisfies (a presumption based on how satisfying we perceive the category to be), then Brand B will always come out on top.

Evidence has supported the idea of a duplication of purchases law, which states that a brand’s customer base is shared amongst other brands of the same category. This, according to Byron Sharp, is a natural monopoly.

Brands with more market share attract more light buyers, and light buyers will over time default to the leading brand.

So, both loyalists and light buyers are more likely to go for leading brands, which means smaller brands get less of both. This is the foundation of double jeopardy, another marketing law supported by research from Ehrenberg-Bass.

With regard to customer satisfaction, these laws throw a spanner in the works of metrics such as the NPS. If the frequency of loyalists depends on market share and popularity, then it follows that customers will be more likely to recommend bigger, more widely known brands over smaller ones, simply because they are bigger and more widely known.

They more easily come to mind.

And the final issue is that in order for someone to recommend a brand, they presumably have to have used it in the first place, which means any responses will be conflated with a fair degree of ownership bias.

If we use it, if it works, and if we’ve heard of it, we are likely to recommend it. These factors take precedence over our perceived satisfaction.

This is not to say that all customer satisfaction metrics are a waste of time, but that survey-structured metrics such as the NPS will only ever measure what people think they are going to do, and not what they will actually do.

Hard sales data such as frequency of purchase, frequency of use, basket sizes, and time spent browsing are much more telling. Thinking is erratic. Behaviour is predictable and habitual.

a.ce