Does the iPhone defy the Double Jeopardy law?
In a recent debate I had with Mark Ritson he made one empirical claim: that the iPhone was an outlier from Double Jeopardy — a smaller brand than Samsung, but with higher loyalty. Mark referred to the Philosophy of Science he did during his PhD where he learnt that scientific laws only need a single deviation to falsify them: “When you play the science game, once you disprove it once, it’s over”. It was a good debating point, good fun, but unfortunately wrong. This mistake is known as naive falsificationism.
Scientific laws, like Double Jeopardy, are patterns that have been carefully documented over a wide range of conditions. This means we have much knowledge about the conditions that don’t affect the law, and ideally also we know about conditions that do affect the law (boundary conditions). For example, if you heat water to 100 degrees celsius it boils. It’s a law-like pattern — surprisingly neither the size nor the shape of the container matters, nor does the heating source (sun, fire, lasers, or nuclear energy), but oddly enough altitude does — which helps us understand why water boils (it also matters for those few who might need to make a cup of tea on Mount Everest, not so much for the rest of us).
Knowing about a factor that affects the law, i.e. a boundary condition where the law is bent, gives us even greater powers of prediction and a deeper understanding of why the law exists and what it means.
So, observing an outlier from an expected pattern seldom falsifies the law but instead raises some interesting questions. Perhaps our observation is wrong, perhaps we made an error in our measurement, or (much more interestingly) perhaps we have discovered a boundary condition?
The powerful thing about a scientific law is that it allows us to spot oddities. Laws also help us interpret them, even small deviations. Rather than just shrugging our shoulders and saying “well maybe the iPhone’s a bit different”. Or falling for popular fallacies e.g. Apple has magical extraordinary loyalty.
So, Mark’s broader point that one instance of an observation refutes a scientific law was wrong, but what of his specific claim that Apple’s loyalty is a deviation from the Double Jeopardy law?
Mark based his conclusion having read that the iPhone has a mere 13% market share and yet a 92% retention rate. Now, Double Jeopardy says that smaller brands should have slightly lower loyalty, so this sounds like a potential deviation, which would be really interesting if it were true, but it’s not. I analysed this sort of mistaken claim a few years back. But let’s look again with today’s data…
The 92% retention figure comes from a recent Morgan Stanley survey. It’s probably in the ballpark, but it’s important to note that it’s not real sales data, it’s of intentions, not behaviour. And very important to note that it’s only of US consumers. The iPhone’s retention rate varies around the world. The Economist reports 80% iPhone retention rate in the US but only 50% in China (Apple has much lower market share in China, so lower retention would be in line with what the law of Double Jeopardy would predict).
In the US, Apple and Samsung have very similar market share, about 35% each (yes, a staggering 70% combined share), and both enjoy very high retention rates – in spite of Samsung’s exploding batteries. Loyalty is a fact of every market and very apparent here). But wait, this is share of recent sales, which is not the correct metric to compare with retention. What matters is each brand’s “installed base” i.e. how many smartphone users have an iPhone. Comscore data shows that Apple is by far the most-used smartphone brand in the US currently, with 45% share compared to 29% for Samsung. The Double Jeopardy law says Apple should have higher loyalty than Samsung, as the data in the following table makes clear.
US smartphone market follows the Double Jeopardy law — smaller brands have lower retention rates
By analysing like-for-like data from a single market, the US, and sorting the table data meaningfully based on market share, a clear, familiar pattern emerges. Smaller brands, those with fewer customers, also have lower repurchase rates; they are penalised twice. Exactly as the law of Double Jeopardy would predict.
So, in fact, after looking at the evidence, the Double Jeopardy law remains intact.