Estimating Apple and Android US market share

Apple knows how many iPhones have been sold in the USA (and disclosed it in the Samsung patent lawsuit) and probably also knows how many are in active use, but isn’t saying. Google knows how many Android phones are in active use in the USA but also isn’t saying.

Equally, the operators collectively know the figure for each of their networks, but again, they’re not saying (with a few exceptions: for example T-Mobile recently stated that it has around 1m iPhones on its network). 

So how can we work it out? There are two possibilities

  • You can do a consumer survey, perhaps combined with tracking retail sales through POS records (though I believe this latter will not capture Apple retail or Amazon sales)
  • You can make an estimate, based on the fact that the operators report how many iPhones they sell and they also report their total smartphone base, and that Apple has also reported US iPhone unit sales

This doesn’t tell us Android share directly, but knowing iPhone share tells us the maximum that Android can have by inference. 

Operator data

A good starting point is the assumption that since operators sell iPhones almost exclusively on 24m contracts in the USA, any iPhone they sold in the last 24m should still be in use. Therefore, very roughly, the last 24m of iPhone sales in the USA is a pretty good proxy for the active base. This data is disclosed, so in theory all we have to do is add it up… 

However, there are four problems with this: 

  • It excludes the ‘grey market’ of phones bought for export, but I would expect that to centre on sales direct from Apple, not phones bought under contract from operators. 
  • It also excludes people buying a phone under contact and then buying a newer model from the operator within 24m at full price - but this is a small market and anyway centres on direct purchase from Apple (see below) 
  • It excludes phones older than 24m that are still in use (a total of 20.5m units were sold in the USA longer ago than 24m)
  • It excludes the 1m iPhones that T-Mobile reported are now on its network. 

If we compare the US operators’ figures for phones sold on contract with Apple’s figure for phones sold in the USA, there’s not an enormous gap: 3.8m more Apple sales than operator activations in the last 24m. Part of this is inventory and repair replacements, part of it is grey market, part of it is T-Mobile and part of it is replacements (the second bullet).

Hence, if we net all of these out, I don’t think that these caveats are enough to move the figures in the following calculation:

  • Trailing 24m operator iPhone sales in the USA, as reported by the AT&T, Verizon and Sprint, were 56.3m to June 2012
  • AT&T, Verizon, Sprint and T-Mobile reported 120.2m smartphone users in June 2012
  • We have to estimate smartphone users for the many smaller carriers with a further collective 10% of the US base. I have estimated 10m. 

This gives us 56.3/130.2, which is 43%. 

Depending on what you think of the various caveats I outlined above, the right figure for the iPhone base might be varied by 2-3m in either direction, giving a range of, say, 41%-45% of smartphones. 

(This number is unarguably rising, because iPhone distribution expanded to include both Verizon Wireless and Sprint in the course of 2011: they do not have a full 24m of sales to include in the base yet. )

Surveys and double-counting

At this point things get complicated. There are now 300m ‘human’ mobile accounts reported by the US operators (i.e excluding M2M, dongles etc). But there are only 250m adults in the USA: many people have multiple accounts, for various reasons. 

So how do those 56m or so iPhones relate to actual penetration? This is where surveys should shed some light, since they relate to actual humans without double-counting. 

Comscore estimates that in June 2012, there were 234m “Americans aged 13 and older” using mobile phones, of which 110m were using smartphones. This is coherent: a higher duplicate rate in lower-value phones, and hence 110m smartphone users excluding the double-counting (eg people with Blackberries and iPhones or Androids as well) versus the 130m the operators report. 

Comscore estimates that Apple had 32.4% of these 110m smartphone users. That would imply an estimate of 35.6m iPhone users. NPD estimates that iOS has 31% of the US smartphone base, which is pretty similar. 

And here we run into a real problem.

If there are only 36m iPhone users in the USA, then of 56m iPhones sold in the USA in the last 24m, 20m - 35% - are either not being used (at least, not there) or are duplicates (i.e. people have and are using two iPhones). 

Have a third of the iPhones sold in the USA in the last 24 months been replaced with newer models, and then either put in a drawer and forgotten or unlocked and exported? Or is there something wrong with the survey? I’m open to suggestions…

Benedict Evans