How To Lie With Graphics: Different Graphics from the Same Data to tell Smartphone Migration Story

I'm nearly done with the 2016 edition to my mobile handset industry volume that comes out every two years, the TomiAhonen Phone Book. And I was doing an update to one of the graphics and played around with some different ways to tell the same story. Please note, that all these graphics related to the same data set, I am not talking about current market share sales (that is the easy number you can get most anywhere). This is the far harder number to find, the actual installed base of phones in use, accounting for the different ages of phones still being used, and in many countries the strong resale market where old Nokias, iPhones and Galaxies go to find a second or third pocket to live in. This is the type of data my Phone Book is full of, data you always wanted but can't find in most public sources. All these graphics on this blog posting related to one data set, and all of it is not current sales of smartphones vs dumbphones ('featurephones') but rather, it is the full global installed base of handsets, smart or dumb.

In absolute numbers, I project the year ends with 5.8 Billion handsets in use, 2.95B of those will be smartphones, 2,85B will be dumbphones. As we know, in the third quarter of this year, we passed the 50/50 point in the migration of all phones in use on the planet being smartphones (in new sales smartphones have outsold dumbphones now for a few years already). So I will be showing a graphic in the TomiAhonen Phone Book 2016 that tells this story. And like many of the graphs in the two statistical volumes that I publish, the annual Almanac and every-two-years-out Phone Book, the story is best told in a time-series so you see the trends that are currently going on. I typically have about 10 year or 12 year periods I tend to show. So lets do a bit of story-telling or 'lying with graphics' to illustrate what will be learned, if we pick one method or another in telling the story. So why trend-line? Why not just pie graph? I could show the statistic like this:



Source: TomiAhonen Phone Book 2016 (adapted graphic)
This graphic may be freely shared


That is honestly true on the numbers I report. This is the split of the world's mobile handset installed base, by type. 51% are smartphones, 49% are dumbphones at December 2016. And there is definitely value in that info but it does not show the time-series. Instead of pie graph, now lets do time series. This is how the same data has shown in the TomiAhonen Almanac series for a decade (note the colors are different) but those who have read the Almanac know this graphic well.



Source: TomiAhonen Almanac 2016 (updated graphic, TomiAhonen Consulting December 2016)
This graphic may be freely shared

And now we can see the changes over time, as well as the current moment. Let me convert that graphic to the same colors I use in the Phone Book series (in the Almanac we have many more areas to cover, so the the total handset side only gets one chapter in the Almanac, and back in the previous decade when I launched the Almanac, the smartphone business was a quiet corner of the mobile industry haha. I have kept the colors then consistent over time so my readers can see the familiar items with familiar colors when they get the newer edition of the Almanac). So this is the same graphic for this blog now, as the one used in the Almanac for the past years, but converted to the colors I use for dumbphone vs smartphone race, in the TomiAhonen Phone Book series. Light blue are dumbphones, bright red are smartphones.



Source: TomiAhonen Almanac 2016 and TomiAhonen Phone Book 2016 (updated graphic, TomiAhonen Consulting December 2016)
This graphic may be freely shared

This is a fair representation of what has happened and this is how I have illustrated the race for years in the Almanac (and also in the Phone Book in the past). There is a shortcoming to this illustration and a misleading side to it. The shortcoming is, that we do not see the changes in the overall market. It seems like the market for phones overall has been stable. The misleading part is the optical illusion. Because most of the graphic has the blue part far larger than the red part, our brain is deceived into thinking even at the right end the blue part is larger. We have to 'think' to get to notice, that actually, red and blue are the same by the end. This version of showing the race, does not draw attention to how much significant the red part is currently, rather it draws our attention to how big the blue part was in the past. Let me convert that same picture to one that is not 'market share' balanced to even 100% every year, but rather, a graphic of absolute numbers that allows the growth (and/or decline) in the numbers annually. Now the same numbers over the same period of time give us this picture:

Source: TomiAhonen Phone Book 2016 (alternate graphic, TomiAhonen Consulting December 2016)
This graphic may be freely shared

This is a far more revealing picture to tell us about the whole handset industry, and the scale of smartphones in it, how long the fight has taken place and at what pace the smartphone has taken share from the dumbphone. The loss vs the previous graphic is that we now do not see the percentage annually, of the shift from dumbphones to smartphones. But now the optical illusion is diminished, and the 'threat' of the red part is far more pronounced. What this graphic does well, is illustrate how the growth of smartphones has done over time but it does not draw attention to how much dumbphone sales have fallen. An alternate positioning of the same data, tells us that. This switches the positions, putting dumbphones on the bottom and smartphones on the top. Same data again of course:



Source: TomiAhonen Phone Book 2016 (alternate graphic, TomiAhonen Consulting December 2016)
This graphic may be freely shared

See how now we are clearly drawn to the 'alarming' peak and fall of dumbphone sales that was quite 'invisible' in the previous graphs! Its the same data series. If you wanted to talk to a dumbphone maker (or its suppliers or retailers or accessories makers etc) this is the way to scare them, using the same data. Its not just that you are in decline, you lost the race (already). That is clear from this data. It was not obvious in the above data while the trend was yes, clear before.

All of those are valid ways to illustrate the race. I think there are two better ways, depending on perhaps what is the message you would like to communicate. If you want just a fair analysis 'unbiased' type of view, of the time series and the overall market, if we use staggered pillar graphs with absolute numbers (vs percentages), we can eliminate most of the optical illusion of the early problem and get to a relatively fair representation in one graphic. It includes the story of the growth of the overall size of the handset installed base. I like this view myself and will include this in the TomiAhonen Phone Book 2016 edition:




Source: TomiAhonen Phone Book 2016
This graphic may be freely shared

The beauty of this type of graphic for this particular data set, is that it lets us focus on the last pillar set on the right, and we get more accurately the immediate observation, that roughly speaking blue and red are about the same size today. But it also shows the overall handset installed base growth over time, and the rate by which smartphones now are taking over. But there is another way to look at the same data, as not the total market but as a cannibalization of one phone type taking over from the older phone type. So again, its the same data, but here we can put the pillar graphs next to each other, by year. Now we see the decline of dumbphones from its peak installed base, and the growth of smartphones but we lose the overall handset market size data, that we had in the previous graphic.  



Source: TomiAhonen Phone Book 2016
This graphic may be freely shared


If this was say VCR vs DVD player graphic it would probably be the better illustration for the race; but any smartphone has the utility of the full dumbphone (a DVD player cannot play our old Disney VHS tapes for the latest child). So this chart does give a misleading conclusion on the total size, where a reader could be thinking there were a TOTAL of 3.8 Blilion handsets in use at its peak (when they were dumbphones) not instantly understanding that these two sets should be added together and the total market is 5.8B today, and still growing. It is a more dramatic slide but it loses a key point of the total market growth, which is why I would not show this slide alone, without the previous slide shown first. As to the Phone Book, I will be including these two last graphs together in the new TomiAhonen Phone Book 2016 edition, to illustrate the migration of dumbphones to smartphones. I will be placing these two graphics right next to each other so the reader can compare the two and of course will be making the point that it is the same data set, telling us the same story but focusing on different aspects of it.

With that rare adventure into graphics on this mostly text-based CDB blog, I now return to working feverishly to finish the 2016 edition of the TomiAhonen Phone Book 2016 so we can get it out soon. If you want to order yours now, you know I do this special deal, where if you buy the current edition (2014 edition) now in days before the new edition comes out, you will get both. Its a ridiculously cheap resource costing only 10 Euros and having every number you ever hoped for about the industry, from screen sizes to camera resolutions to tech specs from NFC capability to WiFi, to installed bases of operating systems to regional splits of market leader brands etc etc etc. To see more about the current edition, go here to TomiAhonen Phone Book information and ordering page. 

Comments

[Wayne Borean]

Yep. Sure makes a difference how you lay things out. I've seen this sort of problem many times, where the visuals can mislead the viewer, despite being accurate.

[Santiago]

Great post, very explanatory to the ones learning.

[Marcel Pol]

Not knowing how thick the book will become, but you could include a short version of this blog post into the chapter. I really enjoy this write-up because it makes you think twice about the stats.

[Cycnus]

HMD Global announced the first Nokia-branded phones:
http://hmd.global/press/2016-12-13-nokia-150/

[E.Casais]

The new Nokia 150 runs S30+ -- which apparently is an OS from MediaTek, not a descendent of the Nokia S30 software. S30+ has been the OS for Nokia-branded feature phones for the past 2-3 years already.

Some time earlier, I raised here the question of how much of the former Nokia system software development is taken over into the HMD/Foxconn venture. The answer is clear: nothing remains from it.

Has anybody ever used one of those S30+ devices? How do they compare with previous S30/S40 mobile phones?

[John A]

The statistics show that the dumphone market shrinking so I suppose its no point to put much resources to the new Nokia low budget devices. I guess HMD Global will keep them alive some years to compete with Samsung that got similar devices. The more important part will be the Nokia android smartphones that will come in a near future.

[chithanh]

The dumbphones are important even if they make no profit. A dumbphone user will likely at some point buy a smartphone. That's millions of future smartphone customers right there, and historically Nokia has been very good at migrating their dumbphone users to smartphone users.

[Thomas Wright]

I have to second: Marcel Pol, that including this blog post in your book, might be an different and interesting way to both tell the story of that particular statistic, and to demonstrate another of your particular skills in the industry. Your love of the business shows in the clearly evident voluminous quantity and quality of time you spend, not only thinking about your industry, but also trying to find more effective ways to pass on to others your valuable insights about it.

[Just realized this may possibly be my first or possibly second post in ~ 4-5 years of carefully and thoroughly following your blog]

[Timmy T]

Well done. This is how all graphics should be presented.

[johnleslie]

Nice answers in replace of the question with real point of view and explaining about that.
telephone code 0116