Determining The Average Apple Device Lifespan

In The Number I provided a model for estimating the number of active Apple devices at any point in time. The relationship between active devices and cumulative devices sold gives us a rule of thumb that says that 2 out of 3 devices ever sold are active.

I propose now that knowing active devices and cumulative devices sold allows one to determine the average device lifespan.

The following graph is a visual representation of how to obtain it:

Note that the second graph shows how the lifespan evolved over time.

Here’s how to compute this yourself: Visually, the lifespan is the distance horizontally between the two vertical bars such that the bars are the same length. The top vertical bar measures the gap between the area (cumulative devices) and the curve (active devices) and the lower bar is the gap between the area and the x-axis, i.e. the cumulative devices. When those two bars are the same size the distance between them is the lifespan (at the time of the top bar.)

Arithmetically, the average lifespan at a given time t is the duration between t and the moment when the cumulative devices sold reached the cumulative retired devices at time t.[1]

For example today–as the visual above represents–the lifespan is the time since cumulative devices sold reached the current total retired devices. The cumulative retired devices can be calculated as 2.05 billion cumulative sold minus 1.3 billion active or 750 million. The time when cumulative devices sold reached 750 million was the third quarter 2013. The lifespan is thus estimated at the time between now and Q3 2013 or 17 quarters or about 4 years and three months.

Note that cumulative devices sold includes Macs, iPhones, iPads, Apple Watches and iPod touch. Since (apart from the Watch[2] ) this information is public, the only figure needed to determine lifespan at any time is active devices and that total is available through the logistic fit described in The Number.

In The Number I argued that active devices is a breakthrough in quantifying the value of Apple’s business. The first insight is into the size of device (read: user)  base, the second is in the resilience of that device base (2/3 in use). Now we see the third insight: the specific length of time or duration of use per device–a proxy for user satisfaction and loyalty. This I (and Deming) argue, is the most important measure of the health of a business because it speaks of the future and not, as all other figures do, the past.

I’ll show in the next post how this single number allows us to calculate the net present value of Apple’s future cash flows, or, by definition, Apple’s enterprise value.

  1. For a simulation and tabular explanation see below []
  2. Which I calculate to a satisfactory precision here []
  • Duane Bemister

    Do you ever sleep? Great Work!

  • Craig Parietti

    Assuming your lifespan is correct, this is the most important measure of the hardware slice of Apple’s business value. For example, If customers previously purchased a new iPhone every 4 quarters and now do this every 16 quarters, that implies iPhones may be either used by one customer for 4 years or handed down in families or sold as used. Are you able to calculate iPhone lifespan separately?

    • James Steel

      I do not know if apple publishes the data, however if not looking at statistics for how many devices are still active with a given version of iOS could be a good estimator.

    • Sacto_Joe

      First, bear in mind that this is all Apple devices, including desktop computers that are known to have longer lifespans than, say, iPods. Countering that is the relative quantity of the devices; there are many more iPhones than any other device Apple makes, for example.

      Second, I worked for a year at an Apple “refurbishing” plant, and can personally affirm that there are active processes in place for extending the lifetime usefulness of Apple devices. That will make it even more difficult to work out the specific lifetime of any one device.

      But far more important than any of this is the clear picture of the value proposition for quality devices of any type. A cheaply made device will yield very little value via refurbishing. Not so with well made devices.

      I’ve calculated (in Horace’s previous article’s comments) that ~60% of iPhones ever sold are still in use, versus ~20% for non-iPhone smartphones. Why? Because most “smartphones” aren’t built to last. As a result, as the smartphone market tops out (which it seems to be doing), higher quality smartphones inevitably gain a proportionally larger share of the installed base. And every single iPhone that’s ever been built has been built to the highest quality Apple could manage. By my calculation, that gives iPhones somewhere between 25% and 35% of the installed base right now. More importantly, that Apple installed base percentage is growing.

      There’s not the slightest doubt in my mind that Steve Jobs was intimately aware of this phenomenon. He practiced it with the Mac long before the iPhone ever existed.

      That raises an intriguing question, btw: Just what is happening to the Mac/iPad installed base? My bet is that the same thing is happening there.

      And then there’s the Apple Watch….

      • Glaurung-Quena

        “First, bear in mind that this is all Apple devices, including desktop
        computers… there are many more iPhones than any other device Apple makes.”

        Total macs ever sold (since 1984) is 230 million. Total since 2009 is 152m. So figure they represent 10-15% of the total active devices.

      • melci

        Incorrect, there are only about 100m active Macs which only accounts for 7% of the 1.3 billion active apple devices worldwide.

        “Apple Senior Vice President of Worldwide Marketing Phil Schiller revealed that the company’s operating system is growing in popularity. In pure numbers, close to 100 million people are active Mac users” 5th of April 2017.

      • Sacto_Joe

        We can also say that 5% or less are non-iOS devices. And I think it’s a safe assumption that those non-Mac, non-iOS devices have no more average lifetime than the average iOS device. Finally, I think it’s fair to say that all iOS devices have about the same general lifetime. At that point, the ~7% Mac contingent has very little impact on overall Apple device average lifetime.

      • Glaurung-Quena

        Thanks, I’d forgotten that they revealed that figure. Interestingly, this means that while over 60% of IOS devices are still in use, for Macs the ratio is reversed: only about 40% of macs ever sold are reporting that they are in use (this stat could be flawed if there are a significant number of Macs which are not phoning home, either because phoning home has been disabled or because they’re running windows/linux).

      • Sacto_Joe

        “…only about 40% of macs ever sold are reporting that they are in use…”

        Only? Macs go back to 1984, my friend. Besides, you need to compare Apple and Apples. We can only go back 10 years for iPhones, as an example. Over the last ten years, 151 M macs have been sold. I can’t imagine more than a few million pre-2007 Macs are still in use.

        I’d say that’s pretty darned close to Horace’s 64% figure….

    • Clinton Weir

      Lifespan has nothing to do with upgrade cycle. If I buy the new iPhone every year and sell the old one just as frequently then that iPhone still has a multi-year lifespan.

      This analysis seems wrong to me somehow. At some point, Apple sells the first iPhone device. Then they sell thousands more the first day. #10001 is DOA. The time from the sale of the first iPhone to the time of the first DOA iPhone (less than 1 day) is the average lifespan of all iPhones? I think not.

      This suggests it’s an average of the lifespan of all iPhones that have already lived and “died”, meaning it’s asymptotic with the true value we’re interested in.

      It’s probably an increasingly useful measurement over time, but with the caveat that it averages together old models with really old models and says very little about the new models with which you’re most interested. A comparative approach would be useful to detect DOA problems.

      • Sacto_Joe

        I don’t think #10001 is “DOA”, as you put it. Recall that Apple is not so much selling a computer device as it is selling a computer ecosystem. Among other things, that includes “Services”, which is the fastest growing segment of Apple revenue at the present time.

        Also, as I said below, I can verify that Apple is intimately involved in the “refurbishing” business, although it’s profit from that business is off the top end, where it recycles essentially perfect “used” (often just unused old inventory, btw) products and sells them or uses them for trade-outs for flawed (a tiny number), returned, or Apple Care insured damaged products (I’ve seen a heck of a lot of brand new products that were “engraved” but were otherwise brand new, for example, that don’t pass the “perfect” test). So it’s not DOA there either.

        So yes, older Apple products still makes Apple money, and so they can’t be termed DOA. But even where they don’t make money directly, they increase the installed base. Eventually, every device dies. Those that have given their user good service, however, will tend to reward Apple long term with a replacement.

      • Clinton Weir

        I think you misunderstood my DOA comment. I was saying, if you use this analysis, then the first time you ship a product that is DOA (customer receives a non-working device) then you’re going to get a wild underestimate of the lifespan of the product line (assuming that’s the first device to go out of service; it’s also possible that the device was delivered functional but the customer dropped it in the ocean or something).

        More generally, the less time the product has been on the market, the more of the left tail you’ll be including in the calculation, and the more time the product has been on the market, the more of the right tail you’ll be including, but you never know where you are along the distribution. Perhaps 50% of the devices will last less than 10 years, and the rest will last forever.

        Incidentally, I have a drawer full of inactive iPhone devices. No amount of refurbishment counteracts customer’s decision not to use the device or pass it on.

      • This is not an estimate of product line lifespan. It’s an *actual* measurement of utilization. There are no estimates or assumptions or guesses required.

      • Clinton Weir

        Your response shows that you clearly are not a scientist and I have to dismiss everything you did here.

      • Sacto_Joe

        I think you need to seriously reconsider that statement.

      • Clinton Weir

        Referring to the current analysis (measuring the temporal distance from the present to the moment in the past when the current number of inactivate units matched the past total number of sold units):

        From my simulations, it seems to badly fail if the sales are not constant (ie if sales of the product are growing or shrinking).

        Imagine a constant sales scenario. For simplicity, we’ll say Apple sells 10 units at midnight on January 1st each year and sells no more units the rest of the year. We’ll also assume that each year, on December 31 at 11:59PM, 10% of active units suddenly fail. Some of the people whose phone break at 11:59PM are some of the people who buy a minute later.

        In this scenario, it’s fairly obvious that the average lifespan of all devices will be 10 years, and if we graph it out, the measured value converges to 10. At year 10, Apple sells they’re 100th unit, and 8 years later there are about 100 inactive units. In year 30, they’ve sold 300 units, and they don’t pass 300 inactive units until year 40. It works pretty well.

        Now, same scenario except with 10% annual growth. In year 10, Apple has sold 159 units, and they pass 159 inactive units in year 17 (just barely not getting there at year 16). At year 30, they have sold 1645 (cumulative) and they pass 1645 inactive in year 37 (7 years later). In year 140, they have sold (cumulative) 62.37 million and they pass 62.37 million inactive devices in year 147 (7 years later).

        Now, imagine shrinking sales. In year 1, apple sells 1.1^30 (about 1.62 million) and sales decline 10% per year. In year 10, they’ll have created 10.54 million and they’ll pass 10.54 million inactive at year 22 (just barely missing it at year 21). They’ll reach 16.09 million sales in year 50, and won’t reach 16.09 million inactive units until year 70.

        (All of this math is thanks in large part to spreadsheet software).

        So, he claims he is making no assumptions (ie he is not a scientist), but his method only works if you assume constant sales, which we KNOW is a bad assumption. Perhaps if he thought about the assumptions he was making, he’d be able to disclaim this methodological error.

      • Sacto_Joe

        From the first graph (which I have no reason to doubt), we clearly see that, for at least the last 6-7 years, growth in devices sold has been close to linear. As you point out, this is not “constant sales”, as you define the term. Note that the percentage of the percentage of active to cumulative sold presently equals ~60%.

        We can likewise draw a roughly linear line through devices active, again, going back about 6-7 years. If we move both of our dark lines back two years, we will see that the line representing what Horace calls “Lifespan” (an arguable term, granted), will decrease. Going back 2 years, active devices looks to be about 900 M out of about 1.5 B. That would only move the line on the left back 1 year or to about Q212. “Lifespan” thus “compresses” from 4.25 to about 3.25 years.

        But I think this is more of a visual cue for people than an actual measure of “lifespan”. At worst, it’s a less than useful label. The green area to the left of the left line shows us graphically the devices that are no longer in service, while the right of the left line shows us those that are in service. That’s it.

        More important is the percentage of active to cumulatively sold. And back 2 years, the percentage of active to cumulative sold would equal ~60%.

        What happens going forward if the active devices line and the cumulative devices sold line stay linear, as they have for the last 6-7 years? The rate of device growth appears to be about 300 M/year. In 2 years, the devices would theoretically grow to 2.65 B. The active devices would appear to increase (by eye) to about 1.5 B devices, following Horace’s chart. That, again, equals ~60%.

        It really doesn’t matter what you call this 60% constant. Horace called it “average device lifetime”. I’d call it the “active Apple device constant” But it’s real, and it’s there. Yes, it depends on a linear increase in devices sold (as others here have pointed out). But there seems to be a decent amount of historical support for the contention that, for Apple, there really has been, and thus can be presumed to continue to be, such a linear increase.

        (Is this linear increase something Apple has actively planned? I can’t see how they could, since sales depends on market forces. About the only thing I can think of that might be relevant is if Apple is implementing a linear increase in production capacity. Which would itself be pretty darned impressive! IMHO.)

      • obarthelemy

        If you start to use that number to predict the future (say, a net preset value), you do run into the same issue that life expectancy has though: today’s life expectancy of 80yrs tells you that people born 80 years ago could on average expect to live 80 years. Using it to assert people born today have a life expectancy of 80 yrs assumes the future and the past are identical. That doesn’t work for life expectancy because of obvious reasons (wars, epidemics, medical progress,…), and doesn’t work for phones either.
        Without guesses and assumptions, the only thing you can calculate is what the NPV of Apple was 4 years ago.

      • Sacto_Joe

        That’s actually a worthwhile point, obarth. We see Android, for example, producing higher quality smartphones now than when they began. That’s a big reason why so many iPhones, which were always produced at the highest quality Apple could manage, are still hanging around and being used, relative to the percentage of Androids. And I strongly suspect that’s why smartphone growth has appeared to “top out”; more and more smartphones of all stripes are able to be useful for longer periods of time.

        But at this exact period of time, we can definitely say that Apple’s strategy of high-quality from the get-go had hidden benefits that we can actually measure. It’s quantifiable.

        I would like to see Horace – or someone – make a similar attempt to quantify smartphones more completely, but there may not be enough data to do more than a guesstimate, like the one I came up with that showed about 21% of Androids are still in service.

      • El Aura

        When Apple replaces a non-working phone with a replacement, does this add to the number of iPhones ‘sold’ (shipped)?

      • Clinton Weir

        I’m not sure whether the replacement phone is counted, but Apple does refurbish the phones that come back to them, so the replacement phone will be counted at some point.

        Incidentally, if you return your phone under warranty they’ll usually swap it out for a refurbished unit. There are definitely exceptions to this, under the general category of “Apple being Awesome”, in which they just swap out your bad phone for a brand new one.

      • Sacto_Joe

        Fair point. I did actually misread it at first. That’s why I posted a second time. Oh, and I have a house-full of inactive Apple devices. I’m sure many of them are worth something to someone, but I just don’t have the time to mess with them.

        I do get your (real) point. I get the “tail” argument. I think of it more this way: Apple has been growing it’s device distribution per year for a long time now. In fact, it is remarkably linear, and has been so for years now (see Horace’s first graph). If at some point device distribution per year levels off (which presumes a lot, including no more new devices developed, which isn’t happening right now), then we would also expect to see installed base level off at some percentage. At that point, the left “line” that Horace has drawn would stabilize after a few years at some given year, and the distance between those two lines would just travel farther and farther apart, with no change in line length. Clearly, then, it would no longer be measuring “average lifetime”.

        But, as I said in my other post to you, I don’t see the 4.5 year “average lifespan” as anything real, except purely by accident. That is, there may be an Apple device with a 4.5 year “lifespan”, but that’s just coincidence.

        Horace is well aware of this, hence his second graph.

        The problem is, I believe, that you are being too literal, or that Horace is using a somewhat misplaced term, or a bit of both. What the first graph is doing is showing, literally graphically, how much of Apple’s total device production is no longer in use versus how much is. The number of devices to the left of the left line equals exactly the number of devices that are no longer in use, while the number of devices to the right of the left line represent the number of devices still in use.

        That’s it.

        Now, where this becomes useful, as I said in my other post to you, is in any “comparative” charts, whether they be charts of smartphones, or cars, or washing machines. Or smartphones, for that matter. Anything where there is sufficient data. The “average lifetime”, to use Horace’s term, of each product or series of products, could be comparable – PROVIDED, as you point out, a similar number of years was used in the comparative analysis, and PROVIDED a similar pattern of growth existed.

        One place all of those exist is in the overall Android smartphone market. Thus, as a means of measuring the difference between those two markets, this approach is quite satisfactory.

      • melci

        Clinton, Apple’s figure for 1.3 billion active Apple devices only includes “iPhone, iPad, iPod touch, Apple TV and Apple Watch devices that have been engaged with our services within the past 90 days”.

        As such, a drawer full of inactive iPhone devices is not counted.

      • Clinton Weir

        They are counted – as inactive.

      • Sacto_Joe

        The other half of your argument is interesting. Clearly, you’re right that #1 doesn’t “die” less than a day after it “was born”. But as a tool for visualizing average lifespan at this moment in time, the “two lines of equal length” approach works right now, by simple virtue of the facts. (2.05 B devices sold, 1.3 B still in service). We could make a comparative chart for computer devices in general, and it would also give us a graphic representation of the actual status.

        IOW, the 4.25 year “lifespan” isn’t in any sense real. But it may allow us to make meaningful comparisons nonetheless.

    • You cannot calculate iPhone lifespan separately without making guesses.

      • El Aura

        Hasn’t Apple at times released the total number of active iOS devices?

      • No.

      • Sacto_Joe

        True, but I think you can make a good stab at isolating iOS devices, and I think it’s a fair surmise to say that all iOS devices of a given year have about the same lifespan. Also, the truly long-lived Apple devices that aren’t iOS are Macs, and there just aren’t that many of them out there, relatively speaking.

        In short, you can make “bigger than a breadbox” comparisons (i.e., a range of possibilities), which is still worthwhile.

        Just my 2 cents….

      • El Aura

        Yes, for example, add the assumption that Macs are used on average at least for years (which I would be willing to bet a lot on), and you could extract the conclusion that iOS (+Watch & TV) devices are used on average less than, maybe 3.7 years (I didn’t do the math). Equally, you could say that Macs on average are used for less than eight years. With that you might get that iOS (+Watch & TV) devices are used on average for something between maybe 3.3 to 3.7 years.

        Horace has modelled Apple using assumptions (eg, on the number of iPod touch sales) before to get a decent estimate of in the form of single number. Adding the same level of assumptions to get a range as a result would be equally valid and I think also very illuminating.

  • Sacto_Joe

    Love the “two lines of equal length” approach! BTW, there’s a bit of an optical illusion going on here. The line on the right looks shorter to me than the line on the left. A ruler proves that’s not the case.

  • Gregg_Thurman

    Excellent work. I’ve been trying to develop a methodology for determine the same thing, obviously without good results.

    I wanted a metric like this for estimating future Services revenue, the logic being that with static revenue per active user a larger active base will result in higher Services revenue. The only fly in that ointment is Apple’s continued expansion into content creation/delivery, thereby increasing Services revenue per active user.

    I believe that Apple Music, AppleTV 4K (HDR) and HomePod play directly into Apple’s content creation strategy. Apple created series and movies will be streamed in 4K (HDR) with sound separated from video by Apple TV 4K and synced in real time with HomePod. In addition to great visual (4K HDR from anyone) Apple active users will get best of breed audio from Apple created content.

    When Apple starts shipping this completed strategy Services and Other Products revenue is going to explode.

    • obarthelemy

      I don’t see why you’d have to go through a lifespan estimation to extrapolate services revenue: services revenue is a function of the installed base, just estimate it directly from that. Adding the lifespan in the middle is pointless, especially since the lifespan itself is derived from the installed base info: it’s just adding complexity and imprecisions.

      • Kizedek

        You’d want to know “active” installed base, in order to estimate services revenue, wouldn’t you?

        The fact that it is becoming clear that the installed base of active iOS users is growing faster than that of Android users is significant.

        It helps demonstrate the success and value of Apple’s approach. And it makes it clear why iOS users are also more valuable to Google than its own Android users.

      • obarthelemy

        “active” is built into the definition of “installed”. “Inactive” is “uninstalled” ^^

        Where do you get that the active iOS base in growing quicker than the active Android base ? Browsing stats are flat (and they over-represent iOS).

        Indeed iOS users are more valuable: they hand out money freely are, it appears, for no strong reason.

      • Kizedek

        “Where do you get that the active iOS base is growing quicker than the active Android base ? Browsing stats are flat (and they over-represent iOS).”

        It’s been discussed throughout this thread — generally, each iPhone sold is staying in service twice as long as each Android. Two thirds of all iPhones sold are still in use, while a much smaller percentage of Android phones are still in use.

      • obarthelemy

        True, but since iPhones are only 15% of phones sold, it’s not obvious to me how you go from iPhones being used 2-3x as long to share of iPhones rising. Please elaborate.

      • Kizedek

        Sacto_Joe already elaborated below:

        “I’ve calculated (in Horace’s previous article’s comments) that ~60% of iPhones ever sold are still in use, versus ~20% for non-iPhone smartphones. Why? Because most “smartphones” aren’t built to last. As a result, as the smartphone market tops out (which it seems to be doing), higher quality smartphones inevitably gain a proportionally larger share of the installed base. And every single iPhone that’s ever been built has been built to the highest quality Apple could manage. By my calculation, that gives iPhones somewhere between 25% and 35% of the installed base right now. More importantly, that Apple installed base percentage is growing.”

      • obarthelemy

        Yet if you look at the web browsing stats there
        %iOS is shrinking.

      • Sacto_Joe

        Yeah, and if you look at the bias of the person who’s running Statista, you’d see it ain’t toward Apple….

        I wouldn’t trust Statista’s numbers when it comes to Apple any farther than I could throw Donald Trump – which may be farther than the average 72 year old, but still isn’t very far….

      • Kizedek

        “Indeed iOS users are more valuable: they hand out money freely are, it appears, for no strong reason.”

        This is another example of how you like to have something both ways. 🙂

        Elsewhere, you said:

        “…Worth it for a $700+ phone, not worth it for a $150 phone.”
        “So that’s not along “Android vs iOS” lines, but along “How much did I sink into the thing” lines.”

        So, the average iPhone is in use 4 years, cracked screen or not (such is the value to its owner), and the average Android is in use for 2 years and retired even if unbroken.

        According to you, the iPhone user finds it worth putting 32 bucks into a new screen, while the Android user will just replace his whole $150 phone. So, the iPhone user has another 120 bucks to spend for services, media and several useful apps (say video editing or business finance, at maybe 20 bucks a crack)…

        Yep, it sure sounds to me like iOS users hand out money for no strong reason, while Android users don’t 😉

      • obarthelemy

        Who said the average Android is retired after 2 years even if unbroken ? Apparently breakage is 10-20% per year, so roughly (not compounded) 30% after 2 years. Those are retired. That doesn’t leave much room for many unbroken phones to be retired after 2 years. I’m upgrading my 2yr-old Mi Max v1 when the v3 comes out this summer, but that’s purely because I’m capricious, the v1 certainly has 2 more years in it.

        As for costs, indeed Android users will on average pay $20*15% = $3 for broken screens. Apple’s windfall is that a lot of their users will pay $75 for aCare + the $30 deductible, or $100+ for out of warranty in-shop repair. For a $30 screen. The situation arose just recently with my nephew. A whole new, better-specced Android costs less than a screen repair on an old iClunker.

        Yougot the $120 wrong: the Apple user already spent $500 extra just getting the iPhone. He’s merely making a small slice of that back. Unless he bought Apple Care or paid Apple service prices.

      • melci

        The stats say Android devices are retired on average after significantly *less* than 2 years. Witness the fact that Google reports only 2 Billion active Android devices (phones + tablets) despite well over 1 billion Android phones and several hundred million Android tablets being sold every year.


Analyst Benedict Evans had this to say a few years ago: “It seems clear that Android phones remain in use for well below the 24m average for the market, and during the peak growth period the replacement rate was closer to one year.”

      • obarthelemy

        That was years ago, and that was Ben Evans.

        Statista says
        – smartphones shipments: 1.4b in 2015, 1.5b in 2017. Of which about 0.4×2=0.8 were in China (no Google) and 15%=0.45 were iOS; so net google-Android = 2.9-0.8-0.45=1.65

        – tablets shipments 0.2 and 0.2, let’s assume 50% google-Android so 0.2

        Total Google-Android sold over 2 years = 1.85.
        Total Google-Android devices in use early 2017 = 2bn

        How does that say devices are used for less than 2 yrs on average ?

      • klahanas

        As an Android (and Apple) user that’s a measure of independence and success. Diversification is good.

  • melci

    In comparison, Android devices look like having lifespans somewhere between 1-2 years. Witness the fact that Google reports only 2 Billion active Android devices (phones + tablets) despite well over 1 billion Android phones and several hundred million Android tablets being sold every year.

    AOSP Android devices only add another 30% or so to Android’s numbers meaning Apple’s iOS installed base at between 1.1 – 1.2 billion active devices is about 60% the size of Google’s Android and 50% of Google+AOSP Android.

    Benedict Evans had this to say a few years ago: “June 2013 there were 538m MAUs (+315m), and 900m cumulative activations by May. So, a bit over 500m activations in the year, and roughly a 1 year life span. It seems clear that Android phones remain in use for well below the 24m average for the market, and during the peak growth period the replacement rate was closer to one year.”

    Horace, I’d love it if you could revisit Android’s active installed base and lifespan for comparison?

    • obarthelemy

      I always feel that “the Android market” (ie the non-Apple market) doesn’t exist any more than the “non-BMW market” for cars: that’s putting very cheap Yugos and <$100 phones in the same basket as Lexuses and $1k Galaxy Ses, then because cheaper phones are overwhelmingly more represented, drawing an "Android vs iOS" conclusion that really should be a "cheap vs expensive" comparison.

      For example, fixing a phone costs about the same: an iPhone 8 screen is $32 on ebay, a Xiaomi Redmi Note 4 screen is $25, plus tools & bother, or expert service. Worth it for a $700+ phone, not worth it for a $150 phone. And probably worth it for a $700+ Samsung S (though OLED screens are a lot more expensive than LCDs).

      And even if you don't fix the screen, the incentive to keep using an expensive phone with a broken screen is much stronger than for a cheap phone.
      So that's not along "Android vs iOS" lines, but along "How much did I sink into the thing" lines. With breakage at 10-20% / yr, that's a very important consideration.

      • Kizedek

        It seems Macs (as Apple devices) are not significantly affecting the 4yr calculation of usability span of the iPhone — if Macs were, it would raise the span closer to 6 yrs, because many Macs last 10.

        Yes, as you note, the Android side of the mobile phone market already includes the whole gamut of Android phones, from the cheapest to the most expensive.

        Yet, the figures and analysis for active installed base of both platforms show a smaller percentage for Android than market share of sales would suggest.

        Therefore, the data seems to suggest some combination of:

        A) high-end Android smartphones are a very small part of the Android market share
        B) the usable life-span of a high-end Android smartphone is not significantly more than the 2yrs average given for all Android smartphones. The quality and updatability that would lengthen its usable life is just not up there with the most basic iPhone.

      • obarthelemy

        Nothing has to seem: we invented maths !
        – cumulative Mac sales over the last 10 yrs = 4x20m + 4x15m + 2x10m = 150m. (eyeballed from statista)
        – useful life (your figure) = 10yrs = 120 months.

        Since the useful aggregate life of 1.3bn aDevices is 48 months, (1.15 X + .15 x 120) / 1.3 = 48 solving for X, results in slightly less than 34 months.Not insignificant: once the Macs are taken out, the remaining iDevices have a useful life of 34 months = slightly less than 3 years.

        That’s a massive effect, and explains why I was complaining about aggregates not meaning much earlier in the topic.You’ve nicely illustrated both the numbers issue and the fandom issue about misapplying or misunderstanding numbers.

        Given the 3x ASP of iPhones (before service costs), a 1.5 longer useful life the $230-average Androids is rather lackluster. And probably close to Android flagships. But they do manage to sell more than one every minute ;-p

      • El Aura

        I think you overestimate average Mac and iPad lifetimes. Some Macs will be used for ten years, most will not. I think even seven years (as the average) is already pushing it. I know the average is not the same as the median, but if ten years were the average this would mean that half of the computers are ten years or older. And since OS support tends to last not longer than eight years, more than half would be running not the current (or even the previous) OS version. And that is not what the OS statistics that Apple releases show.

        You’d have to pick an average time span for how long Macs are supported by the latest OS (or latest + previous OS), but then you could use the OS statistics (what percentage of Macs runs the which OS version) to calculate what is the maximum percentage of Macs older than that time span (maximum because by far not all Macs run the latest OS they technically could).

      • obarthelemy

        You’re probably right, but it’s hard to not take the “Macs last FOREVER” claims in those very comments at face value just for the own-goal. It’s fairly easy to use the formula with other values.

      • melci

        Apple reported that there were close to 100m active Mac users worldwide a few months ago. Not 150m.

      • obarthelemy

        I wanted to see if I remembered how to do a sum on a range of variable length (I don’t), but did a dirty hack to parameterize the calculation anyway, just enter the estimated Mac and iPad lifespans:

        The spreadsheet isn’t protected, so make a copy ^^

      • El Aura

        I’ve just checked some OS statistics. Last September, just before the release of High Sierra, Sierra achieved an installed base of 60% (+2% for High Sierra betas).

      • gigo

        The iPad lifespan is currently set to longer than the iPad has existed in the market. Great work as usual. Have you ever heard of the acronym GIGO?

      • obarthelemy

        It shows 7, iPad was launched in 2010. 7 > 8 ?

        My 2011 Galaxy Tab is still running fine, I assumed iPad could do as much. My bad.

        Also, please note the “spreadsheet isn’t locked”. It show 7 right now, but someone might have put 10 in a while back.

        Great reading comprehension as usual.

      • gigo

        Who cares what your Galaxy Tab is doing or what an iPad could possibly do. This is about average lifespan, not maximum lifespan, genius. The idea that almost every single iPad ever produced is still in use is obviously ridiculous.

      • obarthelemy

        Every single tablet I ever bought or saw around me (that’s about 15) is still active except a Nook Color and a couple of Asuses (predestined name I guess) that got sat on. That includes an HP Touchpad. I think assuming tablets are kept in use until destroyed is probably the less crazy assumption, but please do redo the calcs with 1 year lifespan if you want to fake boost the iPhone’s lifespan.

        Apple says it’s 3 yrs, you say they’re lying. OK !

      • gigo

        What did I say that you somehow interpreted as me saying “Apple” is “lying” about “3 years”? Are you okay?

      • obarthelemy

        It’s OK, you probably didn’t realize what you were saying. I’ll spell it out really slowly for you.

        Aggregate lifetime is 4 years. Apple says iPhone lifespan is 3 years. The only lifespans for other devices that yield iPhone = 3yrs are the ones I used. You say I’m dead wrong. That implies Apple is lying about the 3 yrs.
        Got it now ? Always happy to help you figure out what you’re saying.

      • gigo

        Point to a single document from Apple that says the average lifespan of an iPhone is 3 years. And obviously don’t confuse this with supported life, life before a battery replacement may be required, etc, as that would be obviously stupid.

      • obarthelemy

        It’s already in the links

      • klahanas

        Not my field of interest, but don’t technical barriers introduce a distinct nature and thus separate market. More so than price tier alone?

      • obarthelemy

        Partly. But mostly, the $20/meal fast food market is distinct from the $200/meal restaurant market tough technically the same thing. Ditto paté vs foie gras, table wine vs fancy wines and spirits, no-name bags vs designer bags…

        I get very distinct requests for “the best cameraphone” or “anything that makes calls and texts, maybe with Skype, FB, and Candy Crush”. Treating those are the same would be like taking my 7yo nephew to a fancy restaurant: probably cost me money for nothing, and quite a few “cool uncle” points ^^

      • klahanas

        Interesting, because you can segment on the topmost level based on technical barriers. Android is unambiguously different than iOS. Different OS alone distinguishes them, if not the hardware which is largely the same or similar.

        Within Android you have both Prestige and Mass as a subcategory. This conversation would have one believe Apple is Prestige only. That’s ambiguous.

      • obarthelemy

        I think we can segment on tech because we’re tech-heads; but most people a) aren’t and b) care about other things such as brand image and looks. In the end, care about something that will make them look good, or not look bad.

        I’ve bought/reco’d 4 phones since Yuletide:

        – for my niece. Her current 1yo Samsung works perfectly and she only does calls and texts. But she’s flush with Christmas cash, wants something cool for school, and has been impressed by her mom’s new phone which I spent 3hrs setting up. Key ask: be new and like her mom’s. I translated that to 18:9 small bezel + TouchID, and a bit powerful because she’ll be away weekdays next year and might want media/games, and coolness.

        – for my sister, a very low-end messaging machine, plus a bit of podcasting/radio. Got her the smallest one I could – she’s got Trump hands, and made sure it was Real Easy.
        – for my elderly mom. Her Moto E2’s battery is dying. She wants the same phone because she knows how to use it. She’s getting my niece’s old phone and I’ll set it up the same (she hasn’t clicked on that, I caught her asking how that phone worked again last week^^ she’s worried.)
        – my 12yo nephew. He wants a gamer phone because he’s a gamer who games. So I got him a very loud bulky case in black and red. I think as for his gaming PC from this summer, the loudness is more important than the insides.

        Not one of these is about tech. Apple would have worked fine. They’re getting $100-$150 Xiaomi Redmis (5 Plus, 4X, and Note 4 Plus in order). That’ll work fine too (none of them takes pics^^).

      • Sacto_Joe

        You know, it’s such a laugh hearing an Android fan talk about how it’s “not fair” to compare all Android smartphones against all Apple smartphones. For literally years, many of us have been pointing that salient fact out, and that judging “market share” by including anything that labels itself a smartphone is just self-deception on the part of Android supporters.

        How refreshing to actually hear an Android supporter finally admitting it!

      • obarthelemy

        Actually, you’re seeing stuff that’s not there. This is about users, not devices.

        The markets are different (people who want to spend money vs people who want their money’s worth), but the uses are similar. I’ve said it umpteen times I can say it again: what/how much people do with their phones is barely correlated to how much they paid for them.

      • Sacto_Joe

        That’s really not the argument here, is it, obarth? The argument is that there are distinct advantages, over time, to buying a quality device, one of those advantages being the subject of Horace’s last two articles.

        But you can’t “see” that advantage when all anyone in a position to impact people’s opinions wants to talk about is “market share”, and then lumps in crap along with quality.

        And now that we’ve dragged you, kicking and screaming, into the discussion about quality, you pull this!


      • obarthelemy

        You’re devolving, that post is childish from the shortening of my handle to the final letter jumble.

        I don’t see that. I see that paying a lot more upfront + a lot for service for a little bit more useful life. That’s an issue with all luxury phones, not just Apple. A few are certainly getting their money’s worth by using the advanced features. Most aren’t, but must pretend they do.

      • usage

        “A few are certainly getting their money’s worth by using the advanced features. Most aren’t, but must pretend they do.”

        All usage statistics disagree with you here (app usage & spending, web browsing, social media activity, etc). There is a huge disparity between the usage of Android and iPhone devices.

      • usage

        Actually, even you admit this in a comment below:

        “Browsing stats are flat (and they over-represent iOS).”


        “The markets are different (people who want to spend money vs people who want their money’s worth), but the uses are similar. I’ve said it umpteen times I can say it again: what/how much people do with their phones is barely correlated to how much they paid for them.”

      • obarthelemy

        This merely says that iUsers like spending more money, doing more social, paying more for apps. I spend hours on my phone each day w/o doing any of that.

      • usage

        Yes, in other words, it merely contradicts your assertions in this thread about usage, value, etc.

      • obarthelemy

        Not when you understand what I’m saying: those things can be done just as well on cheaper devices, but some people just like to spend money.

      • usage

        Spend money to browse the internet for free? You’re asserting that usage differences are solely explained by what people want to do, rather than what they find easy to do. This assertion seem ill-supported and many datapoints, e.g. e-commerce and website design and loading speed, seem to lean against you here.

      • obarthelemy

        I kind of feel sorry I have to spell that out but to “browse the internet for free” you have to buy a device.

      • usage

        Are you okay? Did you forget the context of the discussion, which is comparison between the usage of devices which, by definition, have already been bought?

      • Sacto_Joe

        My apologies if you find the “shortening” of your handle objectionable, obarthelemy, but it is rather long. I’ll spell it out from now on.

        Regarding the Rolling On Floor Laughing My Ashcan Off jumble, you may call it childish, but I still get a kick out of what you said. Just too funny.

        For the rest, here’s a tiny example of a usefulness (there are many, many more such examples) of a full-blown computer ecosystem, from just last night: We recently bought a HomePod (for my birthday). I’ve been enjoying asking it to play stuff, but got tired of saying “Hey Siri, play it half that loud”, or some such, which caused Siri to pause the music, then turn down the volume, then restart the music from the point it paused. I spent a little time checking on the net, via my iPhone, and found out that I can actually go into my iPhone control panel, go into the music pane, pick our HomePod, and manually adjust my volume with my iPhone. I may even be able to do so with my Watch, though I haven’t tried that yet.

        And literally, just as I am writing this, I thought; “Hey. I wonder if, from another room, I can pick another song, and play it on my HomePod?”

        Answer, yes.

        THAT’S the power of a complete ecosystem.

      • obarthelemy

        Are you saying you can’t do the same thing in more open ecosystems ? I don’t have an Amazon or Google speaker nor a ChromeCast to check, but I’ve been remote-controlling VLC and DLNA stuff from my phone for years.

        I’ll do you 2 betters:
        there are probably smart speakers that don’t pause to adjust sound in both ecosystems
        my phone even acts as a remote for my dumb TV and stereo via its IR blaster. Works on pretty much any brand.

        You’re ascribing imagined exclusive qualities to your ecosystem, while not even aware of other qualities present in other ecosystems.

        That’s the power of the iBubble.

      • Sacto_Joe

        I was not “ascribing” any such thing. I was pointing out the GROWTH in usefulness for quality devices like iPhones. Literally, the value of my iPhone just went up for me.

        But interesting that you took it that way, obarthelemy….

      • obarthelemy

        “THAT’S the power of a complete ecosystem.” you said. Except that power is in all ecosystems it seems, even open ones. And the open ones also offer other “powers”.

        Anyone with a phone who learns a new trick with it gets more value. If that value is not phone-specific, it cannot be ascribed to the phone itself.

      • Sacto_Joe

        Are we talking about those ecosystems that don’t have anywhere near the personal security of Apple ecosystems, that come with security built in? Those ecosystems?

      • obarthelemy

        That’s a weird pivot. I’m taking you’re dropping the “magical closed ecosystem does stuff open ecosystems can’t” argument ?

        As for security, apparently there’s malware on Macs with a 10-yr lifespan ( ) and Apple itself has distributed malware-ladden apps ( , ).

        Google is publicly saying 0.4% of Android (unrooted, playstore-only) phones have a Potentially Harmful App, ad is actively monitoring each device. That’s good enough for me. Apple isn’t saying anything and is actively suppressing info ( ). That’s good enough for you, it seems.

      • Sacto_Joe

        That’s a weird pivot on your part, obarthelemy. The issue I’m talking about is personal privacy, not malware.

        Android is NOTORIOUS for giving away people’s info to the advantage of the vendor (Google, Amazon, etc). Take the voice-activated speaker devices from both. Forget about privacy anywhere near those buggers!

        I have zero issue on that scale with Apple HomePod.

      • obarthelemy

        You said “security”. You meant “privacy” ? Those are not the same. It’s getting hard to follow you.

        Re-pivot to privacy then (we’re getting even more off track, is that your goal ? I’d like to recap my point that that power is in all ecosystems it seems, even open ones. And the open ones also offer other “powers”, before we get utterly sidetracked) . Is there a doc somewhere about what info Apple gathers, what they do with it, how they secure it, and how they share it with 3rd parties ? I never could find any.

      • Sacto_Joe

        I have zero time to educate you to the facts about personal privacy on computer systems, obarthelemy. As to why you can’t find info on Apple sharing info with 3rd parties, it’s probably because it doesn’t happen.

        I know you’d like to discount the value of personal privacy, but it’s a major advantage that Apple gives it’s users. Just admit it, and move on.

      • obarthelemy

        “As to why you can’t find info on Apple sharing info with 3rd parties, it’s probably because it doesn’t happen.” Seems like you’re almost right:

        That means, wrong.

        That’s after 10s of googling. Apparently you’ve got zero time to educate yourself either, which has become kind of obvious.

      • Sacto_Joe

        1. Old data.

        2. Apple disassociates the person and the data. Does Google or Amazon?

        Speaking of time, time for me to go shopping. We’ll continue the “conversation” later.

  • HotelQuebec

    How much did Apple pay for this fake research?

    • melci

      What facts of Horace’s are you contesting? It is Apple who has publicly reported 1.3 billion active devices on their audited SEC quarterly report. And it is not hard to add up cumulative sales figures over the years to arrive at that figure of just over 2 billion sold.

      Do these facts that Google’s Android is far less dominant with only 2 billion active devices compared to Apple’s 1.1 -1.2 billion iOS devices perhaps shake your worldview a bit too much?

      • Sacto_Joe

        There’s an old saying: “Don’t feed the troll”. Eventually, they lose interest and disappear.

      • melci

        This is true Joe.

      • GlennC777

        “Never argue with a fool. He’ll drag you down to his level and beat you with experience.”

  • Jeff Liebermann

    Incidentally, Apple puts an artificial limit on how long a device is usable at 5 years after end of manufacture..
    This is much longer than the life of the typical iPhone. It’s also likely to be a design goal, where everything fails simultaneously at 5 years.

    • Glaurung-Quena

      “Apple puts an artificial limit on how long a device is usable at 5 years after end of manufacture.”

      Don’t be a troll. All that page says is that Apple doesn’t provide support or service for devices made more than 5 years ago. I own macs that are over 10 years old that still work fine. And I can still get parts for them, just not from Apple.

    • End of support is not an ‘artificial limit on how long a device is usable’. It’s an end of support date.

  • Glaurung-Quena

    The lifespan over time graph seems to start out being just for IOS devices (when we see a 1-2 year lifespan for years when the total number of macs sold still dwarfed the number of iphones sold). But in the latest numbers (unless I am wrong) it includes Macs as well as IOS. So there’s a shift somewhere in there, which should be noted.

    • Cumulative units calculation begins in June 2007 quarter. If including all Mac sales since 1984 the percent active is 60%.

  • Arnaud Campion


    Thank you for this research as well as the number.

    Is there a mistake about total device sold that should not include mac ?

    As far as I can see Tim cook speaks about iOs active device which exclude Macs.

    You are not speaking of mac anywhere apport from when describing total sold unit

    Could you please confirm ?

    • melci

      Not quite. Apple has stated the original 1 Billion active Apple devices were composed of “iPhone, iPad, iPod touch, Apple TV and Apple Watch devices that have been engaged with our services within the past 90 days” back in Jan 2016.

      • Arnaud Campion

        Yep that’s what I think, so the total sold device should not include Macs. And I think they do not. I believe it is just a typo from the Author

      • melci

        Sorry Arnaud, I made a typo. Refresh the page and you’ll see the quote did include the Mac.

      • Arnaud Campion

        Ok thank you melci, so there is a bia on the calculation because computer are kept longer but macs are outsold 1 to 20 including iPhone, iPad, iPod, Apple TV and Apple watch so that should not distord too much

  • ak

    The numbers are suspect, because the math is suspect. I.e : “17 quarters or about 4 years and three months” – 17 quarters is 4 years and one month. But this is minor. How do you calculate the average lifespan? From the text, it seems like you just define it to be so for no apparent reason. If you look at your graph, discarding the devices which are not in service, you can see that the number of devices 10 years or older is around 30,000 units. Roughly. Now, number of devices which are six years old is 120,000; 5 years old 180,000; 3 years old 420,000; 2 years old 300,000; 1 year old 400,000; less than a year old – 300,000. Now, roughly, if you calculate the average age of the device (multiply and sum the above numbers and then divide by the total number of devices) you should get 1525700000/1750000 ~ 872 days which works out to be around 2 years and 5 months. Lets just say two years, assuming that devices will die faster the older they are – accounting for the “bath tub” effect. Now, this number is much more believable, rather than your four years life span, on average. They, have enormously expanded their production only recently, since 2012 or so about. And so most of the devices are at most 5 y.o. now. A lot of them, I suspect around 60% are newer than two years old. So, something is wrong with your math and therefore numbers. Cheers.

    • By definition, 4 quarters = 1 year. 17 quarters is 4 years plus one quarter.

      • ak

        The “quarters” settled, I was faulting by aliasing quarters for months after the the mod operation 🙂

        The rest of the critique stands – are you going to address it?

    • The graph is cumulative sales not quarterly sales. The number of devices that are any age is the obtained by looking at the green area graph. 5 years old is equal to the cumulative sales up to end of 2012 which is 612,125,000.

      • ak

        I don’t see how what you replied has anything to do with the point I posted above. Sorry.

        So, if you will, can you please tell how many devices are now being used that are roughly 10 years or older. And how many devices are there in use which are made from Q412 to Q414 (roughly 300,000 units = 600k-300k, roughly). So if you do this and tabulate you can come up with some rough numbers that I posted above, If I didn’t commit a huge error. Then calculate the average lifespan of an apple device. I just don’t see how you come up with 4 years.

        Can you please work out how you get > 4 years with numbers. Like quarter to quarter incremental difference vs age, then calculate the average and please share the number.

        Also, I have no bone in this – I am arguing this simply because my brain was feeling being tricked by looking at that graph and then when I read the text – I couldn’t believe it 🙂

      • discarded

        Your calculation makes no sense to me, but it’s not clear exactly what you’re doing. You seem to be averaging over devices that are still in use, so of course you’re going to seriously underestimate the lifespan.

    • Sacto_Joe

      You, and others, are “missing it”: “Average lifespan” of any individual device is one thing, although even then you miss the impact of improving quality, so it’s ALWAYS a moving target.

      But more importantly, this is not a measurement of an individual device. This is all Apple devices. Yes, the number of devices grows each year. But your “calculation” is meaningless, precisely because not only the longevity of individual devices is continually changing but because more devices are added or subtracted each year.

      What’s important to focus on here is that the growth in all devices has been literally linear for years now. You clearly see that in the top graph. Devices sold are growing at a linear rate. Why? Because Apple is continually developing new devices.

      With that as a given (the linear rate), it is then possible to look at the past (slide those two dark lines in the top graph to the left) and see a certain pattern. It’s that “pattern” that Horace is looking at, and then projecting into the future. He uses the term “average lifespan” and mentions 4.25 years, but note that in his previous article Horace says “Note that the ratio remains remarkably constant. It’s currently about 64%.” That’s 64% of a ten year chart, which comes out to an average “lifespan”, by that measure, of 6.4 years.

      Back to the point about the linear rate of growth of Apple devices: For all we know, it could be a very long time indeed before device growth slows. If that’s the case, then it’s likely that the 64% “pattern” that Horace recognized is indeed, as Horace says in his other article, “…a rule of thumb which says that two out of every three devices ever sold by Apple is still in use. And that this rule is always true.”.

      But even that’s not the important point. We can use this same approach on all other collections of devices, whether they be from Samsung or from Hewlett Packard or wherever, and then make active comparisons of similar “average device lifespans”. That’s the worth of this approach.

      Now, it may be that Horace needs to define more clearly what he means by the term, or it may be that he even needs to use a different term. But if you can understand what I’ve said, then you’re bright enough to know what I mean, and thus what Horace means, and to not make false equivalencies.

      • ak

        A term “average life span” (of an average device) is quite well defined. As well as the concept known as average age of some certain population, for instance. So, do you see the graph charting the number of devices on Y axis and time on X axis? See? Well then, please calculate the number. This is not a trick question. In fact it is a simple number. If you can’t do that, how anybody is supposed to have any credibility in more convoluted concepts being talked about? Legit question.

      • Sacto_Joe

        I don’t care about what’s “quite well defined”. I do care about the point the Horace is making and that you seem determined to ignore because you are obsessed with terminology.

        It’s really very simple: The green area to the left of the thick left line in Horace’s first graph is the amount of iPhones not presently in use. The green area to the right of the thick left line in Horace’s first graph is the amount of iPhones presently in use.

        Really not hard to figure out. Oh, and btw, that comes out to roughly 60% of iPhones ever sold being presently in use. You’ll note that, if we go back 2 years, roughly 60% of iPhones cumulatively sold at that time were still in use, and if we project forward 2 years, roughly 60% of iPhones cumulatively sold at that time will still be in use.

        Get it yet?

  • apple refugee

    What about the throttling that apple did with iOS 11.1?!
    My iPhone 6 virtually stopped working – reception so weak as to be useless in all the same places that it worked fine for two yrs up to the day i upgraded…

    new battery gives the same life as the two yr old one it replaced!

    • Get Serious

      Yep I had an iPhone 6 on iOS 10 that crashed shooting slow motion/time lapse video and also in other high demand Apps due to a depleted battery. Missed recording a Grand Canyon sunrise because of it. I much prefer the “throttling” to a crash/restart.
      Maybe if you read release notes before doing updates you wouldn’t feel so butt hurt.