Apple’s Graphic Failure

2013.01.20   prev     next

FROM Bare Figures, here’s Mac unit sales over the past dozen years:

Here’s the same graph with iPods added:

Here it is with iPads:

And finally, with iPhones too:

Notice that all of these graphs have something in common. Except for the iPod, they’re all currently going up. And except for the iPod’s sudden rocket-blast-off around 2004-2005, there’s almost no sudden change in direction. The trajectories are all very consistent, and experience only gradual change in direction over a period of several years.

This all strongly suggests, of course, that Apple is very unlikely to find itself in any kind of serious trouble for many years to come, at least. But you wouldn’t know it to read many pundits’ dire predictions of near-term Apple decline.

Graphic Failure

Most of these down-on-Apple commentators don’t grace us with a graph of their predictions, but every now and then they do. Here’s exactly such a graph from Pyramid Research:

Graphs look authoritative, don’t they? And this one seems to indicate that Windows and Android will be neck-and-neck winners in the mobile computing space (Windows arbitrarily slightly ahead of Android), iOS and Blackberry will be tied for a very distant third-and-fourth place, and Symbian will be dead and forgotten.

Now, let’s draw some lines over this graph and see what’s happening with it:

The green line represents the present, i.e. when the graph was created. Everything to the left of the green line is actual data (perhaps skewed with “shipments” tricks, but for the sake of the argument, we’ll assume it’s valid). Everything to the right of the green line is predictions (projections, forecasts, whatever).

The emphasized line represents Android. As you can see, up to the present it is increasing, but progressively more slowly. And after the green line, it continues to do exactly that, with no noticeable inflection point at the green line. So this can be considered a reasonable projection of the past into the future. Of course, there’s no guarantee that that’s exactly what will happen. And the farther you project the past into the future, the less likely it will be to happen that way. (And this graph projects pretty far into the future, you’ll notice.) But still — a reasonable projection.

Next, here’s what they did for BlackBerry:

Again, a reasonable projection. The line is going down, but progressively more slowly. The future line continues that same trend with no noticeable change.

Now — here’s Symbian:

That’s weird. Why the big inflection point right at the present? Don’t get me wrong; at the time this graph was made, I had no quibble with the general idea that Symbian was in severe decline, nor that it would be effectively dead several years later. Nor do I dispute that thesis today. But still — why the big change in direction right at the green line — the time when the graph was created? That’s very suspicious. I suggest that this big dip is necessary to make the graph add up to 100% over its whole range: What we’re about to see below (regarding iOS and Windows) is why Symbian has to take a suddenly steeper dive at the green line.

Now for iOS:

The past showed iOS climbing at a slightly increasing rate. Then suddenly, at the green line, it changes direction to downward. And does so at just the right speed so that it can be arbitrarily ahead of BlackBerry but otherwise as small as possible.

All of these predictions leave just enough room in the 100% total for Windows to do this:

It takes a huge and abrupt upturn at the green line, and winds up barely in first place, just the tiniest bit ahead of Android.

Now, I leave it to the reader to ponder: What are the chances that this graph was made by any predictive technique more sophisticated than, “Somehow, we have to predict market victory for Microsoft. How can we bend and twist these lines to make that outcome occur, while still keeping the graph total at 100%?”

Backhanded Compliment

Even when market analysts seem to be saying something positive about Apple, it’s not always so. Consider this graph of “iPhone Estimated Installed Base” from Rob Cihra of Evercore Partners:

Hey, that looks positive, doesn’t it? Climbing at a nice, healthy rate. No abrupt downturns. What could be anti-Apple about that?

But notice that the graph’s title included the word “Estimated.” And notice all those “E”s starting in December of 2012 — they’re easy to overlook, and they’re the only indication that those quarters are “estimated,” not actual. So now take a look at the exact same graph with a big, fat, green present-line on it, plus a reasonable (my freehand) continuation of the past into the future:

Ugh. The reasonable projection turns out to be about 70% greater than Cihra’s “estimate.” His graph has no obvious change in direction that pops out as in Pyramid’s, but still: an arbitrary and massive reduction in Apple’s future potential, with no basis in past performance.

Future Uncertain

Now, again — and I can’t emphasize this enough — I have no way of knowing what Apple’s future holds. It might be much better than a reasonable projection of its past, or it might be much worse. It might even be worse than Pyramid and Evercore said it would be; who knows?

My point is simply that when you look at forecasts of how Apple and its competitors will be doing in the next several years, keep in mind that there’s a good chance that the forecast is being made by people who have little or no interest in how their predictions will look a few years from now. They simply want to be negative about Apple today, and they use their analyst position to do so.

Update 2013.09.10 — The Wall Street Journal’s recent graphs showing the very-near-future of all three iOS devices:

To their credit, they’re going only one year (or less) out, and they make the not-actual-data portion very visibly different from the actual-data portion. And the iPod Touch curve is only the tiniest bit worse than a reasonable projection of its very smooth, many-year curve. But what’s up with the other two graphs? Can you say, “massive downward-bending cusp at the now-line, with no other noticeable cusp in the history of the product?” I knew you could.

prev     next