Analysts' categorical failure

Gartner explicitly explained so in its press release: “Gartner’s PC group does not track media tablet sales in this PC shipment data, so iPad sales are not included in these results

via Is Apple the real U.S. PC market share leader — or soon will be? | Betanews.

The reluctance of  industry analysts to measure the iPad as a computer is a fascinating and vivid symptom of how analysts conspire with their customers to smother visibility of impending failure.

Before we dive into the motivation to ignore iPads, we need to understand how analysts in general and companies in particular group products into comparable piles. This process of grouping is called categorization. It should be distinguished from segmentation which groups buyers in a market. Categorization is essential for competitive analysis and measurement of the performance of a product (relative to other products).

Categorization is a challenging problem for most companies. To illustrate why, consider four possible methods for categorizing products:

  1. By product attributes. The assumption here is that products with similar attributes compete in the same market. As an example hard disk drives might be categorized by dimensions of the cases and sub-categorized by capacity and perhaps transfer speeds.
  2. By user attributes. This method is similar to segmentation where products are grouped by the type of buyer. Magazine titles might be properly categorized by the age and gender of the readers. This can also cover geographic data.
  3. By user behavior. This method, sometimes called psychographics is a more advanced method than demographics where user personality, values, attitudes, interests or lifestyles are measured. You might have seen examples where users are defined with a “prototype” person and the products are attached to this prototype. This is a common way to categorize mobile phones (Nokia is a known leader in this method).
  4. By modes of consumption. Determine what products are hired to do the same job. In this case you have to have real insight into what people actually do with their products and try to look broadly at what they substitute. The classic example of this is to understand that a power drill competes with products what make holes in walls not just with other drills.

When reading this list, it’s easy to see that it’s ranked in order of difficulty. The easiest thing to do is group according to things you can measure easily. Products are easier to measure than demographics which are easer to measure than psychographics and that’s easier to capture than observed behavior.

Consider how you would structure a study to capture this data.

  1. In the first case, you need an analyst to collect products (or just their descriptions) and record values in a spreadsheet.
  2. In the second case, users need to be surveyed for basic data. This is hard because you have to find users and then you have to get a statistically significant unbiased sample.
  3. In the psychographic study you need to ask a lot more questions about lifestyle in your survey and hope that the subject answers honestly. You still need a large sample.
  4. In the case of consumption analysis you need to observe a subject use a product for a prolonged period of time in the usual context where the product is used and be able to interject questions when certain behavior is encountered. The observation of one person can take one research person-day per subject. To get statistically significant data requires literally man-years of work.

Now it should be clear why an analyst will only categorize according to product attributes. It’s not only economically and temporally expensive to perform research on methods 2 to 4 but it also requires interpretation and insight into study development.

A few companies go beyond buying analyst market data and do their own research. As far as I know user studies are more commonly done by consumer products companies. But it’s extremely rare to see this research done by low-margin technology companies. They don’t have the budgets or skill sets to do any research.

However, in my experience, nobody performs the fourth, most difficult, form of categorization.

The result is that good, smart managers trained to always base their decisions on hard facts, pick up these lowest-deonominator analyst reports to help decide which products to develop. And it should be obvious why they are surprised when products that did not show up on any market study suddenly take all their profits away.

Thanks Gartner.

  • It was this kind of rigid thinking – categorisation and segmentation – which led to the product line nightmare of Spindler's 1990's Apple. Quadra / Centris / Performa XXXX±yz. Ugh. Who would be mad enough to run with that nowadays? Ahem.

    One of the first things Steve Jobs did on coming back was enforce a strict four-spot product matrix. A classic sign of things to come.

  • Jason Grigsby

    The first thing you have to do to use an iPad is sync it to a real computer. You can't do system updates or backups without tethering.

    Until that changes, it seems to makes sense to keep the iPad separate.

    If not, then why not include the iPod touch in the computer number comparisons? It is simply an iPad with a smaller screen?

    If you include the iPad, then why not include the iPhone as well? It is simply an iPad with a smaller screen and voice.

    Of course, if you include phones in the mix, then suddenly Apple isn't the largest computer manufacturer. Nokia is. And unsurprisingly, they claim to be the world's biggest computer manufacturer.

    For my money, keeping the iPad separate makes sense until the ipadcan function without requiring tethering to do basic functions.

    • So you are arguing that products need to be categorized according to their attributes and competitive assessments must not stray into discussions of substitutability.

      The argument of whether a device (be it a Nokia phone or not) belongs in a category of use has nothing to do with its attributes. Indeed, it has nothing to do with being a device at all.

      • No, I'm arguing that a product that cannot function without another product should not be lumped in with that product as far as market share is concerned.

        An iPad has more in common with a printer than it does a netbook. An iPad requires a computer in order for you to be able to use it. So does a printer. A netbook does not require that you already own a computer in order to use it.

        This may very well change in the future. In fact, I expect it will. But until it does, it's relevant that the iPad is still built on the digital hub strategy that Steve Jobs articulated in his 2001 Keynote. Here's a pretty good retrospective on it:

        My point is from an analyst's perspective, it makes no sense to bundle an iPad with computers. From a company perspective, especially if you sell netbooks primarily to people who already have computers, then you better be looking at the iPad.

        My breakdown of the market would be something like this:

        * Desktops
        * Laptops
        * Netbooks and Tablets
        * Smartphones
        * Feature phones

        The question of what constitutes a computer I find much more difficult to answer.

      • "More in common with" is too strong of a statement. Wish I hadn't wrote that. But you get my point. 🙂

    • Tom Ross

      Actually Tomi Ahonen made a list of biggest computer manufacturers in the manner you are describing.

      Apple was #4 in 2008, #3 in 2009 and I have a feeling they'll give Nokia a run for the #1 spot in 2010, with 80 to 100 million units shipping this year across Mac and iOS devices.

      • I would not say that every smartphone is competing with PCs, but this gives a boundary.

    • Ted T.

      Actually, beyond the initial activation, which just as easily can be done using a computer at the local Apple Store, the iPad functions just fine without ever again synching to a computer. Putting it in a separate category because of the activation requirement is silly.

      Also, if via iOS update, Apple eliminates even those minimal requirements, does the iPad that was not until now a computer suddenly become one?

  • ChuckO

    These categories seem to have far more value for analysts than companies building products which is probably why Apple has been successful building what Apple exec's would want to buy and not what market research thinks consumers want to buy.

    The problem for most companies is the central problem of our times. Most companies start with the premise that they need to be successful so what can they build that people will buy. They don't start from taking a chance on an inspired idea they deeply believe people (as opposed to consumers) NEED.

    Sorry to sound pretentious.

  • ericgen

    Thanks for the explanation and insights into different ways to categorize things!

    It seems like Apple consistently aims for 'mode of consumption'. I wonder how much of this is an intuitive sense and how much is conscious. I guess it's irrelevant since the lines between the two tend to blur given enough practice and experience.

    It would be interesting to see a 10-20 years sales analysis of what, for a better name, I'll call "stereo equipment". I seem to remember reading things that indicated that even by 2000 much of the sales of stereo equipment were starting to be for home theater. I wonder how much, over time, the iPod affected these trends and stereo equipment sales. I would guess that outside of home theater sales (which the iPod may have actually helped by simplifying the choices for just music listening), that they probably significantly affected the sales of non-home theater audio equipment, and not just portable things like the Walkman or the other media players.

    I wonder if the iPod experience on stereo equipment might be a reasonable view of what might happen to traditional computing. When there is another important component, such as large-screen video in home theater (hence lack of mobility – at least currently), then the more traditional computing equipment will be required. But, for almost all other less constrained uses, the iPad will become a sufficient substitute for traditional computing and completely disrupt the existing structure.

    There are frequent complaints in the media about what's missing in the iPad. There have been complaints since its initial debut about what's missing in the iPod. It made no effective difference to the iPod's dominance and it's looking like it will have little effect on the rise of the iPad. Both are 'good enough' in traditional areas and fantastic in areas that no one ever really considered.

  • ChuckO

    "Things are seldom what they seem, skim milk masquerades as cream"
    – Gilbert & Sullivan

    "The reluctance of industry analysts to measure the iPad as a computer is a fascinating and vivid symptom of how analysts conspire with their customers to smother visibility of impending failure."

    So are they "industry analysts" if they "conspire with their customers"? I'm assuming the "customer" here is specific tech companies not say investors, correct?

    We seem to have hit a point where a lot of our information is really a denial of service attack for the brain. Figure out ways to preoccupy and overwhelm peoples logical abilities by starting arguments based on nonsensical premises hoping they become the baseline thinking people use. I can't personally see the value in categorizing based on any of those 4 criteria or any criteria. You be better off going with fusion not fission and just compare companies that are selling processor based products.

  • jane Zweig

    Another question/issue is what is the definition of an industry analyst. There are analysts who are paid by companies to say what a company wants them to say. Then there are analysts who drive share holder prices. Then there are analysts who "forecast" based on only today, hit some button on a computer, and do not take into account how politics, technology change, markets and economics will change the impact but these are firms that must "supply some number". Then there are analysts who are totally independent and say what they believe the truth is but companies don't hire them back as it isn't what companies want to hear.

  • Pingback: asymco | Mobile devices overtaking PCs()

  • Sir,
    We track iPad (and other slate) shipments and include them in our Quarterly Notebook PC Shipment and Forecast Report (, which I happen to be the lead author of. At present, we track slate shipments in a combined "mini-note/netbook/slate" category but will, eventually, break slates out into their own category. This was a topic of significant discussion within our organization and we decided the best place to track them was with mobile PCs.
    On your last point (#4) , our parent company, The NPD Group ( has a vast consumer panel which is used to track various aspects of buying and product usage behavior.
    -john jacobs
    Director, Notebook Market Research, DisplaySearch, an NPD Group Company

    • You're doing the right thing. There was a time when smartphones were new that they were also mis-categorized as PDAs by some analyst groups. The first company to do it right was Canalys and they prospered as a result.

  • If “derived” Android tablets become a significant proportion, lumping them into “other” would mean the report would be less useful as it would not shed light on market size for apps on such devices. It seems wise to be the first to pick a new category for these devices. The interested could be enticed to buy another report that further classifies, quantifies “derived” Android tablets.

    edit … posted in the wrong thread. I meant it for