Categories

Measuring iPad enterprise use

I’ve noticed that there is a lot of speculation on the prospects for iPad use “in the enterprise”. Business users could sure benefit from the device and anecdotal evidence does point to many cases of use in business.

But anecdotes are not conclusive evidence. How can one make a better guess? Based on reader input, I thought I’d test this hypothesis directly on my own stats.

As pointed out a recent posting on my stats, the iPad has been a popular device in accessing this site. 102k out of 1.06m views came from iPad users. A statistically interesting number.

One way to check enterprise use is to measure the ratio of use of the product during weekends vs. use during weekdays. The logic being that weekend use is almost strictly consumer use and weekday use is mixed. Any normalized significant difference between the two measures should indicate a bias toward consumer use or vice versa.

The following charts show the percent of each platform’s use with weekends separated from weekdays.

The larger the delta the more likely the more the bias one way or the other. To that end I measured the average percent of visits from iPad (as a total of all visits) for weekends and weekdays and measured the delta between the two. On the weekend the iPad was accounting for 13.3% of visits. On weekdays it was 8.8%. A difference of about 4.5% in favor of weekend use.

To calibrate that figure I did the same for iPhone, Mac and Windows use.

The following chart shows this ad-hoc index of “enterprise use”.

Essentially the data is inline with assumptions. The iPad is the most likely to be hired by consumers. The iPhone is more than the Mac and they’re all far more likely to be consumerized than Windows.

The average consumption of Asymco.com through Windows was 32% of during weekdays but only 22% during weekends.

In order to make this measure more meaningful, the evolution over time of these figures should be tracked. If the Blue bar above moves more toward the center then the chances are that iPad will be more adapted by business users, or at least more in use at work.

Does this seem like a valid method for measuring relative enterprise use?

  • asymco

    Thanks. Fixed.

  • FalKirk

    An interesting experiment. However, the primary underlying assumption – that business users don't use their devices on the weekend – has to be seriously questioned. I think that with regard to desktop computers, the assumption is accurate, less so with notebooks and not at all true with iPads. An iPad is the very type of device that a business user might continue to use throughout the weekend.

    • asymco

      I don't think the measure assumes that business users don't use the device on the weekend. It only measures that a certain device is in use over the weekend. The assumption is only on the *use* during a timeframe. The chances are that weekend use is not for business use. Therefore it's more likely being hired to do consumer jobs.

  • Ivor

    This is an interesting analysis, but I think you could get a better estimate by looking at the owner of each IP address associated with a visit and then bucketing them into coming from residential or commercial ISPs. The data wouldn't be be perfect, and a large portion of IPs would be difficult to classify, but it would paint a general pictures of the sources of your traffic.

    You may even be able to find an IP geolocation database that makes this classification for you, although I was unable to find one with 5 minutes of investigation.

  • asymco

    The 102k is indeed views (fixed now). My measurements however were made on visits.

  • http://aaplmodel.blogspot.com deagol

    If you have the hourly stats, one pattern I remember noticing when I used to do web analytics was, a morning peak for business use and a lesser evening peak for consumer traffic. My experience was with mostly US/Americas web users, so you might see a different pattern if your readers' regional mix is more homogeneous.

    Also, do you have enough data for a baseline of these weekend/weekdays ratios before April?

  • vangrieg

    A very interesting experiment indeed. However, I’m afraid it doesn’t say anything about enterprise use, because the underlying assumption that if people do something during work hours it is work related is, should I say, optimistic. I personally visit this site for anything but work.

    • Ralph Melton

      I agree that the assumption you've stated is not true, but I don't think that you need to make an assumption that strong to draw conclusions from this.
      You don't need to assume that these weekday people are visiting asymco for work; you just need to assume that they're visiting asymco from their work computer. And you don't even need to assume that; you just need to assume that they're more likely to visit asymco from their work computer when they're at work than when they're at home. I consider these assumptions plausible.

  • Ralph Melton

    If I understand correctly, the graph you've posted shows changes in share between weekday and weekend. What about considering changes in number of visits instead of changes in share?
    If there's a substantial difference in total visits between weekday and weekend, then it would be possible that iPad visits increased during the week, but didn't increase as much as Windows. That would still substantiate your claim that the iPad is much more likely to be consumerized than Windows among asymco's readership, but it would make an interesting statement about the use of iPad among business.

    • asymco

      Plotting the absolute visits does show iPad spikes during the weekend. That's what tipped me off to this method. I will update the post with absolute figures.

  • Rob Scott

    The methodology make sense, that is, closer to the centre means the iPad is used equally between weekday and weekends which might suggest business use. Using Windows as a control seem to suggest that your methodology works.

  • MattF

    How do you model the per-person visits (rather than the per-device visits)? For instance, there will be a (Windows-at-work and (iPad or iPhone-at-home)) population and an (iPad-at-work and (Windows or iPad-at-home)) population. Discriminating between these might be ill-posed without additional data.

  • http://twitter.com/Niilolainen @Niilolainen

    I buy it.

  • famousringo

    If somebody is using their iPad at work, they probably aren't surfing the web with it. They're probably in a meeting and either taking notes or reading/presenting documents.

    If they had time to surf the web at work, they'd be in their office, where a PC sits waiting at their desk.

    I think it's quite likely that iPad owners get more use out of their device outside work hours, even if they justify the purchase for work reasons, I just doubt this is a useful data point for analyzing that.

  • WaltFrench

    Well, you write well and all that, but I imagine most visitors are here because understanding the mobile internet is important to developing apps, investing or general business education. Now if I load up asymco on my MacBookPro from home @ 5AM before taking the train to work, is that different from hitting Starbucks with my wifi extender from my office midday? I think not. Neither the time-of-day nor IP approach will be perfect, either.

    But the info is already good enough that I'd love to see some other comparatives: "real" PC Chrome vs FF vs IE vs Safari. Android vs iPhone. etc.

    Thanks as always.

  • gctwnl

    The assumption is that weekend use is less work use and hence it says something about the question if the device is used to do a business/consumer job. However, reading asymco is different from watching funny YouTube movies. That is entertaining, but probably all readers are in some way interested from a business perspective. So how does asymco's content influence the equation?

  • gctwnl

    I mean, asymco is entertaining but probably more business oriented content.

  • Kristian

    It is unbelievable how COO's, CIO's and CEO's can skip the fact that the iPad is the ultimate INPUT device.

  • Kristian

    When The Apple adds the cameras in to the iPad.. well it will be the killer device… Can you say AAPL $2000? ;D

  • wpt

    These charts are confounded by the population bias of readers of asymco, who are probably much more likely to have *personal* computers and devices from apple than is the general population. I suspect that's a large part of the heavy bias of windows toward presumed business use in this particular population. It would be interesting to compare this to statistics on a more general tech site.

  • djbtak

    Exactly, I do a lot of my "business reading" on the weekend because I don't have time to do it during the week.

    Further, I think most of the interesting "enterprise" use is in verticals as an input device (healthcare esp), not as a replacement for desktop web browsing. I never browse the web on my ipad unless I'm in a cafe etc. or on my couch.

    Love your analysis of the market figures, but I think the methodology is a bit suspect on this one, due to some assumptions about "enterprise" and "weekend" that have too many different use cases that would need to be researched to provide valid data.

  • http://twitter.com/aegisdesign @aegisdesign

    I'm not sure of the merit of this.

    You'd have to

    a) account for international time zones

    b) account for 'office hours'

    c) survey more than asymco which is essentially a business site only.

    ie. 'enterprise use' would make sense if you geo-located every IP and matched it up with the 9to5 for that locale. I would guess most weekday usage of the iPad is not in the enterprise at all and simply evening use at home.

    I'm sure Hitwise or somebody like that has worked this out already and have a better guess at it.

  • Ziad Fazel

    Promising analysis, Horace.

    I second Ivor's suggestion about looking at whether the IP addresses are commercial/government or retail, but watch out for the small businesses who use retail ISPs. You've used both terms "enterprise" and "business" above, so maybe this analysis would clearly identify the "enterprise" but miss the use in "business" not large enough to have their own block of IP addresses.

    You could poll your readers directly for their mix.

    Perhaps you could compare your results with those of friendly bloggers such as Daring Fireball, whom I consider more of an entertainment read (no shot there, it is a nice break on a busy day, and still informative). I can see more people in the computer and mobile telephony industries reading your blog for industry insight.

  • EricE

    I work at a rather large (90K seat) enterprise and there is huge interest in the iPad. What you are seeing is the delayed reaction of corporate IT to support new technology in the environment – I don't think the tip of the iceberg has even been exposed on the potential for the iPad in business – many of the new enterprise management features in 4.2 will go a long way to accelerating testing and deployment.

  • JohnR

    I think there is a fundamental flaw in your assumptions. You assume that people looking at your web site are the "normal" business user. That is likely wrong. The reason is that because you have your readers who are all highly interested in mobile, they are more slanted toward the mobile device in the first place. So they may have purchased iPads as it is/was the hot mobile technology item, and it was not purchased by their company. The underlying question really is "will enterprises use the iPad" and although I don't profess to know the answer, I don't think your way is a good way of measuring. I do think your conclusion is right for the time being, which is no, the iPad is for the consumer.