Our post last week regarding Android’s increased share of mobile web consumption got quite a bit of attention, and we had many comments. Several of these asked for clarification of the methodology we use and, rightfully, highlighted our lack of caveats.
So, here’s is a reminder of our approach and important things to keep in mind when using this data.
Where does the data come from? All data comes from the Quantcast Measurement service. This free service offers any web site owner detailed information relating to their site’s traffic patterns, and demographic, business, geographic and lifestyle data about their audience. The service operates via a measurement tag that is embedded in the HTML of a website. Millions of web destinations participate in our program and every month we observe hundreds of billions of pageviews worldwide. By analyzing the user agent strings in the HTTP request we can determine the originating operating system, browser and for many mobile devices the manufacturer and model number. Our most recent August analysis was based on more than 4 billion mobile pageviews.
What are you measuring? The measurements we provide are the relative proportion of pageviews attributable to the given device, browser or operating system. We provide either the relative proportion of all web consumption, or the relative proportion of all mobile web consumption.
Why do you measure the relative share, not the absolute levels? We use share because our data set is not static. Every week thousands, even tens of thousands of new web sites might sign-up for our free Quantcast Measurement program – after all it’s the first, and only, syndicated online traffic measurement service to gain Media Ratings Council accreditation – which means our absolute numbers are always going up. By presenting this data as a share we normalize for the continual underlying growth in our program.
When using our numbers it’s important to keep this share approach in mind. For example, the amount of mobile web consumption is growing very rapidly compared to all consumption, so a vendor who’s share of mobile web consumption hasn’t changed in a given time period may still have have seen much higher overall usage and an increase in their share of all web consumption. Because mobile is growing quickly and the space is highly competitive, we feel that the mobile only comparisons provide a useful perspective on this important market’s development.
Is the iPad included in mobile? No, we do not count the iPad as a mobile device. This is a judgment call and we fully expect that over time the distinction between device classes will continue to blur and these decisions will get even harder (as they will with the forthcoming slew of Android tablets). As mobile becomes a more significant portion of all web consumption and as more devices appear we may drop the distinction and shift solely to total web consumption share (if you have thoughts on this please let us know).
If a Quantcast chart shows iOS in respect of mobile web consumption, then it includes both iPhones and iPod Touch devices, but not iPads.
Why only website consumption, why not mobile apps? There’s no doubt that mobile app usage is growing rapidly, and we’d love to include it in this analysis, but there are several reasons why we don’t at this time:
i) Apps have to be individually instrumented and, often, then approved. This is more involved that a simple HTML tag and so we have less distribution for app measurement than we do for general website measurement.
ii) Apps may be platform specific making it harder to compare usage. When a website is browsed we’re able to measure the consumption regardless of the originating device.
iii) A pageview is a generally well understood metric for website media consumption and therefore provides a good basis for comparative analysis. Mobile application usage does not offer the same standard consumption unit, though we do expect standards to emerge over time.
So, for today at least, we feel that a comparative analysis of browser based media consumption offers the fairest comparison basis for our data set.