Using Data to Create a Global Browser and Device Matrix

In an ideal world, any technology project would be able to support all browser and device combinations. The reality is that there are always going to be some users who have ‘the wrong browser’. Clients want to reach as many people as possible, but especially for global brands who operate in a large number of local markets, the combination of supported browsers, devices and operating systems can be enormous.

There’s a practical limit to the number of device and browser combinations that it is cost effective to test against. So how do you come up with a client list of supported browsers and devices.

How we solve the problem
There are several solutions to this problem, and different types of projects have varying constraints. For some applications the core idea will dictate a basic device capability (such as having a camera). For other projects the objective is to look at the client audience across different markets and see what capabilities the current and potential audience have and ensure that the solution meets that baseline.

Using the raw data from current traffic data and also the browser and device profile of each market we can get a picture of this browser and device landscape by market. By feeding this data into a simple programme we can then automatically generate a global and local browser matrix based on pre-defined rules. This has a number of benefits:

  • It’s repeatable, so we can re-sample in the future and get a view as to how the client’s browser and device landscape is changing
  • It’s data driven. Whilst you can prove anything with statistics, having some cold hard facts that everyone buys into can be very helpful
  • By manipulating the business rules, you can modify the weighting between global and local, which helps clients manage the amount of effort required to support local market variations.

So what do these rules look like?
Despite having a data led approach, there’s still more of an art than a science to coming up with the initial rules. As a starting point we can set the bar to support any browser or device that makes up more than 3% percent of the clients current web traffic (using data from their current analytics). The same rule can also applied to the global browser and device composition for the clients key markets. Using this rule will give us a good picture of the browsers used across the selected key markets.

However, the danger with this top down approach is that local market variations are missed. So if a browser or device has a very high penetration in a specific market, but globally is less than 3% it would be excluded from the generated list of supported browsers. To counter this we can also include any device that makes up more than 3% of the local market and also more than 0.25% of the global market. This allows us to take into account local market variations, without having to support devices that represent a very small number of global users – and avoid small local markets distorting the global browser and device support.

Ultimately the aim is to come to a compromise that maximises the number of browsers and devices supported in order to give the best user experience to the largest number of users, within the constraints of a realistic amount of cost and effort.