John Resig has just put out a nice post about browser share. It’s well worth reading and I won’t repeat what he says there. I do, however, think the cost benefit balancing graph is a worth examining further. If you take a look, it seems to say that IE7 is a lot of work, but worth targetting, whilst IE6 is not worth the effort. However, this is very dependent upon the browser share and the traffic to your site. The yellow bars, the cost, are pretty much constant across browsers. The blue bars, the benefit, can be expressed as follows:
Number of users of this browser x Revenue per user for this browser
Now, for a seriously low traffic site, that might suggest that you shouldn’t even bother supporting IE7. In practice, you will, because you won’t feel like you can take yourself seriously if you don’t. IE6 becomes a complete waste of time. If you’re google, on the other hand, all of the blue bars are in the stratosphere, and you should support browsers that people haven’t even thought of yet.
In practice, the cost benfit nearly always favours supporting IE6. Let’s assume it took three months to support IE6 to the extent that someone could use the site, that the conversion rate for IE6 users is 7% and that you got £60 per action. (These are reasonable numbers for some dotcoms.) Finally, assume the contractor that does the work costs £300 a day and demands payment up front. Then, if you had 137 visitors a month after that point, you’d get a 5% annual return on your investment. That’s assuming a three-year planning horizon. Obviously, the numbers go down if you have a shorter horizon and up if you have a longer one.
137 visitors a month is nothing. It’s more than this blog manages, but a half-decent commercial business with a bit of TV advertising would have thousands of visitors. You only need 4500 visitors to make 5% back the month you go live. Obviously, you need to run your own numbers: some people will see the quick numbers I’ve made up and not even vaguely recognize their own business. But I can reasonably guarantee you, if you’ve got a public website, you’re not going to be able to convincingly argue for the demise of IE6 any time soon.