When we talk about web performance, we often talk about the user experience. A slow website means visitors are less likely to be happy. Unhappy visitors might leave and go to a competitor or, at the very least, be inclined to view your organisation less favourably.
With this in mind, simple common sense suggests an obvious link between web performance and revenue, and there are a number of well-publicised case studies that back this up.
Nevertheless, from time to time, we still encounter a bit of scepticism. Typical arguments run along the lines of: “Sure, I understand how performance matters to X. But our customers are different. They won’t mind waiting because our products are unique.” In some ways, this is fair enough. It’s possible that these site owners are right. The weight of evidence may be against them, but without testing fast and slow versions of their site, it’s impossible to say for sure.
However, customers (and prospects) aren’t the only ones accessing your site. Search engines also crawl and index sites, and Google has for some time used site speed as a factor in search ranking (http://googlewebmastercentral.blogspot.co.uk/2010/04/using-site-speed-in-web-search-ranking.html).
What’s more, site speed is also a factor in Google’s Quality Score for landing pages in PPC (pay-per-click) advertising. It is, of course, one of many factors, and you certainly won’t want to sacrifice relevance, for example, for the sake of speed. However, if your landing page is too slow, the chances are that it’s costing you either a) money (in higher costs per click) or b) click-throughs and conversions (because your ad isn’t ranked as highly as your competitors’).
This, then, is at least one argument for the financial benefits of web performance that can’t be dismissed out of hand. Even if your customers don’t care about a slow website (and that’s a big “if”), Google certainly does.
What do we mean by a slow site?
We’ve been talking about “slow” and “fast” sites as if these terms were simple to define. Of course, they’re not. Do we mean time to first byte, render start time, time to visually complete, time to interact or total load time?
At this point, we have to admit that we just don’t know for sure, nor can we. The precise nature of Google’s algorithms is a closely guarded secret. It’s been suggested that time to first byte is what matters, at least when it comes to organic search, and in some ways this would make sense. It would mean that you’re not penalised for too much content or for poor performance of third-party objects. However, it could also weigh excessively in favour of organisations with dedicated, high-performing web servers, to the detriment of smaller organisations or individuals on shared hosting accounts.
Another possibility is that the window.onload event is what matters. This can be inferred from an article by Matt Cutts, Google incorporating site speed in search rankings, which hints (without explicitly saying) that performance is based on the same figure as the one used in Google Webmaster Tools. Time to the window.onload event would probably be preferable to time to first byte, since it would allow front-end optimisation to be taken into account.
Despite the fact that we have to accept a degree of uncertainty, the important point is that it matters if your landing page is slower than it needs to be. And it matters not just because of its impact on user experience, but also because – in the case of PPC advertising – there’s a definitive link between performance and profit. It might be difficult to define, but it’s impossible to deny.