Whilst working at my last job, I was surrounded by developers and not much else in terms of staff which was a great a learning environment for me, I think that is one of the great things I got out of that last job as there was little else going for it (more on that in another blog). Being surrounded by developers meant I had to listen to their thoughts, ideas and workflows all day, at times I was yawning a lot and at other times I was in awe!
The company I worked for specialised in building online marketplaces like Gumtree or Craigslist in underserved markets such as Ghana, Bangladesh and Sri Lanka. The fascinating part of this was that we had to totally double-down on making the user experience (UX) perfect for mobile users in these markets. This “mobile-first” approach should be the priorty for all online business these days but for us it was absolutely essential to get this part right as pretty much the majority of our users used mobile devices to access our web platforms, many using “dumbphone” which presented all sorts of interesting technical challenges which I won’t dwell on here.
When dealing with mobile-first and you also consider the SEO implications it pays to actually understand the users position and location. Many industry SEO folks refer to the Google PageSpeed Insights tool as the place to go to, this I am afraid is not sage advice for several reasons. This tool does not know where your users are, as we had users in Bangladesh, how would it test for these people, it doesn’t. The Google PageSpeed Insights tool as far as I am aware is run from a few random data centres and does not test locally to where your website is aimed at in terms of target users. We actually installed local machines into each country to run our tests locally on the local internet infrastructure and monitor here from Sweden, that is how obsessed we were on it. We ditched Social media buttons and much more for loading speed, it was like one of those movies where there is an aircraft in trouble and they start ripping out the seats and throwing them out the door to maintain their altitude!
The next issue is that in terms of page loading speed, from a user perspective or UX it is more to do with how quickly you can render the page to the user, many of the page speed analysis tools use a crude measurement of when the page is requested to the time it finishes to load all the calls to the server(s), where as a better measurement is actually how quickly the page renders for the user on their given device in their location. Good developers know that the order in which they put their code can make elements of a webpage load quicker for a user and have other less important elements load in the background after the main UI (user interface) has been presented to the user.
Here is an example of this site compared to some of my industry peers and people I look up to using a tool called WebPageTest I think it gives more attention to the user when we are considering page speed or loading time. This website is hosted in Canada as I have a good relationship with a hosting company there. So to make this fair I only added sites from North America, the test was performed from Texas and I am still ahead of the industry gurus. That is fair, because at this stage my site has not got much on it, but I have decided to not install countless WordPress pluggins and I don’t have too many on page graphical elements yet. But what you are seeing in the image below is how a page renders for the users and I believe this to the primary metric that should be followed and not simply the rather basic Google PageSpeed tool.
This uses what is called the Speed Index which I find a much better KPI.
There are lots more things that can be said on this subject by wiser and more educated people than me, but for now think about what SEO metrics you are using and how you can get led down the wrong path. Put the user first and test from a users perspective, think about KPIs, are they suitable and using the right methods to capture the data needed?
Some useful tools for analysing page speed: