Lessons to be learned when tuning websites
Whenever we have been involved in tuning Databases and front ends for Clients we have always taken the approach that getting fast performance out of the system is a little like using a cheese cutter, sometimes you have to whittle away patiently until an acceptable performance point is reached, and with the notable exception of a council in London where we got a slow running report down from eleven hours to 30 seconds in a matter of an hours work, this has always been the case.
This week however the old dog learned a new trick or two, every 18 months or so Google contact us to get the best out of our not unsubstantial Ad-words campaigns and while this is obviously designed to get more clicks and therefore more revenue for both parties, following the consultation we focused our attention to the performance of our website, now over 17 years the website has gone through many different configurations and styles taking on board the new techniques available and the latest version (15) is designed to render on both desktops and handhelds equally well.
So bearing this in mind we looked at the performance of the main page (there are over 200) and it got a score of 86/100 for the smartphone and 36 for the desktop !, so something was going wrong with this page, we turned to the GTMetrix (https://gtmetrix.com/) site (there are lots of others) which performs timing analysis and gives advice as to improving the metrics of the page, now a lot of it is quite specialised techie stuff which we took on board and applied, but strange techniques like placing all the small images into one single file and using the CSS Sprites facility is a new one on me and we gained 10% by doing that alone, anyway to cut a long story short and a lot of work over the last week or so, the entire website is now running at an average of 92% per page so very pleased to report no waiting for https://www.seercomputing.com to load and everything is back to normal.
Do a speed test on your site you might be surprised, no point in paying for clicks if the page takes too long to load.