Jobin John: If I click on a link and it does not open up in 5-10 seconds then I just hit back and then click on a different link. Now connect the dots... at this stage Google thinks the link I clicked did not have relevant content which made the visitor go back and find something relevant... So Google just puts you down as for the visitor he/she just hates your website for wasting their time by not loading fast enough.
Casey Markee: When we talk about page speed you should definitely focus on load time over a 3G connection here in the US. Which is the dominant standard until 2020.
Make sure to review the Google Test My Site tool here:
Google doesn`t really have a "page speed benefit" as opposed to a "page speed detriment." Once you get to a certain level of page speed you don`t get some kind of "magical benefit."
But if you are very slow, then that will definitely hurt you.
When Google introduced Page Speed as a ranking factor in 2010 they also said less than 1% of all sites, at the time, would be penalized by this. Most likely though, that number has gone up considerably.
Google likes to say that they want pages to load fully in "2 seconds or less." And this is because the chances of a "bounce" rises considerably when the user goes from 1 to 3 seconds (I believe it`s like 30+% or more).
As for the Google Pagespeed Insights tool, that is more of a USABILITY tool, not a PAGE SPEED tool. Avail yourself of the Test My Site tool above then use tools like WebPageTest and GTMetrix to refine the recommendations down.
Bottom Line: focus on speed over a 3G connection as Google prepares to flip everything to the Mobile-First Index now and in 2018. The faster you can get, the better.
Michael Stricker: If you can’t get users to stay, you can’t get users to pay! So be sure to consider speed-related bounces as lost opportunity. That frustration builds as users go deeper into the site... more pages, more delays, until *poof* you’ve lost them. There is your “upside” to “faster than the competition. Don’t risk losing a conversion from traffic that you paid and labored for.
Alan Bleiweiss: I love ALL of these responses! They`re all valid considerations - speed matters. What I tell clients is this - even when the "average" over 30 days, according to Google Analytics own speed data, shows speeds in the three to five second range, you need to drill down into individual page data, then add the secondary dimension for device category. Then plot those numbers. You`ll sometimes find that speeds are intermittently shocking in how slow they are.
And don`t rely on Pingdom for accurate speed data. It`s good for basic insights into speed fix opportunities, however consensus across the industry from dozens of professionals I am connected with (me included), is that testing with GTMetrix or WebPageTest.org (emulating 3G) is the only way to get fairly accurate insight into cases where you`ll see more accurate speeds.