Website speed tests – Part 2

Awhile back I bemoaned the very inconsistent results I get from online website speed checkers. Not just inconsistency from checker to checker, but also from the same checker at different times. I test-drove a few of the checkers and …

In summary, results are all over the damn place. My site’s speed grade is A, B+, B, or C. Load time is less than or second, or over 4 seconds. Also, there seems to be no correlation between grade and actual speed. Pingdom gives me a ‘C’ with less than a second load time, Uptrends gives me an ‘A’ with a 4.4 second load time.

Furthermore, if I use the same 10 checkers again, I will get different – sometimes wildly divergent – results from the same checker at different times.

 

Today, Michael Bely, on his superb blog Research as a Hobby,  published The Uncomfortable Truth About Testing Website Speed And Hosting Speed. As anyone familiar with Michael’s blog would expect, his article is immeasurably better researched than mine.

Among his conclusions … “Single speed tests that you probably run to know your website or hosting speed make little sense.”

Micheal recommends using a paid service for constant, around-the-clock speed testing – he uses Monitis. For manual testing, “I run the manual tests about 100 times for each configuration from each testing location.”

OK, I’m sure that’s great advice but I’m not going to pay and I’m not going to make time to test hundreds of times a month. Michael seems to realize there are a lot of cheapskates like me, so he also thoroughly tested a half-dozen of the free online website speed checkers, in The Best Free Online Tool To Check Website Speed.

He concluded … “In my opinion, WebPageTest.org gives the most realistic website speed estimation.”

WebPageTest allows up to nine test runs at once of both first view and repeat view, and provides results based on the median test run. Using the median rather than the mean prevents an outlier test run from overly skewing the results.

So …

  • Based on Michael’s research and advice, I’m switching to WebPageTest for my monthly speed testing.
  • I still focus on raw speed results, not grades, though I do like getting ‘A’s.
  • To track a site’s speed over time, I do not use the raw load time. Rather, I use the delta between my site’s load time and that of a baseline site. My sites are WP, so I use WordPress.org as a baseline to measure my site’s speed against. I figure the experts at WordPress have optimized the speed of their site pretty well. If my site can match or exceed that speed – and they do, consistently – I think I am doing OK.

Update 2019-05-28: I just learned that Cloudflare added a speed check feature to the Speed section of their dashboard.

Cloudflare’s speed check is powered by WebPageTest.org – I am taking that as another credible vote of confidence over other checkers.

WPPOV supports freedom from Net Neutrality and the GDPR. The Internet of the people, by the people, for the people, shall not perish from the Earth.