Updated: November 2019
There are no affiliate links on PagePipe.
I got 100/100 in GTMetrix. Is my site now good enough?
Scores representative principles that make little speed difference for most user experience.
- load time in milliseconds
- page weight in kilobytes
- the number of requests
Scores aren’t on the list of importance for evaluation. Even the number of requests isn’t super important because of browsers loading assets in parallel.
Speed scores are artificial or superficial criteria based on the original concepts of the Yahoo Performance team. In particular, Steve Souders the mastermind behind Yslow Score. Google pirated Sounder away to their team where he invented PageSpeed Score. It wasn’t innovative but a clone of what he’d done before. Now more extreme. (Impossible?) He coined the title performance engineer and page speed. At Google, he established the same obsessive-compulsive criteria that evolved into PageSpeed Insights test. He also influenced other ideas like introducing speed into the Google ranking algorithm to persuade compliance.
Google hand-waving about speed is self-serving. It’s about them making advertising money. Making the web a better place is a noble cause to get site owner buy in. Ads slow down websites the most and are unmanageable. So Google has us focus on other distracting trivialities.
Google never let speed influence page rankings more than 1 percent. In fact, it’s closer to 0.5 percent. But no one knows for sure because Google doesn’t publish this proprietary secret information. The idea that a “standard” exists influences myths regarding web speed. And pushes site owners to waste time attempting to achieve a Utopian ideal. It’s inefficient.
All online speed tests (Pingdom, GTMetrix, WebPagetest.org and many others) are interpreting Google criteria for goodness. It’s extreme or invalid to the point even Google can’t pass their own insane tests. This ivory-tower speed snobbery is out of control. In other words, it’s excessive and impractical.
The good news: scores are NOT used in the Google ranking algorithm. Only one speed parameter counts and that is Time to First Byte. TTFB is host-server dependent and beyond the control of website owners. They can only cherry-pick a better host with lower server overhead and change hosts. A good test for TTFB is ByteCheck.com.
Our best-practice suggestion is to take 6 consecutive TTFB readings and average the results. Assume your worst TTFB in milliseconds is more common than the average reflects. How does that affect your 2-second performance budget? Take 2000 milliseconds and subtract the worst-case TTFB. That’s how much time is left to load your best pages. Hurts doesn’t it.
Speed doesn’t have the same valuation on all pages. You don’t have to be under 2 seconds all the time. What? Not all pages have equal importance to visitors? There is the primacy effect or halo effects influencing how a site is perceived. First impression counts. After the first page experience, people are more forgiving if speed was initially exceptional. Your 10 most popular posts or pages need to be the fastest possible. Less trafficked pages can be a little slower. If you’re using WooCommerce cut yourself some slack. Relax the performance goal to 3 seconds.
Speed does NOT affect ranking immediately. It affects SEO over time. Speed improves the user experience (UX). That’s measured by user intent. That’s derived from metrics like dwell time, bounce rate, and return visitors. Google search then knows visitors are finding what they search for on your site.
A few articles help explain Google’s speed oddity:
“I’m still not happy. My PageSpeed Insights score is low.”
“Think-with-Google ranked my site as just average … above 2 seconds.”
Please don’t use Think-with-Google test:
Even Google.com scored average with this Think-With-Google test. Only 1 percent of websites have speeds of less than 1 second. Focus on website content instead. If you’ve made it under 1 second: Time to stop. Remember less than 1 percent of the Internet achieves this perfection.
Try not to use CDN band-aids for speed. If you are using free Cloudflare, it’s not good. It’s especially a waste of money if you’re paying. You can achieve the same results with free origin optimization. Avoid annual or monthly paid edge optimization.
On WebPagetest.org, my site’s TTFB scores a big red F. Does TTFB really matter?
Yes in a big way. You can’t get under 2 seconds load time if your TTFB is 1.5 to 2 seconds. TTFB is server speed overhead.
Mobile-first ranking is a different and undefined criterion. Its influence on your ranking is still a mysterious and long-term strategic plan by Google. Punitive shaming affects those not complying. Google experiments on the little guys first. They can’t afford to upset their biggest advertising client accounts with inferior rank changes. Google isn’t stupid – even when it appears they are.