Does PageSpeed Actually Impact SEO? [New Experiment]

Does page loading speed actually affect Google rankings? I decided to find out.

And the results will probably surprise you.

First, a little bit of background. Google has stated several times on the record that site loading speed impacts SEO. The slower your site is, in general, the worse it will rank in the search results.

But they’ve never said how important speed actually is. Is it a Google ranking factor that’s a key part of the algorithm? Or a minor thing that won’t make much of a difference in practice.

Well, I decided to find out. Specifically, I ran a little SEO experiment. This experiment took one page that loaded SUPER slowly according to Google PageSpeed Insights… and sped it up significantly.

Then, I tracked the rankings and organic traffic to that page.

That way, we could control speed as variable (as much as you can control any variable in an SEO experiment!).

The results were… interesting to say the least.

SUBSCRIBE to get higher rankings and more traffic!

Follow me on Twitter!

[BrianDean]

Have you seen PageSpeed make a difference in your rankings? Let me know in the comments!

[unmiss-com]

Everything depends on many factors.
If you run an experiment with one page - many circumstances play significant roles than simple ones.
I increased the PageSpeed Insights parameter for the whole website without decreasing UX - the traffic has increased significantly.
I always create content first for people, then jump with optimization without changing design.
It helps.

[aloeveraforever]

Thank you Brian, that will improve significantly my SLEEPING QUALITY because I tried really hard to speed up my pages but with so little success.
Greetings from Italy, Martina

[wojtekszulimowski]

Ha last time I was checking site speed of your pages and conclusion was if backlinko can have 28, then site speed doesn't matter. Now you confirmed that. Thanks.

[TravelTipsbyLaurie]

“There’s no such thing as a free lunch.” So true-just one change will not propel us. Thanks for your time with these vids!!

[miklosmayerphoto]

Great news Brian! And thanks for putting all this crazy effort into it (I would've never gone that far :))
I have the same experience, although I haven't gone this far checking it. I have an image-heavy post that ranks in first 3, and it did equally well when it had half as many photos.

[MikeKwal]

Hey dude! Awesome video! Your engagement is off the chart. Love the energy. I work with local service businesses, and my results clash with your experiment. When we furnish faster loading content with customers on their old websites, we up-rank on Google. Maybe it's our content game, or maybe it's because local customer's are not on the first page of Google. Our PageSpeed makes a difference with websites that rank in the keyword abyss on page 10 of Google, where nobody shops. Hope that helps, and again, really love the high energy in the video.

[ContentCreatorPlus]

You're videos are a goldmine.
Out of all the misleading information you find everywhere on the internet, your videos are just pure gold.
Thank you so much for teaching me so valuable information. Support will be Grateful!

[rustyglaze]

One thing missing from the experiment is testing the speed of the pages ranking above the page.


There are also differences in user intent if I am going to a blog site to get information - I would be more willing to wait to read the post, so a slow load time would not deter me as easily. If I am on an e-commerce site shopping (moving pages quickly and often as I shop) a slow time would be a killer and I would leave. So the type of website should match the user intent in terms of speed.

[WildBoys]

We love that you run your own experiments to test theories! It helps so much! Thanks again! -EliWild1 and LukeWild2

[nikilsingh]

I am Brian Dean: founder of the Backlinko.



I love this line.

[bluelinemorphs]

I just started my channel a little over a week ago and I’ve been following your channel to help it grow - and simply put it works !

[toothlessdentist]

The correct way to conduct an SEO experiment:

1) Set up two similar pages that you have control over

2) Change only one thing,

3) Compare how the two pages rank relative to each other regardless of the rest of the web.

4) To speed things up, make the two pages rank for a unique word of your choice (example: iuewibewfewfiuiebiewbfe).

5) Also to speed things up, search with exact phrase just to make sure that only your pages show in the results.

6) Also, make sure to augment the difference whatever it is that you are testing. For example, if you are testing for page speed, make one page super slow and the other super fast. Then reverse the changes to confirm that what you did actually caused the rankings to change.

[ThrottleThrill]

Brian the page is at #5 now.
We rebuilt a WP website from ground up with SPEED as the only focus of the rebuild. The website is in the medical space in the US. Within a week of relaunching the website we went to #1 on all 5-7 major keywords that were previously ranking at a much lower position (7/8 on average). We overran all other players with authority and age and backlinks. Important to state that we went from being 3x slower than the top competitor to be being 3x faster with the rebuild — a 6x change overall and it paid off.

[jakeballinger]

This is a great video. Really appreciate how you acknowledged some of the limitations there—I’d love to replicate this for a less competitive keyword on a less competitive website.

4 questions for you:

1. Were the increases in pages per session and session duration statistically significant?

2. Why did you remove the images instead of lazy loading them? (Or were they already lazy loaded?)

3. Was 2 weeks an arbitrary length of time you chose? Was there some reason behind it?

4. I like your explanation for the increased bounce rate. Balancing UX and SEO has always been tricky for us, especially since I work at a web design agency. But since we assume bounce rate is a factor, is it possible that the increase in bounce rate was enough to offset the page speed gain (if any)?

[MarkBadran]

I've long-suspected this to be the case based on informal observation with our own websites and several clients. It helps to see others like yourself giving it a more formal test ... I can now confidently avoid wasting any more time tweaking pages based on lackluster Google PagesSpeed reports. Thanks Brian!

[superfastpress]

1st: it's not the same page without images nor is it necessary to pull them out. Lazy loading is the solution here. 2nd: you DO NOT optimize for Google (but you do obviously) but you make a site or page faster to serve your audience better and to avoid wasting their life time. But finanlly to roll back to the damn slow version makes clear that you care more about Google than your visitors. AND: Those Google SERPS are not static. They constantly improve. So good luck with that approach.

[membershipmovement]

I've been thinking about your test and I don't think the "slimmed down" version of the page was an "apples to apples" test.
While the text content didn't change, the visual experience changed drastically. You said in the video that maybe visitors found what they wanted quicker and thus, affected the bounce rate. Another effect could be the lack of visuals. So I think it is hard to measure if the effect of the page speed in regards to the bounce rate.

[itsinfactagift]

That was surprising. Good job as allways brian and team.

I really hope you redo this experiment on a lower keyword competition to se if this is really impacting the seo positioning and if it will be worth it by the way.

Cheers!

[ABeardedDad]

Nice high quality video Brian, and a surprising result.


A few questions came to mind while watching.


- did take into account any confounding effects like removig images (assuming good low size images are a potential ranking factor)
- what about time? Maybe two weeks simply isn't long enough
- it seems as if page speed may be a proxy for other relevant ranking factors. You mentioned bounce rate, and other user behaviour. Perhaps page loading speed has more of an influence on UX, which in turn may affect your rankings. if page speed improves UX it might require more than two weeks to examine the spill over effects?


Overall it's good information to know though. Losing sleep over page speed is a bad idea, especially if it comes at the expence of the overall quality of your post.


Thanks for doing this work Brian, and for sharing it!