The effects of Google's page load time factor?

Forum Moderators: Robert Charlton & goodroi

Message Too Old, No Replies

The effects of Google's page load time factor?

dickbaker

11:02 pm on Jan 18, 2011 (gmt 0)

For several months it's seemed as though there was a ceiling on the number of visitors my site was getting and the number of page views. I'd get 27,000 or 28,000 page views a day, but never broke the 30,000 number to get to where I was last year. Visitor counts also seem to have hit a ceiling.

Back in August an advertiser of mine had me put an OTP ad on the site. If floated an animated ad with a talking sponsor for about a minute. It showed just once for a new visitor, and had a cookie set for a couple of weeks.

It definitely affected load time, and I have no doubt affected the number of page views.

The ad was dropped on December 31st and a week later I saw visitor count and page views rise. The rise over the past several days has been dramatic. The graph looks great.

Some of this I attribute to user behavior, but there's also signs that Google's page speed calculation was a factor.

Just an FYI.

tedster

12:14 am on Jan 19, 2011 (gmt 0)

Thanks for that - very interesting. If indeed it was the "talking sponsor" ad, that would be a shame (except that I personally hate the things). It seems there should be a way to program that kind of advertising so that page registers as loaded before the ad even starts - assuming of course that the page is still fully functional during the ad.

dickbaker

3:48 am on Jan 19, 2011 (gmt 0)

The page was loaded before the ad starts. I don't think Google sees it that way, though. I could be wrong, but the change in visitors and page views is awfully dramatic to be a coincidence.

rowtc2

4:05 am on Jan 19, 2011 (gmt 0)

This thing i was experiencing.

I've got a ceiling in visits and after change of PHP code site was running faster and visits has increased. I could't bet it was a signal that load speed can influence, but keeped in mind.

Now i am at another ceiling, i have a script which search for 4 related articles in entire database and displays them at the end of the page (and source code). I will not risk to eliminate those related articles now, i think i will move the images on a CDN.

Product page has in left a list with all categories, 20 other products from the same category listed under article and in right side.

Waht do you think, removing these 4 related articles/posts (link +snippet) is a risk?

tedster

4:43 am on Jan 19, 2011 (gmt 0)

Isn't there a way for you not to perform the search every time? Similar to what I said about the ad on dickbaker's page - it seems like the source should be able to trigger "page is loaded" before the ad starts doing its dance.

I've seen a similar improvement in all kinds of stats (sometimes including ranking) by using a caching plug-in for Wordpress. It's like you have everything ready to go for visitors instead of running around grabbing things every time someone does show up.

indyank

12:56 pm on Jan 19, 2011 (gmt 0)

Is it permissible to show some scripts (take for example the FB scripts) to only the visitors and not to show it for googlebot? I was thinking of doing this but not sure how acceptable it is..

tedster

1:44 pm on Jan 19, 2011 (gmt 0)

Page load time is mostly NOT determined by googlebot - the core data comes from real visits by browsers using the Google toolbar.

xnavigator

1:57 pm on Jan 19, 2011 (gmt 0)

i think this facotr isn't included in rankign.. maybe if ur site takes more than 10 seconds google will penalize it but if stay under a resonable time dont' worry..

consider only that on my website the main facotr of load time is the slowness of adsense.. pages without adsense are loaded in less than 1sec, with adsense over 3 sec

rowtc2

6:51 pm on Jan 19, 2011 (gmt 0)

I've seen a similar improvement in all kinds of stats (sometimes including ranking) by using a caching plug-in for Wordpress.

Thank you tedster for the idea! I will upload these related articles on a separate table in database and list them from there.Will be better for sure.

Page load time is mostly NOT determined by googlebot - the core data comes from real visits by browsers using the Google toolbar.

Speaking about page load time as a Google ranking factor, i think is normal to be determined by spider, not Google toolbar.

tedster

7:09 pm on Jan 19, 2011 (gmt 0)

A spider does not render a page visually, so load time MUST use other methods than the server's response to googlebot.

Simsi

7:44 pm on Jan 19, 2011 (gmt 0)

I run all my ads through a custom script running in 468x60 or 120x600 iframes, on the basis that I assume this is then deemed a seperate entity to the main page when load times are calculated. Not sure if that's worth considering or not but I've never seen any ranking fluctuations that I can attribute to ad load times, even on video ads.

xnavigator

8:42 pm on Jan 19, 2011 (gmt 0)

iframes time load counts in the main load time i guess.

the only thing that doenst' affect page load time is loading the ads after the window.onload (with dom appendChild)

Simsi

8:51 pm on Jan 19, 2011 (gmt 0)

iframes time load counts in the main load time i guess

I'd guess its not myself but not sure how it is proved either way.

azn romeo 4u

6:22 pm on Jan 20, 2011 (gmt 0)

Question,

Does google pagespeed and yslow take into account stuff that it self provides? Like how adsense, analytic slows down the website.

If I remove all the ads, and javascript provided by third parties, I get a score of 95/100 from yslow and really good pagespeed as well.

xnavigator

8:34 pm on Jan 20, 2011 (gmt 0)

as I said above adsense are the main parts of mu load time. they take about 2seconds to load..

so except ur website doesnt' load in more than 10-15 seconds imporving page speed imo isn't so worthy

misterjinx

9:02 pm on Jan 20, 2011 (gmt 0)

Hi tedster,

we know that a crawler doesnt' visually render a page.

Google in webmaster tools experimental voice about load time declares to use at least 200 nodes from its toolbar.

But Google has also Chrome.
And Chrome can render web pages from a client side perspective, send back informations about URLs, page load time and other informations.

Informations that Google could elaborate (in aggregate form) ...

tedster

9:09 pm on Jan 20, 2011 (gmt 0)

Informations that Google could elaborate (in aggregate form)

Yes. Google hasn't told us their "formula", but they have said enough that we can assume they are aggregating data from several sources.

koan

10:23 pm on Jan 20, 2011 (gmt 0)

Page load time is mostly NOT determined by googlebot - the core data comes from real visits by browsers using the Google toolbar.

This has been confirmed to me while using the Google Webmaster Tools, which gave me examples of speed for non public, protected pages, and I have the toolbar installed.

dickbaker

10:54 pm on Jan 20, 2011 (gmt 0)

Well, my site is faster, but the traffic has gone back to the levels it was at before the spike.

I didn't think of it at the time, but there was probably an uptick in traffic to my site due to a major news event that involved widgets in my niche.

Back to figuring out how to get back to year-ago traffic levels. :(

rogerd

2:37 pm on Jan 21, 2011 (gmt 0)

Whatever the ranking effects of page speed, the effects on user behavior are well established. And, with the visibility that Google is giving page speed, you can be sure that sites other than yours will be getting faster.

I'd say forget ranking effects, but work on fast-loading pages purely from a user standpoint. On multiple sites, I've implemented more server power, server-side and browser side caching, and a CDN for stuff like images and script files.

ponyboy96

3:05 pm on Jan 21, 2011 (gmt 0)

I don't think it matters that much for rankings. It's really a go/no go situation. Just because your site is faster than everyone else doesn't mean you will be #1.

However, a faster page and site is better for users and helps conversion.

dickbaker

5:41 pm on Jan 21, 2011 (gmt 0)

I don't think it matters that much for rankings. It's really a go/no go situation. Just because your site is faster than everyone else doesn't mean you will be #1.

However, a faster page and site is better for users and helps conversion.

I've found that to be the case in my niche. Most of the top-ranking sites are ecommerce sites with a lot of images per page, and are very slow loading. The page speed analyzer puts them in the 70's, while my pages are in the 80's.

ashear

1:39 am on Feb 2, 2011 (gmt 0)

I have personally been testing and tracking this for the past 5 years and can tell you without a doubt that server responsiveness, DNS configuration and page load time are extremely important to Google natural rankings.

If your page load time exceeds <4 seconds within the Google Webmaster Tools you can safely bet that you are being scored differently than a site with an average of less than >4 seconds.

If you are a blog, I highly suggest installing W3 Total Cache on Wordpress. It has some amazing speed settings that will make a huge difference. As well as CDN connection capabilities.

koan

4:02 am on Feb 2, 2011 (gmt 0)

Thanks I'll give W3 Total Cache a whirl, I try not to use too many WP plugins, but finding a good cache plugin was on my to-do list.

ashear

6:54 am on Feb 2, 2011 (gmt 0)

My test blogs, with a CDN dropped from 9 seconds to 1.7 in Google Webmaster Tools by just installing that plugin.

trinorthlighting

4:40 pm on Feb 2, 2011 (gmt 0)

I just took all my pages from the 70's to high 90's this past week and Google is crawling the sites like crazy. I am watching to see if there is a bump in the serps but I have not seen it really. I will keep you guys posted.