Myth busting: delays in page loading

Someone asked about this, so here’s a quick one. It’s fine to run your website off your home DSL, as long as you configure your webserver correctly. This is an extension of the “virtual hosting vs. dedicated IP doesn’t matter” idea. As long as Google can load your web pages, it doesn’t really make a different whether the pages load in half a second or 5 seconds.

29 Responses to Myth busting: delays in page loading (Leave a comment)

  1. So what you are saying is that loading time of a webpage doens’t matter in search engine rankings?

    From what I’ve found is that running a website on a dedicated server tend to rank higher than a website running on a shared hosting with 5000 other websites.

  2. What’s the timeout on Googles client?

  3. LOL. I wish my pages would load in 7 seconds, let alone 5 or zero seconds!

  4. Hm, according to Alexa our website loads in 1.1seconds – as “Borat” would say, NICE…..VERY NICEE

    😉

  5. Matt,

    This is interesting news to say the least. I setup a website on a DSL connection using apache and regularly check the logs and Google visits every day.

    And your previous article on More webmaster console goodness http://www.mattcutts.com/blog/more-webmaster-console-goodness proves this, as the site in question shows a good level of downloading from Google.

    One concern I do have is serving pages up on a contended line though. Is this a no no? Someone I spoke to recently says that DSL web serving is great, but should only be done on a uncontended line?

    Thanks,
    Christian

  6. Dude, I’ve had perfectly well indexed pages run from a dial-up connection (we used a dynamic DNS service to maintain the DNS on the domain name), believe it or not (yes, I was the Queen of Minimalist Design :))

    My question is: you keep saying that virtual vs dedicated doesn’t matter (good!) as long as the host serves the website correctly.
    I don’t believe I have ever been hit by this issue, but am I correct in thinking that this is not specific to Google, that visitors to the site would also at times be getting ‘wrong site’ loads when they click through a link in these circumstances?
    The way you generally phrase this sounds like its only Google that would pick up this issue, but I’m sure that’s not what you mean?

  7. Sander, you’re confusing causation with correlation.

    It may be true that dedicated sites rank higher than virtual sites, but it’s probably a result of work put into them.

    Typically, sites that are hosted on dedicated servers have bigger budgets, get more time spent on them, and probably have more visitors (something had to justify spending on a dedicated server right?)

    Bigger sites are hosted on dedicated servers. Biggers sites rank better. But there’s no causation here.

  8. I may just be bringing up another myth to bust, but I think that the dedicated IP thing depends on your strategy. You want inbound links from different IP ranges. Now if you have 10 different satellite sites that you want to link to your main site, putting them all on a different dedicated IP would cause the links to pack more of a punch. Of course, then you’re going to be paying more for those other sites.

  9. There is no way for Google or anybody else to know for sure if your web site is on a DSL line, cable connection or in the data center(s) of your provider.

    You know, your provider knows. However, unless you understand the routing and ip address allocation in great detail you just cannot be sure when you look from outside the providers network.

    Some ip’s are portable between customer sites and data centers and some are not. ( because of routing, virtual networks and other unique to the provider ip trickery ).

  10. Hey Matt,

    I have heard from a few people that Adam Lasnik has said quite the opposite to what you have said here.

    Look forward to a bit of clarification from you 🙂

  11. It would be good to note that Matt’s comment is qualified with “doesn’t really make a different whether the pages load in half a second or 5 seconds”. So, if the site has heavy use (causing slowing page serves) or just a dsl line (causing slowing page serves) so as to slow page loading time to an extreme (over 5 seconds), Matt still leaves the door open for this to alter rankings.

    But Ryan does make a good point in that this could be just correlation, since slower loading could cause a higher bounce rate – again altering rankings. Never-the-less, I’d imagine Google would have a ‘hey – wait a minute, this page is taking to long to load – maybe this site sucks’ rating. Possibly gradually kicking in at 6 seconds 🙂 .

  12. My friend has a website hosted on his Newton from his cable connection and it is indexed. My website is hosted on a cable modem (don’t tell Comcast). However, my average page serving time for Googlebot has been going up because Googlebot has been requesting more and more processor-heavy pages (I can easily see this by looking at Sitemaps).

  13. Yeah cool, I know a guy who had a Quake server under his bed at college and it was sporting massive PR until he moved on to bigger and better thingers…

  14. Matt,

    Can you further clarify whether time spend downloading a page impacts the number of pages crawled per day (rather than SERP of the individual page)?

    For instance, if average load time is reduced from 3000 miliseconds to 500 miliseconds (through either on page and off page changes), would the number of pages Google crawls per day go up? Or is this inconsequential to number of pages indexed per day?

    Thanks

  15. Don’t get tangled in webs of cause and effect.

    A slow loading site not ranking well might not be the direct result of a search engine penalty. A site that loads slowly is going to be annoying to use. Less people will use it, less people will recommend it, less people will link to it, less directories will list it.

    Or we could go further and say that slow loading pages are likely run by less professional webmaster, and less professional webmasters are less likely to know about things like SEO.

    Or a slow loading page could be said to likely be less profitable than a fast loading page (or else, why wouldn’t they upgrade?) and so there is likely less money to spend on new content/products/features so the site is possibly of a lower quality, and so it may not get as much links or recognition.

    There are many reasons why slow loading sites could be ranked lower in search engines, the possibility of a direct penalty is only one of them, do not forget about the others.

  16. Pages load in half a second or 5 seconds!
    May be Google or other search engines does not take it seriously, but recent news by BBC affirms that the visitors are likely to leave a site if it takes more than four seconds to load. So webmaster needs to see it also from the customers point of view instead of just search engines.

  17. Well spoken Chris! There are many ways to Rome…

  18. does that mean the page size doesn’t affect the way SE crawls a site

  19. Are there any recent postings elsewhere that clarifies on the matter of page load time and whether it affects how your pages get ranked or crawled by SEs? I wasn’t able to find any postings by Matt on this matter solely.

  20. does page size calculate page load time?

  21. I have no reliable information on search engine ranking but I just wanted to indicate useful tools to check your Web pages load delay: curl (http://curl.haxx.se/) and echoping (http://echoping.sourceforge.net/) Both are free software.

    For instance, with curl, I can do:

    % curl –output /dev/null –write-out “Total time: %{time_total} s, DNS time %{time_namelookup} s, connect time %{time_connect} s, %{size_download} bytes transferredn” http://www.mattcutts.com/

    % Total % Received % Xferd Average Speed Time Time Time Current
    Dload Upload Total Spent Left Speed
    100 471 100 471 0 0 2382 0 –:–:– –:–:– –:–:– 0
    Total time: 0.198 s, DNS time 0.003 s, connect time 0.004 s, 471 bytes transferred

    And see that the network is fine and the download time reasonable.

    curl did not get images, so if images are an important part of the page, you may have to use another tricks (left as an exercice for the reader).

    Also, if you consider user experience and not only network time, very often, the time to display depends, not on the lot time, but of the display time (Javascript to run, complicated CSS, layout, etc).

    Disclaimer: I’m the author of echoping so the moderator is free to regard this as unsollicited advertisment 🙂

  22. When my page loading time was more than 5 sec, 90% of visitors used to leave beofore seeing my page – now that it’s about 3sec I can see about 50 % sticking around for longer
    I can’t stress anough on page loading speed: more important than any other seo practice

  23. Who exactly is jameno123? Why are links with my domain name and urls with my domain name showing “&username=jameno123” tagged onto the end of them. The Google sitemap errors shown with this tag at the end of urls are no longer a coincidence.

    Someone is doing this on purpose. Got the time to investigate? If my investigation has turned up the right person, I have no choice but to contact authortities, unless you can assure me in some fashion that he isn’t harming my business……it seems as though he is…..late September, all msn traffic gone…..a few weeks later, all yahoo traffic gone. If all google traffic disappears, his name makes headlines.

    I’ve been in contact with a .edu domain administrator who is willing to help since fake p*rn links to other sites were found within a wiki ( wiki.cs.uiuc.edu ) on their .edu domain. They’ve altered the page ( wiki.cs.uiuc.edu/CISTUG/&username%3Djameno123 ) since I notified them, after a google search, about this person. I think he lives in Amelia, OH….

  24. I don’t understand the people which host their sites at home with DSL. If they not want to pay the 10 bucks for a standard webhosting package they should close the project.

  25. I have a question related to this topic:

    Does the google bot parse just the HTML or actually load an image? Meaning, if my HTML page size is 50KB vs the actual page size with all the images and flash and what not is like 500KB, does the google bot load 500KB like a browser or just the 50KB?

    And how will this affect indexing?

    Appreciate a response.
    Cheers.

  26. If I’m a Google search engineer I sure would take into consideration of the average loading time (or average loading Kbps) of a website. Because it’s a big part of the end user experience.

    I would rank fast sites over slow sites for the sake of good user exeprience to the Google users. Though the ranking nuances might be slim but there it should be in my opinion.

    So are you saying, site A that’s constantly 5 times slower than site B will be no inferior to site B in Google search rankings at all?

    With due respect, I don’t know if you have close contact to the technical staff of Google (or you are one of them?), or maybe you are just busting the myth by taking the myth automatically incorrect, please if you can, can you confirm this with someone who’s directly responsible for the ranking algorithms.

  27. I take it things have moved on a little since you wrote this article originally?

    At the Pubcon conference in Las Vegas (Nov ’09) you are reported to have indicated that good loading times may positively influence a site’s SERPs (though the opposite does not hold true).

    Source: http://www.seomoz.org/blog/new-interesting-insights-into-google-rankings-spam-from-pubcon – second bullet point.

    It would be great if you could confirm that here for anyone stumbling over an olb blog entry and mistaking it for current advice.

    Cheers

  28. Hey Matt, I read a Blog piece this morning that you indicated in an interview that the speed in which a page loads may soon be a factor in the search engine rankings. I am not sure how accurate that statement is and I am not a techie by any stretch of the imagination, but I do have concerns if Google starts to move away from content for ranking toward cosmetics. One of the reasons I use Google almost exclusively is because I know I will get the most relevant results without having to dig through pages of results. If you start to bury better content because Joe SEO knows how to make his page load faster, I am afraid you will sacrifice what makes Google stand out above everyone else.

    Just my two-cents worth.

  29. Hey Matt, So what you are saying is site performance in itself isn’t a metric Google uses to rank a website. But, bounce rate as it relates to slow page load does effect page rank?
    Thanks!

css.php