Archive for the ‘SEO Testing’ Category
Over the last week I have been making some changes to by blog, these changes were made for three reasons – to decrease indexation times, increase crawler activity and hopefully increase rankings and traffic over time. Sadly I cannot publish any results because as usual I have managed to get a -30 Page penalty on a few keywords during a test which was not related to my new changes (The Keyword spam which I tweeted about a few days ago – so instead of posting results I would like to go through the changes that I’ve made then do a follow-up post on the results.
Plug-ins Installed
Google XML Sitemap
Does exactly what you think it does, auto generates a new XML sitemap and pings search engines with the updated version.
Majestic SEO Dashboard Widget
This is a great plug-in from Dan Taylor that displays Majestic’s link growth graphs within your WordPress Dashboard, giving you a good idea of how your blog is growing over time without having to leave your website – this is something which will be important for me during testing the new setup 
nStatistics
If you want a plug-in which will display crawler activity on your blog then look no further, install this plug-in and get pretty graphs displaying both user and crawler activity over time
RSS Footer
I have been doing a bit of work with RSS directories recently and wanted a way to put static links at the bottom of every post in my RSS feed – job done
Feed2tweet
Just your every day auto Tweeter for your new posts.
Tweet old post
Ok so everyone has been wondering what plug-in Michael Grey has been using to display his archived posts, it’s just most people couldn’t be bothered to search for this obvious plug-in (says allot for the SEO community).
WP OnlyWire Auto Poster
Most SEO’s have an OnlyWire account for getting those bookmarks quickly – now you can auto post to OnlyWire via this plug-in every time you publish a blog post.
I’ve Done The Plug-ins Now The Fun Bit
As some people know I have been setting up quite a few profiles recently, most of which have ways to aggregate either your Twitter feed or your blog RSS feed, some of which will even accept your social bookmarking accounts (Example Here) and then give you an additional RSS feed output – which ten leads to some fun with RSS directories and so on.
With this in mind and the fact I’m auto posting links to OnlyWire (which sends out to quite a few places) you can imagine the 1000’s of links that could instantly be built to your new blog post as soon as you hit that publish button .
Now The Boring Bit
Now because I have just suggested a way to fully automate link building for a WordPress blog I thought I should put in a little disclaimer – In no way am I saying go out and do this, as most people know I test things with this blog and sometimes I will blog about these tests, I have not fully tested the above and would never use this technique for any clients. There are people who do this kind of thing already, out of which some make mistakes and get penalties, so be warned!
2 months ago today Google Caffeine went live, it promised live indexing which would lead to fresher results, sadly between 25th July right up to now Google’s crawling seems to be at snail pace, for example before and just after caffeine any blog post I published took around 10 minutes to be cached and indexed.
The most recent pages/post published took just a little longer, up to 5 days to be indexed. Why would a search engine which promotes live indexing take 5 days to cache and index a blog post or page of content, when the site-map has been updated and submitted and Google has been pinged the URL? personally I think the servers are under strain with an unexpected amount of pages being published and a heck of allot of crawl errors slowing down the bot.
I make these assumptions because many people will be redeveloping their websites due to everyone wanting to use HTML5 , because of this many people will just redo the content, change the URL structure (maybe mess up 301′s along the way). Anyway that’s just what I think, if you have any opinions please post a comment 
P.S Im going to track how long this post takes to be indexed!
For a long time now the majority of SEO’s have been led to believe that Google Suggests is based on keyword search volume. Because of this many try to manipulate it to suggest the keywords they want and to be honest I’m really tired of these guys messing up the search volumes on keyword data.
Google has stated that Google Suggests is based on keyword popularity, this term can be easily read as search volume but in reality Google are not stupid, they don’t want people manipulating data easily so they use an algorithm for it. The algorithm itself is a ranking algorithm, possibly similar to that of the organic search algorithm.
Through testing I have seen that the algorithm uses 3 factors at least when ranking each keyword, these are Content, Anchor Text and Search Volume. Search Volume seems to play the smallest role within this algorithm with mentions of the keyword within content and anchor text playing a much larger role.
The problem with manipulating the data in Google Suggests this far on is that due to the mass volume of content on the web, adding more content with new keywords is not going to make too much difference on the money making keywords. Another reason why the method is so effective is because the long tail versions of keywords contain the short tail keywords so long tail shouldn’t outrank its partnering short tail keyword (some keywords have exceptions), which gives the illusion of search volume playing a key factor.
At the end of the day Google Suggests as mentioned above has got to the stage where the majority of it cannot be messed with so please stop ruining my search volumes (more than Google already does).
P.S If you have seen any additional ranking factors for Google Suggests please let me know.
Over the last 2 weeks I have been doing tests on pages that have the noindex nofollow meta attribute, in theory Google should ignore these pages but the truth is actually a little different.
My recent obsession with this type of page started two weeks ago when I noticed that a links page on one of my blogs (which was set to noindex nofollow) was publicly cached by Google, which I thought was a bit strange. After around 5 minutes of being deep in thought about this discovery I thought I would run a test to see how Google treats the links from these pages.
Since I knew the page on my blog was being cached I added a link which contained unique anchor text pointing towards one of my test blogs, which also contained mentions of similar text on the page. The next step was to get the page reached by Google which was sadly quite a difficult task.
To get the page re-cached I had to use a combination few social bookmarks and a few blog comments then wait a week for the Google bot to find these links and follow them, luckily that did the job. Once the page was cached I typed my unique key phrase into Google to find that my test blog was the first result.
Just to make sure this wasn’t a one off I had a look through the link data on web master tools for a few of my other blogs, to my supprise there were around 10 links listed which were from noindex no follow pages.
Looking at the test data so far it would seem that Google has devalued the strength of the noindex nofollow attribute due to overuse of web masters, just like what happened to the nofollow attribute. I am still running tests on this to try and get clear proof that Google is now counting links from these types of pages so watch this space for further updates.
Over recent months I have been developing a process for website speed tests, the reason for this is that sometime in the near future Google may implement website speed into its growing list of quality signals, if it hasn’t already that is.
If and when this is implemented, fast loading websites will start to be favoured upon more, affecting rankings across the board. You may ask why Google is adding this as a quality signal; there are three reasons for this:
- Google’s servers are feeling the strain of crawling the web, fast loading website will also be quicker to crawl and index, basically you scratch Google’s back they will scratch yours.
- Quality results in the eyes of users – Users don’t like waiting for websites to load, the slower the website the less chance of click through.
- Less spam in the listings – Spam websites are usually a lot slower than say a corporate website, less time is taken to develop them so adding this could remove some of the spam results from the listings.
Reasons for a Slow Website
There are many reasons for a website having performance related issues, the most common are:
- Large amounts of CSS and JS files – the more files there are the more server requests are made and the slower the website becomes so it is recommended that these files are put together forming no more than 3 of each.
- Poorly Optimised CSS and JS files – Once a website is ready to go live, any unneeded whitespace should be removed, this will lower the file size and reduce the load time of the files
- Poorly Optimised Images – Most images can have their file size cut in half by simply changing the image type, reducing the quality can lower the file size further without any noticeable quality difference.
- Too many resources being served from the same domain – Browsers can only download so many resources from the same domain at the same time; using external domains for some resources will allow them to be downloaded in parallel. This is also known as parallelizing downloads across hostnames.
Benchmarks
The best place to start with any process such as this is benchmarks, taking note of how the website compares to the top 5 competitors will give you the goal of beating them, achieving this can give a good sense of accomplishment .
There are a few page load time tests available on the internet, but none are 100% accurate. I suggest using the test at iWeb Tools, this lets you run a page load test on up to 10 websites at the same time, but as this is not accurate it is recommended that you run the test a few times and record the average.
The numbers from the IWeb test is only for the loading of the html code, this does not include external files. To benchmark the actual browser render time of the website I recommend installing the Firebug add-on for Firefox. The Net tab of this add-on will record the load time for each resource a page requests and also gives a total load time, this is good information and gives you a good insight into where the website needs improved before using any additional tools.
Optimising a Website for Speed
When it comes to optimising a website for speed there are two steps, first you should have a look at the way in which the website is coded, there are more than one way to code a website and some techniques are better than others.
Inspect the code and see what you can refine, chances are that a bulk of code could be replaced by more refined, light-weight code. Swapping JavaScript for jQuery where possible and removing comments from code are good examples of improvements you could make that would benefit the website.
The next port of call should be to install the PageSpeed Firebug add-on, this tool was released by Google to help webmasters find problems that affect website speed. When the tool is run it will scan the page you are viewing, analyzing resources such JavaScript, CSS, Images and cookies, looking for anything that could be better optimised.
After the scan it will list any potential problems, down to singular images and suggest how best to improve them. The tool will even optimise most of the resources, making them available to download so the majority of the work is already done.
If you use files provided by PageSpeed it is important to test these files offline first to make sure there are no noticeable changes in the website, but so far the tool has provided error free, fully optimised files in tests on 5 different websites.
Conclusion
Even if Google decide not to introduce website seed as a quality signal these improvements should still be made for users alone as this could provide lower bounce rates, greater click through and possibly better revenue.
