serps

Google includes site speed as a ranking factor

Google has ended months of speculation by officially announcing that site speed is now being used as a ranking factor.

For some time now Google has been promoting a faster web. Going back to May ’09 Google co-founder Sergey Brin told the attendees at Google’s I/O developer event about wanting to speed up the web. This was soon followed by a Google blog post introducing the Page Speed application.  In the same month Google published the following blog post which focused on how Google has strived to make its services faster and has identified a variety of performance best practices, which Google shares with the web community at http://code.google.com/speed/.

So, later in the year it wasn’t a total surprise when Google’s Matt Cutts mentioned in an interview with Web Pro News that site speed would be an important factor for SEO in 2010.  This was followed by the addition of an experimental feature to Google Webmaster Tools called Site Performance in December 2009.

In the announcement on its blog post, Google has provided a list of applications that can help in evaluating the speed of a website as well as a link to its own Web Performance Best Practices page.

Google does, however, say that as a signal it does not carry as much weight as the relevance of the page and that "…the signal it only applies for visitors searching in English on Google.com at this point".  Google also say that it only affects fewer than 1% of search queries, and that it was launched a few weeks back. So if you haven’t seen any changes to your rankings, the site speed change probably didn’t impact your site.

Tags: , , , , , ,

0 comments Add This

Capitalisation in Search

Continuing our series of frequently asked questions, this article looks at capitalisation with regards to SEO, common problems, and how the search engines handle capitalised keywords.

The specific question we were asked was "What impact do the use of capitals have on search engine results pages (SERPs), if any?"

This particular question is often asked in relation to town and location names, such as the English town of Reading in Berkshire, which can be swapped with the word ‘reading’. I will address this specific question to begin with before taking a broader look at capitalisation in Search.

All of the major search engines are case insensitive.  That is to say whether you type [BOX], [Box] or [box] as a search query, it doesn’t matter, as you are more than likely to get the same results. So from an SEO point of view the best practice is to optimise your page so that it is grammatically correct, as you would any other typed document.  As we always recommend you should write for the user and not search engine spiders.

One place where letters written in different cases can be an issue is within URLs, which are in fact case sensitive according to the HTTP specifications. Case sensitivity affects everything after the domain, which is case insensitive, i.e. whether you have http://www.example-url.com/ or http://www.Example-Url.com/ doesn’t really matter, as this is only used by DNS to find the web server address.  What does matter is what you have after the domain, as different cases will indicate requests for different files. For example, http://www.example-url.com/Folder-Name/, http://www.example-url.com/FOLDER-NAME/ and http://www.example-url.com/folder-name/ are all different URLs and are treated as such by the search engines.

If all three versions of the above URL existed, it could lead to them being identified as duplicate content and there is a good chance that this will dilute the page’s link equity. For this reason, as well as to promote uniformity in order to make the process of creating URLs more straightforward, the recommended best practice here is to stick with lower case for all URLs. As an aside, lower case URLs are considered more aesthetically pleasing and are easier to read.

Case sensitive issues tend to arise if you use a server which is case insensitive, such as Microsoft IIS. With a Microsoft IIS server, the three URLs above would be treated as the same URL. Again the best practice here is to stick to using lower case in your URLs.

However, there are occasions when Google does return different results depending on the case used. This seems to be mainly where the letters could be either a word or an acronym. Compare [BAR],[bar] and [Bar] for example.  The results produced are split into three sections, and it is in the third section where we found differences.

Comparing search results of BAR, bar and Bar

Differences were also seen when comparing results for [AND],[and] and [And].

Another oddity that came to light was seen when searching for [MAD] and [mad]. For [MAD] Google returns a currency exchange rate one box but not for [mad].

Therefore a best practice for including acronyms on a page is to include the full form with the acronym in brackets, at least in the first mention, as Google often highlights this in the search snippet.

Tags: , , , , , , ,

4 comments Share

Bing announces updates to search results pages

Bing has announced a number of updates to its search engine at SES New York, including changes to the way that query intent is determined and catered for and the addition of FourSquare information via a new map application.

Microsoft announced a number of updates to its search engine, Bing, at the Search Engine Strategies (SES) conference held in New York this week. The Bing Search Blog also announced the updates, and included information as to why they were needed, summing up the impetus for the updates by saying:

“Our focus remains consistent – we want Bing to be the best place to help you make better decisions. When we launched last June, one of the ways we helped customers get to better decisions was by providing a more visual, more intuitive, and more organized experience. This spring we will continue this evolution by making some changes to our user experience and providing new experiences throughout search.”

The problem, as Bing sees it, is that users don’t always get the results that they want first time. Bing intends to address this by providing richer search results and analysing searches resulting in follow up searches, so that patterns can be identified and used to enhance the relevancy of search results to user intent:

"Our research showed that 42% of sessions require refinements, searching sessions are getting longer, and we see that many of those refinements happen when trying to complete common tasks."

Bing’s response to this is to move and expand its ‘Quick Tabs’ results, these are currently located on the left-hand side of the page, but will be moved to be at the top of the page, as shown in this screen shot.

 

Bing Decision Engine Update

 

Although not included in the image above, where Bing identifies (through anonymous user query analysis) a need, the results will also include more information in order to answer queries on the first search, such as answers to queries about celebrities and sports. Results will also include additional pieces of rich information, such as the current most ‘socially’ shared links within a site, as in this screen shot.

Bing Real Time 

 

In addition to this, Bing previewed an application for it’s mapping platform, which includes data from FourSquare, a popular location-based social networking site.

Bing FourSquare

 

Bing will also be adding a way to apply filters to search results based on various parameters. The Bing blog concludes by saying:

“You’ll see some of this roll out as we begin our testing, and there is more to come as we move into the spring. We are at the beginning of this journey, and are excited to see what you think. Thanks for using Bing for all your decisions, and keep the feedback coming.”

These changes are all part of the Bing strategy to position themselves as a ‘decision engine’, where people can find the information they need in a straightforward and intuitive manner by augmenting their search results with rich data that answers a query more intelligently than a more traditional search engine.

This is all positive from Bing, and we welcome these changes. Bing’s market share currently stands at 3% in the UK, so whether or not its efforts will result in an increase in market share, although they are taking steps in the right direction, remains to be seen. This, along with the Yahoo search deal now being in place, puts Bing in a strong position and makes them the one to watch.

Tags: , , ,

1 comments Share

SEO breadcrumbs for site hierarchies in Google

This latest LBi research explores Google site hierarchies and SEO best practice for implementing breadcrumbs in order for them to be picked up by Google.

Since November, when Google announced the arrival of ‘site hierarchies‘, we have been considering the best method of adding breadcrumbs to your site in order to be picked up by Google and added to its search results pages. The official advice provided in this Google Webmaster video for implementing breadcrumbs is somewhat vague, simply being to use a

“Set of delimited links on your site, that accurately reflect your site’s hierarchy.”

So we decided to investigate exactly what this means in terms of best practice for SEO, in order to provide the best chance of having your breadcrumbs picked up and used within Google search results. It is worth noting that Google indicates that this has not yet been rolled out for all sites:

“By analyzing site breadcrumbs, we’ve been able to improve the search snippet for a small percentage of search results, and we hope to expand in the future.”

Of course, whether having these site links is beneficial to any particular site needs to be addressed on a case-by-case basis. It will depend on whether a particular site is well suited to the use of breadcrumbs, and how your site uses top-level category pages within your site hierarchy.

A few notes on general best practice for breadcrumbs

Breadcrumbs are a secondary navigational feature used in combination with main navigation elements on large hierarchical sites. They provide a reference as to where a user is within a site’s hierarchy and help the user to quickly navigate to a higher level. This also provides the associated benefits of improved usability and reduced bounce rates.

Google’s definition of ‘delimited links that match your site’s hierarchy’ is a good description of breadcrumbs. The ‘delimiting’ character is usually a ‘greater than’ symbol, or ‘>’, which contributes to usability through recognition of uniform navigational elements. There should be no more than a few levels, and the link should start on the homepage and end on the current page. The last ‘breadcrumb’ should not be a link, as it would simply link to the current page and having a description of the current page written with no link provides additional navigational context.

The research

In order to investigate the use of breadcrumbs, we found existing examples of instances in which Google either has or has not picked up a set of delimited links and used them in its results pages. We selected 15 three word queries including keywords with navigational, informational and commercial intent, and analysed the top ten results in Google.co.uk, a total of 150 sites. We initially selected a larger sample, but retrieved enough data to make assertions sooner than expected.

Where to place breadcrumbs on the page

Breadcrumbs are usually seen at the top of the page just under the main navigation. In terms of the Document Object Model (DOM), you can expect to see breadcrumbs some way into the code. They are a secondary navigational element, included after the primary navigation and contained within the hierarchical environment of the DOM. With the addition of CSS, which can place elements anywhere on the page regardless of where they reside in the code, it is unsurprising that we found examples of breadcrumbs lower down the page, and even at the bottom of the page which triggered Google site hierarchies in search results.

However, we recommend placing breadcrumbs at the top of the page for best usability, as this is where most people will search for and recognise them.

Where to place the ‘delimiting’ characters?

Logic prevails here: the ‘delimiting’ characters should be placed in between the links. We saw a couple of sites which had either preceding or trailing delimiting characters, and these did not get picked up by Google.

Which ‘delimiting character’ should be used?

We wouldn’t advise against using any particular characters until further testing has been undertaken, although we found site hierarchy links in the search results for sites using ‘>’, ‘> >’, ‘›’ and ‘»’ characters, with white space and inline elements appearing to be inconsequential (see this example of bold tags on the home link).

The golden rule here is to make whatever is between the breadcrumb links identical, which was the case for 100% of sites which had breadcrumb links in Google results pages. Where this was not the case, breadcrumb links were not attributed to the page.

As a side note, we also found examples which did not use any characters at all, but used block level elements instead, such as the London School of Economics Website, which has a hierarchical structure and no breadcrumb links. Links in the main left hand navigation are separated via individual ‘<div>’s, which were picked up by Google and included in the results pages. This could of course be a false-positive, given that the ‘>’ character is included in the code between the links, but is unlikely.

Other tags which were found in breadcrumbs which were included by Google in the results pages included; ‘<li>’, ‘<div>’, ‘<td>’ and ‘<p>’. In each case, as with the LSE website, they were added between links in a uniform manner.

In summary, the most common method of indicating site hierarchy to Google is by separating links with a symbol such as ‘>’, ‘> >’, ‘›’ or ‘»’, but other methods involving block level elements may be used. This will need to be tested before we can recommend exactly which block level elements to use, and may prove useful where more advanced or decorative breadcrumbs are desired.

Can images be used to separate breadcrumbs?

From the sample we looked at, no images were found in the set of sites which did have Google site hierarchy links, but images were found in the set which did not, even though they had breadcrumb links. The same was the case where CSS was used to insert images as dividers.

As a best practice, we would advise against the use of images as breadcrumb link delimiters. However, we have noted that a site may be able to use images and be picked up by Google if images are contained within block-level elements or combined with relevant characters and appropriate replacement techniques. This is something which we will need to test.

Does a site need to be strictly hierarchical?

The short answer is no. Take this example from South West Water,
where the majority of the pages are numbered using a URL variable. The internal link structure and breadcrumbs are the only clues to the site’s hierarchy. Another example of site hierarchy links showing in Google even though the URL is not hierarchical is Yahoo! Shopping. You will need to follow the link to fully appreciate this. For many reasons, it is best to provide a hierarchical structure, but in a case where this is not so, breadcrumbs make an ideal addition to a site until such time as a redesign structural overhaul is feasible.

Should the first breadcrumb link to the homepage?

We found site hierarchy links in Google search results for pages both with and without breadcrumbs including links to the homepage. From a usability perspective, this is useful, in that where you have a large number of directories, missing the first few may still indicate site hierarchy for search. It is worth noting that the majority of sites do include a link to the homepage in breadcrumbs, and that we would recommend including this as it provides additional context to the hierarchical structure denoted by the breadcrumbs.

Should the last breadcrumb be a link?

We found examples of site hierarchy links in Google from both pages which did and pages which did not use the last breadcrumb to link to the current page.

It provides no benefit to the user if the last link points to the current page, and if the last breadcrumb is simply a description of the page with no link, it can only add to the relevance of the page and context of the breadcrumbs. It says “you are here”, which adds to the usability of the breadcrumbs.

Therefore our recommendation is to have a trailing breadcrumb for the existing page which is not a link. However, if it is, it will not stop you from gaining site hierarchy links in Google.

Can breadcrumbs be cross-domain or cross-subdomain?

Sites which we found with breadcrumbs which contained cross-subdomain links were not included in Google site hierarchies. Bearing in mind that the root domain is listed in the search result, this would render the hierarchy links inaccurate.

Semantically, what code should be used to mark breadcrumbs?

Although HTML lists lend themselves well to writing breadcrumbs, and were used in several examples of sites which were included in Google’s site hierarchy links, breadcrumbs are, by their hierarchical nature, not a list. HTML lists enable the use of customisable bullets to delimit links, but via CSS, which is not written to the page. Only where the list elements were separated by a delimiting character were breadcrumbs included in the search results.

It makes sense to contain the breadcrumbs within a block level element in order to distinguish them in the code, as well as to provide a clear signal to Google as to which content is actually your breadcrumb links. We would recommend using a ‘<div>’ element specifically for this purpose. The breadcrumb links should be placed inside the block level element and should then be separated by the same ‘delimiting’ character (most likely ‘>’). The same number of new lines should be used between each link (no new lines are required, but using one in between each link and each delimiting character will increase code readability).

Should you ‘label’ your breadcrumbs?

Often, breadcrumbs are labelled in the code as IDs, Classes or similar, with names such as ‘breadcrumbs’, ‘crumbs’ or ‘sitenav’. This is not required. We found examples of both labelled and non-labelled breadcrumbs which had been recognised by Google.

That said, there is no reason why you cannot label your breadcrumbs. A label in the code will make it more readable, and it won’t hurt to give an extra clue to search engines as to the content held within the containing element.

Semantically, an ID should be used over a Class, as an ID is a unique identifier to an element on the page, whereas a class can be used multiple times.

How long does it take for breadcrumbs to be found?

A ‘site:’ search in Google, followed by a restriction by date will show that pages are indexed faster than they have the breadcrumb links added. It was noted that this sometimes happens a handful of weeks apart, which suggests that it is a separate process which considers site hierarchy links outside of general indexing procedures.

Best practices for writing breadcrumb code

Based on our findings, an example of the ideal code layout for breadcrumbs is:

<div id="breadcrumbs">

<a href="http://www.cheese.com/">Home</a>

&gt;

<a href="http://www.cheese.com/soft-cheese/">Soft Cheese</a>

&gt;

<strong>Brie</strong>

</div>

A breakdown of the above code elements is as follows:

  • Hierarchical links which match the structure of the site
  • A link to the homepage
  • A reference to the page the site is on rather than a link
  • Contained within a ‘<div>’ element with an ID of ‘breadcrumbs’
  • Uniformly divided links using ‘>’ symbols as ‘delimiting characters’, written as ‘&gt;’
  • Single line break between lines (no line break is required, but adding them makes the code more readable. Remember that if you do add any to keep the number of lines consistent between links.)
  • As per our standard recommendations the inline tag <strong> has also been added to emphasise the keyword in the breadcrumb relating to the current page

A final thought 

Bearing in mind that Google has said it is still experimenting with site hierarchies and that some sites may not be suitable for site hierarchy links (think Wikipedia, or sites with minimal architecture), it does not mean that just because breadcrumb links are implemented a site will receive site hierarchy links in search results (in fact, sites which were setup by LBi to test site hierarchies are still not achieving site hierarchy links).

In addition, the code above is not definitive and, as stated in this article, inline elements such as ‘<b>’ and ‘<span>’ may be added without preventing Google from understanding your breadcrumbs. That said, unless abso
lutely necessary, this is probably best avoided.

For further information on Breadcrumbs, see this article on Smashing Magazine which provides some useful information outside of the scope of this article.

Tags: , , , ,

1 comments Share

Google testing site search boxes within SERPs.

For selected searches Google are inserting a site search input box within the SERPs.

This is only appearing for major organisations and even then, only those who have very distinct, unambiguous names.

Yesterday we started observing search results from Google which include a site search box within the Sitelinks under authority results. This is only showing up for search terms which are not at all generic, so whilst searches for big on-line names such as Amazon or eBay are not displaying the site search feature as yet, some smaller names, such as Argos or Woolworths are.

Google search result page for the keyword [Audi].

Appearing below the Sitelinks, a search here leads to a search results page, ostensibly for the query with an added site: operator, although the URL query string parameters of the SERP does specify that it is a navquery_searchbox.

Google SERP served from a search in the Sitelinks navquery searchbox.

Other example searches include:

Generally I consider Google’s experiments with their interface to increase clutter, but this may prove to be a truly useful addition to the SERPs.


UPDATE: Google have announced that the Google SiteLinks search box is now fully rolled out and that the selection of sites which feature this is algorithmically deduced from implicit user behaviour data.

1The Microsoft example is less surprising than most, with Google already providing a Microsoft specific search page.

Tags: ,

0 comments Share