Sometimes when you search at one of the major search engines, you’ll see an extra set of links showing up under one of the listings in those search results. Referred to as either quicklinks or site links, most often those will show up for the listing at the top of the search results like in the following image:
Sometimes, those extra links will also appear for pages listed a little futher down in search results as well. A lot of questions have been raised about how those search engines decide which pages to show as quicklinks or site links, and the reasons why. I’ve written a number of posts about whitepapers and patent filings from the search engines that have provided some clues to answer those questions, and there’s a list of links to those posts at the bottom of this post.
But, another mystery surrounds those quicklinks or site links, which is how search engines might decide upon the text used in those links. I haven’t seen an answer from any of the search engines previously. At least, until now.
Yahoo published a patent application that provides some hints on their approach to deciding which link text to use in their quicklinks.
The patent application is:
Generating Succinct Titles for Web URLs
Invented by Shanmugasundaram Ravikumar, Deepayan Chakrabarti, and Kunal Punera
Assigned to Yahoo!, Inc.
US Patent Application 20100049709
Published February 25, 2010
Filed: August 19, 2008
Methods, computer programs, and systems for generating a link title for a URL (Uniform Resource Locator) within a context webpage to be shown as a web result are provided. The method evaluates generation parameters for a plurality of sources for picking words from the link title. Further, the method generates candidates for the link title, and a likelihood is computed for each candidate.
When computing the likelihood, the generation parameters, the context webpage and the words are considered. In addition, the method selects a candidate with the highest likelihood from all the computed likelihoods, and presents the URL with the selected candidate as the title.
I usually like to take a patent filing like this, and try to break it down into plain English, cutting out a lot of the legal language and the math presented to give a summary of the ideas behind an approach. Fortunately, Yahoo also published a whitepaper and a powerpoint presentation on the topic, which provide a look at the ideas in the patent filing in plainer language.
The whitepaper was published for KDD 2008 – Knowledge Discovery and Data Mining Conference, held in Las Vegas, Nevada, in August of 2008, around the same time that the patent was filed, and shares the same name – Generating Succinct Titles for Web URLs (pdf).
The powerpoint presentation (of the same name) presents the ideas in the patent filing in even plainer English. See: Generating Succinct Titles for Web URLs (ppt). Note, if you don’t have powerpoint, and you want to see the presentation, consider downloading the free Open Office open source productivity suite, which can show powerpoint presentations.
While people creating web sites often spend a fair amount of time deciding upon titles for pages they create that describe the content of those pages, as well as anchor text for links pointing to those pages, there are sometimes problems with those choices of text that make them less than idea for use as the text for quicklinks.
One of those is that a search engine often wants to use a limited number of words in their quicklinks, and page titles for pages pointed to by quicklinks tend to be longer than what the search engines want to use in their links.
Another problem is that, according to the patent filing’s authors, around 17 percent of the pages they run across on the web are missing page titles.
Yahoo might look at a number of different choices to select text from to use in quicklinks. Some of the choices can include:
- Anchor text pointing to the page from links on the same site
- Anchor text pointing to the page from links from other sites
- Search queries for which the page was returned in the top 10 results
- Search queries for which the page was returned as a result
- Search queries for which the page was the first result
- Search queries for which the page was clicked upon from search results
- Words extracted as key phrases from the page
- Tags for for the page from delicious.com
The patent application describes some of the potential problems that might be found in some of these choices. For example, they tell us that search queries used to find some pages might contain misspellings. They also tell us that those misspellings don’t appear as often in search queries that are clicked upon.
You may have noticed that they don’t include the titles of pages as possible link text in quicklinks. Their reasons for not using page titles:
1. An analysis of one million random URLs they performed showed that at least 17 percent of HTML documents lack titles.
2. Page titles are often erroneous, incomplete, long, or simply not the best title.
They provide an example of a bad title on the SIGIR conference web site at “www.sigir2008.org/schedule.html” which used the title “SIGIR’08-Singapore,” because a much better choice for a quicklink would be “Conference Schedule,” which isn’t very clear from that page title.
We’re also told that long quick links provide a bad user experience on a search results page, because of limited real estate on search results pages.
It’s worth spending some time on the presentation, paper, and patent application if you’re interested in learning more about how Yahoo might decide between the different choices they have when choosing the text for a quicklink.
If you publish a web site, and you won’t to understand more about quicklinks or site links, as well as when and why search engines might show them, some of my previous posts on those kinds of links might answer some questions you may have:
- Have You Ever Seen Delicious Quick Links?
- Microsoft on Navigational Queries and Best Match
- Should Webmasters Pick Their Own Quicklinks in Search Results?
- Yahoo Site Links: Quicklinks for Navigational Queries
- Microsoft Study Takes Navigational Sitelinks a Step Further
- Google’s Listings of Internal Site Links for Top Search Results
35 thoughts on “How a Search Engine Might Choose Text for Quicklinks or Site links”
I observed these quick links are displayed for only highly renowned websites that provides tools or products! However such results are displayed only to the core keywords. As most of the websites cannot be displayed for the core keywords and as the keyword strategy doesn’t work there, I believe there isn’t any benefit considering these results! What say?
I agree with Australia SEO. Most sites will only have one listing and that is it. For a site to get multiple links, such as WordPress homepage, you need an awful lot of links. An awful lot. I just don’t think this is reasonable for non-large scale companies not willing to spend a vast amount of resources on buying links. What I like to aim for is the double link. Two relevant pages on one site showing up in search results and also spreading out of your products on multiple sites. The spreading out allows you to possibly rank high through the use of another’s site PR. Just a couple quick thoughts from SEOs working with the smaller fish.
Hi Australia SEO and Sean,
Those are good questions.
I’ve seen sitelinks for smaller, and less well-known sites that don’t provide tools or products, and that don’t have “an awful lot of links.” And that haven’t engaged in buying links. I’ve worked with a few small businesses that started having sitelinks appear for their businesses once issues like duplicate content and impediments to search engine crawling programs were removed from their sites.
I’ve also seen sitelinks appear for terms other than “core” keywords.
It’s not a question of the search engines favoring large brands or well-linked sites, but rather of those search engines being able to recognize that a specific query is likely navigational in nature, and that sitelinks would be helpful to people using that query term.
You don’t need to have a nationally or internationally known brand to have sitelinks show up for your pages.
Thanks for posting. Sitelinks have been such a mystery! There is a way to block certain sitelinks (who would want to do that)? But there’s no way to suggest different ones – bummer! Thanks for providing some insight.
When I come upon sites with sitelinks, it seems that they typically point to those pages that already have some link love themselves from within the domain. That said, I never thought about how the link text was determined until now. It seems that most of them are pretty accurate, so I guess their patented technology is pretty good at what it does. Thanks for giving me something to ponder.
BTW – As you said, Bill, I’ve seen sites, both large and small, with sitelinks.
That’s a very interesting and comprehensive article, Bill. It’s good to know more about the mysteries of sitelinks.
There is some mystery about sitelinks.
I’ve blocked some sitelinks in Google’s Webmasters Tools for a number of reasons. I didn’t like the text that was used for some. I didn’t really want the page that was showing up as a sitelink for others.
I think it might be a good idea to at least allow webmasters to suggest the text of sitelinks for some pages. Microsoft came out with a patent application that described some ways to do that, and I wrote about it in a post last year:
Should Webmasters Pick Their Own Quicklinks in Search Results?
Thank you. I agree – it does seem that most often pages linked to frequently by other pages on the same site often end up as pages with sitelinks pointing to them. It does seem more likely that pages that are either contained in the main navigation for a page, or in a link that appears as a sitewide link may be chosen as a sitelink or quicklinks. But, I’ve seen exceptions to that as well, such as a popular blog post or page that isn’t linked to widely within the same site.
Thank you. There’s always been some part of SEO concerned with how search results for pages appear in the search engines. With the introduction of things like sitelinks or quicklinks, that’s been expanded even more.
We started to study this one and made some tests. Feedback from some people here are somewhat true that large sites display Quicklinks. Initially that’s what we noticed. We’re developing more strategies to try to identify how Google does it – although it is something not guaranteed.
Bill – hi.
with the advent of Google’s personal preferences showing up on my search results I am finding more and more of these quick-links appearing and realize that because i visit my own sites often – I seem to want to find similar sites and pages, which is why I think i am getting more of these quick-links.
still even though i am aware that the results are skew, i do enjoy it for a mini second. then i have to clear cookies to do it right.
It’s possible that many larger sites do display quicklinks, but I’ve seen large sites that don’t, and small sites that do. I think one of the keys behind whether or not a search engine provides quick links to a site is how well a search engine might be able to crawl a site and get a sense of that site’s structure, and another is whether or not the site and the pages of the site fulfill navigational type queries.
I think those two issues are more important than such factors as whether or not a site is large or small, highly linked to or not.
Interesting point, and something I haven’t seen a lot of discussion on – does personalize search affect what we see as quicklinks or site links? Do we see more of them because a site that otherwise might have not been the top search result now is, and Google tends to like to show sitelinks for pages that are at the tops of search results?
It’s definitely a topic worth thinking about more, and investigating. Thanks.
What people need to understand is just because you get “blessed” with sitelinks it doesn’t mean your site is climbing SERPS. I’ve had many sites attain site links with poor rankings. They (site links) are exciting to get, but don’t celebrate until you see your site #1 for your search terms.
Google now will sometimes show a series of up to four quicklinks in a line under a snippet for a page even if that page isn’t at the top spot in search results rankings. But they usually don’t show the two column set of sitelinks if you aren’t at the top of a search result for a particular query.
As you note, sitelinks aren’t an indication that you are climbing in rankings for other query terms, but sitelinks are an indication that Google is associating your site with a particular query term that it thinks your site is a good match for as a navigational query. Having those sitelinks doesn’t necessarily mean that your site will rank better for other query terms however. But I think it’s a start. It shows that Google has some sense of what your site is about.
Thx for this posting Bill 🙂
I tried to get site links for one of my projects for about two years but it hasn’t worked yet 🙁
In my opinion you need an huge number of trust worth links to get site links or you have to be the only one in your small niche *g*
I suspect that both things you suggest would be helpful – having a huge number of trustworthy links or being the only one in your small niche. I have found that having a somewhat unique name that people will search for can help, as well as having a site that is set up so that search engine crawling programs can crawl and index easily as well.
Being new the “SEO game”, I’ve always wondered about those site links and quick links. Never really understood what was behind them. Much to learn I guess 🙂
Thanks for the information, Bill. I’ll have to stick close to you so I can keep learning the right stuff!
You’re welcome. There is always something new to learn. 🙂
I’ve been reading about SEO for a bit less than a year now and try to apply best practices where I can. But it’s only now that I’ve stumbled upon your blog and this great article! I need to enlist your rss asap 🙂
That being said, I’ve never come across an article about quicklinks nor have I ever wondered about them. I always assumed they used the most popular pages and the main internal (menu-)link referring to that page.
Food for thought! 🙂
Thanks, Webdesign DragolinDesign
It’s easy to make that assumption that link popularity and main navigation being the primary things that a search engine might look at to decide which pages should show as quicklinks, but it’s only part of the picture.
The primary purpose does seem to be that the pages shown are supposed to be ones that someone seeing the quicklinks in the search results would most likely be prone to visit, but that doesn’t always bear out well when looking at the quicklinks that are sometimes assigned to a page.
I think we will eventually come to a standard that will force developers to label menu links just for the sake of being included in the Google results.
I don’t know. There are good sides and bad sides to standards when it comes to web design and development, and there are sometimes competing standards.
Standards can be helpful in having people share best practices that are predictable, and allow for things like common development in HTML editing tools and browsers. But standards can also act to limit innovation and development of new ideas and approaches.
A search engine also needs to be flexible enough when it comes to standards to include pages that searchers want to see even if those pages aren’t coded to include the latest standards, such as pages that validate perfectly as HTML or XHTML.
I have a question, why search engines only show sitelinks for very selective websites? How do they select websites to show sitelinks.
Sorry if it seems like stupid question.
It’s a good question, and there are a number of reasons why only some sites have sitelinks.
The first of them is that Google usually (but not always now) limits sitelinks to pages at the tops of results.
Those pages often tend to be the top results for queries that tend to be navigational in nature. For instance, when someone searches for ESPN, chances are pretty good that they are trying to get to the homepage of ESPN. If Google doesn’t think a query is navigational like that, it may be much less likely to show sitelinks.
If my search is for something like “buy green sneakers,” it’s more likely that I’m not looking for a specific web site.
Since sitelinks are navigational shortcuts to deeper pages within a site, they often show up when that top result is seen as an ideal page for a navigational query.
There are a few different ways that a search engine might decide that a page is the ideal or perfect page to show in response to a query that might be seen to be navigational. One approach, that I’ve seen described by Microsoft, is that most of the people searching for that query tend to choose just one site, even though they might get a number of relevant sites shown to them in search results.
I’ve read your post considering site links.
I do have a website in a small niche and also a huge number of links but I have never got site links. I think it is also important to have the respective keyword you want to rank for in you URL. So if you do want to rank for the keyword “sun”, it will be necessary to have the word “sun” in your domain name.
I don’t think that it is essential to have the keyword in your URL for sitelinks to show up for your query. I’ve seen sitelinks show up in search results for pages for terms that aren’t included in the URL or domain name for a site.
I think Google is as big a hypocrite as it comes. There whole speal is making sure there SERPs are based around relevance and what people want to see. Most of the time, especially around uber competitive terms, the site that shows up number one in the SERPs is a site that has been manipulated by and SEOer to appear to be more relevant. Now what Google is doing is not only showing those sites at the top of the SERPs, but also rewarding them with additional internal links. This just continues to feed into the monopoly that a lot of sites have within the SERPs.
Hi Gray Collar SEO,
I actually think that sitelinks are a good idea, though there are times when I wonder at some of the choices that Google, Yahoo, and Bing makes in their selection of sitelinks.
Sitelinks have evolved so that sites don’t necessarily have to be the top result anymore to appear for some pages – the single row type sitelinks, or quicklinks, as opposed to the two column list of links.
One of the things that I’ve seen that makes a difference as to whether or not sitelinks appear for a site is whether or not the site in question can be crawled easily by a search engine – something that SEOs often help a site in doing that isn’t a manipulative practice.
One of the major goals of SEO is to make it more likely that the objectives of a site owner and the objectives of a searcher can both be fulfilled by making it more likely that the right audience can find what they are looking for, so that the site owner and the searcher can interact in a positive and meaningful way.
Good to see an SEO guy loves to read patent applications and subsequently share with others 🙂
Once again – good analysis Bill!
Quick question – is this post an updated version of this post?
Thanks. I really enjoy digging through the patents to see if there are any hidden gems (or at least new ideas) amongst them.
To a degree the two posts are related, but this post is not an update of the earlier one.
The older post describes a Google patent that tells us about how the search engine might choose to display sitelinks for sites. This post is about a Yahoo patent that describes how the search engine might choose the text that appears as anchor text for sitelink or quicklinks.
So both are about sitelinks, but the focus of each differs – why sitelinks in the earlier post, and how those sitelinks are linked to in the second.
It would also be a mistake to assume that Yahoo and Google both follow the exact same processes when deciding upon which pages to link to, and how to display links to those pages, but it’s possible that a lot of the same thought processes might play a role in each aspect of choosing and showing sitelinks.
Comments are closed.