How Google May Identify Navigational Queries and Resources

I sometimes see people say that paid search is a great way to do keyword research for SEO, but I disagree with that statement. Paid search primarily focuses upon keywords that are transactional in nature – usually the terms chosen are the kind that match an intent to buy something, download something, or take some other kind of action. I’ve asked many people who do search engine advertising, and focus on Adwords if they ever target queries that are informational in nature, and most of the time the answer has been no.

Often searchers will do some research on a product or service before they decide who to buy from. They will perform research to find what kinds of features are available for different products, try to find reviews or opinions from others, They may try to compare different manufacturers as well. These types of queries are more informational in nature, and the same searcher will conduct these types queries that evidence an informational intent before they begin to consider a query with a transactional intent.

An image of a mariner's compass from the Library of Congress.

There’s another type of query that plays a strong role in search and SEO. People often search for pages that they expect to find on the Web, likely from a site that they know exists or assume should exist. For instance, if I search for [IBM services], I’m likely searching for a page on the IBM website that I expect will be something I can find easily. If I look for [WordPress support], chances are in most cases that I want a page on one of the WordPress sites that can answer my questions about the software or the service. If I type [ESPN] into a Google toolbar, I’m expecting that the ESPN website will be the first result that shows up.

The top result on a search for [ESPN] is the ESPN home page and site links.

How does Google determine that a query might be navigational in nature, and that a resource might be a great response to it? A Google patent granted last week describes an algorithmic approach to making those determinations. It begins with looking at queries in search sessions.

Search Sessions

When someone searches, they may enter multiple searches into a search engine looking for answers to their informational or situational needs. A search engine might consider the sequence of queries as a search session, with all of those searches related to one another. Is this something that a search engine can take advantage of?

And how exactly might a Google decide what a search session is exactly?

A patent granted to Google last week explores how to use query sessions to identify Navigational Queries and Navigational Resources.

A navigational query is one that seems to be looking for a specific page in response to that query, and a navigational resource is the specific page being looked for.

So how does a search engine determine that a sequence of queries might be part of a query session?

The patent points to at least three different ways it might make that decision, though also explicitly states that other methods might be used, too:

Revision Time Window – A search session might be defined by queries submitted within a revision window time period between the input of a query and a subsequent query. Someone submits a query, and a current search session is initiated. If the searcher doesn’t submit another query within a certain amount of time, such as five minutes, the session might be closed.

User Defined Session Period – A search session might be defined by a searcher taking specific actions, such as logging into and then out of the search engine.

Common Terms between Queries – A search session could be defined by common terms between new search query and a previous query from the same device. These common terms would contain either some of the same words, or word within the same topic cluster.

Queries are stored in query logs, showing an initial query and revised queries, and the order that a search submitted them during a search session. In addition to looking at what queries might be submitted in a query session, Google also keeps an eye on what happens after each query is submitted, including clicks on results as users select pages in response to their queries.

Query Logs and Click Logs to Map Queries to Resources

For every query in a query session, there’s an associated set of search results filled with a corresponding set of resources. The results that are clicked upon by searchers are collected in a click log, along with an indication of whether the click was a long click or a short click.

A long click is a click-through that either results in further clicks on the page in question, or a a visit that lasts longer than a certain amount of time, such as 30 seconds.

A short click is a click-through to a page where there aren’t any subsequent clicks on that page during that session or is shorter than a certain amount of time, such as 30 seconds.

There may be other ways of deciding upon what a long click or short click might be. For example, in my recent post on reachability scores, long clicks and short clicks where also used as indications of whether or not people found value in pages linked to. Visits to pages with videos that were shorter than 30 seconds were still considered long clicks if the videos were less than 30 seconds long, and a person watched the whole video.

Quality Scores for Navigational Queries and Navigational Resources

The patent tells us that other information might also be used to decide whether or not a page is a navigational result as well, including a quality score for the page, characteristics of the URL for the page, and the click-throughs.

A quality score used to determine whether a query is a navigational query can include an information retrieval score for the results that appear for the query as well as characteristics of the URLs that show up for it as well. For example, the shorter the length of the URL and lower the directory depth of the URL (the closer to the root directory) the higher the quality score of the query.

Queries could be mapped to pages during search sessions based upon information in the query longs and click logs based upon user actions, such as the selections of pages, and whether or not clicks to them were long clicks or short clicks.

If this kind of user behavior is aggregated, and people perform similar query revisions for specific queries, and tend to end up at the same pages, or resources, both the queries and the resources pointed to might be seen as navigational queries and navigational resources. If many of the results for the query also tend to link to a specific page identified as a potential navigational resources, then it’s more likely to be seen as one by Google.

If Google decides that a query is a navigational query, and a result for it is a navigational result, the patent tells us that the page identified as a navigational result might be re-ordered in search results so that it is “always shown on a first page search results provided for the user session.”

So, a navigational result for a query doesn’t necessarily have to be the first result for that query, but may be shown on the first page of search results from Google.

The patent is:

Navigational resources for queries
Invented by Trystan Upstill, Henele I. Adams, Eric Lehman, Neesha Subramaniam, Wensi Xi, Sundeep Tirumalareddy
Assigned to Google
US Patent 8,326,826
Granted December 4, 2012
Filed: January 15, 2009

Abstract

Methods, systems, and apparatus, including computer program products, for identifying navigational resources for queries. In an aspect, a candidate query in a query sequence is selected, and a revised query subsequent to the candidate query in the query sequence is selected.

If a quality score for the revised query is greater than a quality score threshold and a navigation score for the revised query is greater than a navigation score threshold, then a navigational resource for the revised query is identified and associated with the candidate query. The association specifies the navigational resource as being relevant to the candidate query in a search operation.

The patent provides some additional details on how navigational queries and navigational resources might be identified. Some posts I’ve written in the past on navigational queries include:

Take Aways

When searchers perform searches, the intent behind their search may determine what kinds of searches they perform. When they first start exploring a topic of type of product, their queries may be informational in nature, as they attempt to learn more about a product or service. When they may be at the stage where they are likely to buy or perform some other kind of transaction, their queries might be transactional in nature.

When a searcher is looking for information that they believe may exist already on the web, and have an idea of what page it might be on, or a good reason to believe that there should be such a page, their query could be considered to be navigational in nature.

It’s possible that someone can influence whether or not a page does appear as a navigational result, and the patent points at some of the signals that a search engine might look at in making that determination.

Are your pages showing up as navigational results being returned for navigational queries?

Share

29 thoughts on “How Google May Identify Navigational Queries and Resources”

  1. As always, great work !

    I think this could be the reason that we are seeing navigational queries returning multiple results from same website as your pointed out here:

    So, a navigational result for a query doesn’t necessarily have to be the FIRST RESULT for that query, but may be shown on the FIRST PAGE of search results from Google.

    I think it even goes beyond FIRST page and shows it on 2,3 or even 4th page.

    It is an interesting development but Google needs to be absolutely sure that the query is navigational otherwise the user experience will be very bad (returning multiple results from same website is not a good user experience).

  2. I really agree with you that paid traffic is not the best way to marketing business, I think that grown up business that had websites needs to SEO their site to acquire organic traffic from search engines which is a natural way to get converting visitors and lowering the budget on fail advertisement.

  3. Totally agree. it’s quite challenging to “sell in” the importance of non-transactional or highly targeted to a particular product or service. For all the uproar about recent algo updates like Panda, I have found this has actually made it easier to get clients on side when it comes to investing in proper content and not obsessing about ranking for the “cheap whatevermyproductis” type keywords. An oldie, but a goodie on this subject is the Machine is Learning video

  4. My link didn’t work… maybe being being moderated. To see the video, for those who haven’t, just Google “The Machine is Us/ing Us (Final Version)”

  5. Hi Yasir,

    Navigational intent may result in a page appearing on the first page of Google’s results. When we see multiple pages in those results, it could be because they are among the most relevant pages for a query, and Google often used to show a couple of pages in search results when that happened, with the second one indented after the first. When we see three, four, or even more, it’s quite likely that Google has identified an even stronger relationship between a particular query, and a site, and is likely engaging in an entity relationship. See: 10 Most Important SEO Patents: Part 6 – Named Entity Detection in Queries. Google assumes in those instances that the searcher is attempting to perform a “site” search for the site, and find something on it, as opposed to finding the information elsewhere.
    So, it’s very much like a navigational query and navigational result – but there are even stronger signals about searchers wanting the domain in question. An example that I’ve seen that shows that is a search for [spaceneedle hours], where I’ve seen up to 7-8 results from spaceneedle.com showing up on the front page of Google’s search results (presently, I’m seeing 8 results from that domain)) It’s actually a very good user experience in that instance.

  6. Hi Axel,

    I’m not saying anything negative about paid search, but rather against the statement that keyword research for paid search is helpful for organic search. The intent behind the queries is often different, and what might make a great keyword phrase for PPC isn’t necessarily a great term for organic search. And if an phrase gets a lot of clicks through sponsored listings, that doen’t mean that it will get a lot of clicks in organic search results.

  7. I think many informational queries are in fact “early funnel” queries and search marketers ignore them at their peril. You may have early funnel queries that aren’t appearing to bring in conversions, but if you use Google Analytics multi-channel attribution features, often you will uncover that some of these early terms are providing “assists” to later terms.

    The exceptions to this that I always try to put in negative lists though are
    “what is”
    “what are”
    “definition”
    “wiki”
    and
    “wikipedia”

    because those types of queries are so early in the funnel they can barely said to be part of it at all. But queries like “get rid of termites” or “how to get rid of termites” are great terms, regardless of the paid/organic distinction.

    My 2 cents

    - Ted

  8. Hi Ted,

    Agree 100%, and I try my best to make sure that searchers don’t have to conduct transactional searches if I can by being responsive to their informational queries, providing solutions, and making it as easy as possible to conduct those transactions without doing more searching. :)

  9. Pingback: Rien à dire…
  10. Hi.. I read your post and I’m really impressed on how you detailed everything with regards to navigational queries. I am just wondering if Google Knowledge Graph is the same thing as you have mentioned here? I read in other post that this knowledge graph is intended to resolve the user’s query without having to navigate to other sites and assemble the information.

  11. I’ve always wondered what impact rank checkers could have on search session data for the search engines; generally there will be a whole series of similar queries, and in smaller outfits possibly all coming from the same IP/machine. Of course this could be one of the reasons Google is against them, and could also explain why sometimes they can throw up weird results.

  12. I’m from Russia, now more and more often to read foreign blogs, learned a lot, though, and a Google translator from Russian to English is not an exact translation, but the main thrust of this article, I get it. I think that sometimes we get to experiment Google, then Google evaluates the behavior of groups of people and depending on the number of actions using the formula the chances that a particular message, and captures the preferences and interests of users to different subjects of sites, but this is not all analyzes Google. I think that these figures are a reference for the search engine Google.

  13. What would be the correlation between navigational, informational and transactional queries on tablets and smart phones. The whole seo game is changing, where on tablets/smart-phones the proportion of navigational/informational queries are increasing over transactional queries which are more likely handled by applications. Mobile marketing is on the rise. Would appreciate your comments on that, Bill!

  14. Bill, regarding your statement in the article “I sometimes see people say that paid search is a great way to do keyword research for SEO, but I disagree with that statement.”, I would have to disagree.

    PPC is a great way to do RESEARCH. If you find a keyword someone used to find you, and determine it’s not a good keyword to bid on for PPC, how is that not helpful research? PPC is a great way to see what search phrases people are using to find you and broad matching is a great way to get into results for keywords you may not otherwise achieve. The Show Search Phrases tool in Adwords let’s you see which search phrases people used that triggered a broad match, and that in turn give you new keywords to bid on, or new keywords to prevent your ads from being shown, thereby increasing your CTR and quality score.

  15. Hi Norm,

    Keyword research for PPC isn’t as helpful for organic SEO keyword research as many people have suggested in the past. If you’re doing the research for PPC, it doesn’t give you any idea whatsoever what the competition might be like for those keyword phrases in organic search, the “broad” search volumes don’t give you a very good idea of how much organic search volume there are for those terms, and the “competition” indicator is only an indication of how much competition you have against others for paid search.

    The focus of most keyword research for PPC also tends to be much more heavily weighted towards transactional queries than informational queries, which means that you’re much further down the sales funnel to start with.

    PPC is not a research methodology for organic keywords – it’s a mean to its own ends.

  16. Hi Bill,

    I like how you concluded into this: “So, a navigational result for a query doesn’t necessarily have to be the first result for that query, but may be shown on the first page of search results from Google.”

    I will tip a hat for you! I find it very educational.

  17. Hi Bill, Thanks again for a very well researched and informative post. I do think that for transactional keywords Google Adwords is a good place to gather valuable keyword data but as you say you are already getting someone very far down the sales funnel. A better time to influence somebody’s buying decision is at the time when they are gathering information.

  18. The questions and AdWords tests to make the answers could improve offline marketing, inbound telemarketing and the nature of products and services given.

  19. Has Google got to clever for it’s own good?

    The more I read, and experience of Google both as a net user and as a web designer the more I think Google is out thinking itself. Over the last few months Bing seems to have passed Google in terms of relevant searching.

    As an example if I search by name for a hotel at London’s Heathrow Airport I would expect and probably want the official website for that hotel somewhere close to the top of the page, with then perhaps reviews or maybe a third party booking engine further down. What I would not want is Google to return multiple links to the official website where each link is for a different page on that website, and by doing this exclude all other relevant web pages.

    But in my experience Google has gone backwards over the last few months, regressed into a search engine of yesteryear, whilst Bing seems to have improved and surpassed Google.

    Perhaps it’s an algorithm to far!

  20. Search sessions are incredibly powerful, and this patent makes sense with inbound marketing.
    As a product seller with an etablished brand, giving user informational pages would definitly helps transactionnal pages to rank well, as the site could be used by the user in all steps of the funnel, even when the user bounce back to Google at every step of the funnel to make an additional query.

  21. Thanks for the informative post.I read your post and I’m really impressed.I really agree with you that paid traffic is not the best way to marketing business.Best way to get traffic naturally.

  22. Well I also think that search query session performs a crucial role in accessing best navigational queries and navigational resource outcome .I will recommend better to think first before consuming your time and money in following any type of marketing strategy.

Comments are closed.