How Classification of Page Elements and Search Results May Influence Alternative Titles and Snippets Displayed in Google

Has an improvement in how Google understands the layout of pages, and understands and classifies different elements found on page had an impact on the titles and snippets that we see in search results? Google may classify queries to decide what to show for those page titles and snippets in search results, but it’s possible that they might also be classifying the contents of “original titles and snippets and URLs” when deciding to show different titles and expanded snippets. Might Google do that in combination with a classification of page elements (a portion of HTML containing some text) found on the pages in search results to try to determine the best representation of a search result in response to a query?

Google May Chose Titles and Snippets for Pages

When you search at Google, the search results displayed for web pages include titles, URLs, and snippets for the pages listed in the results. In those, the query terms you used, or sometimes synonyms for them, may be included in the title and snippet, and Google will highlight those. As a site owner, you should have unique and engaging titles and meta descriptions for each page you want indexed by search engines. Not only does that make it more likely that search engines will crawl, index, and display those pages, but if you use the keywords you’re optimizing those pages for within those titles and descriptions, Google may show your choice of title and meta description within search results.

But not always…

Google has sometimes been showing titles that are different from the actual page title listed in an HTML <title> element for pages over the past few years, if the search engine might find a different title to be a better match for the query used than the one a site owner chose. A January 12th blog post on the Official Google Webmaster Central blog, Better page titles in search results, told us more about how and why those titles might be chosen:

We use many signals to decide which title to show to users, primarily the <title> tag if the webmaster specified one. But for some pages, a single title might not be the best one to show for all queries, and so we have algorithms that generate alternative titles to make it easier for our users to recognize relevant pages. Our testing has shown that these alternative titles are generally more relevant to the query and can substantially improve the click-through rate to the result, helping both our searchers and webmasters. About half of the time, this is the reason we show an alternative title.

The post also tells us that Google will sometimes choose to display their choice of titles when a webmaster might have forgotten to include a title, or uses a non-descriptive title like “Home.” A couple of the other issues that the post tells us may be the reason for Google deciding upon a different title is when a site uses the same or very similar titles for a large number of pages, or when a title might be “unnecessarily long or hard-to-read.”

Back in November, the Google Inside Search blog also told us of a change to snippets that Google may display, in a post entitled Ten recent algorithm changes, where one of those changes involved snippets shown for pages in search results.

Snippets with more page content and less header/menu content: This change helps us choose more relevant text to use in snippets. As we improve our understanding of web page structure, we are now more likely to pick text from the actual page content, and less likely to use text that is part of a header or menu.

It’s interesting that the blog post mentions how Google is improving “our understanding of web page structure.” Less than a week ago, Google also announced that they would be paying more attention to, and possibly penalizing pages that had too much advertising content “above the fold” on pages, in the post Page layout algorithm improvement. Might Google’s better understanding of the layouts of pages also be influencing some of the decisions that they make when choosing page titles and snippets( or descriptions)?

Google does provide some best practices in a help page on writing page titles and meta descriptions, and Google was granted a patent on expanded snippets last year.

The help page tells us that Google’s choosing of titles and snippets is an automated process:

Google’s generation of page titles and descriptions (or “snippets”) is completely automated and takes into account both the content of a page as well as references to it that appear on the web. The goal of the snippet and title is to best represent and describe each result and explain how it relates to the user’s query.

They also tell us that their decision of which titles and snippets to display might be based upon a number of choices, including information from the page, or from publicly available sources off the page such as “anchor text or listings from the Open Directory Project (DMOZ).”

I started a Google Plus post on when Google might transform page titles yesterday, and received a lot of interesting comments and some examples as well, of times when Google has changed titles for pages.

It’s not hard to see when looking at search results that Google is sometimes showing considerably longer snippets within them than they did in the past as well:

Google search results with many snippets that are much longer than what Google used to show in the past.

How Does Google Decide What to Show in Titles and Snippets?

Google was granted a patent earlier this week that describes a way that they might classify queries, and classify search result snippets to come up with new snippets that might use content found on the pages being returned in results for those queries.

A number of the patent’s inventors have been associated with Google Custom Search Engines. The patent itself is written more broadly than just being about Google Custom Search Engines though, and could be a description of how Google might be identifying at least some titles and snippets for pages shown in search results. There’s also a lot of language about labels in the patent, which is an integral part of the Custom Search Engines. The patent is:

Classifying search results to determine page elements
Invented by Tania Bedrax-Weiss, Ramanthan Guha, Patrick Riley, and Corin Anderson
Assigned to Google
US Patent 8,103,676
Granted January 24, 2012
Filed: October 11, 2007

Abstract

This invention relates to determining page elements to display in response to a search. A method embodiment of this invention determines a page element based on a search result. A classification is determined based on a search result. Page elements are generated based on the classification.

By using the search result, as opposed to just the query, page elements are generated that corresponds to a predominant interpretation of a user’s query within the search results. As result, the page elements may, in most cases, accurately reflect the user’s intent.

The patent provides a very detailed look at how Google might go from classifying queries (using a lookup table, or via some other method) to classifying the different elements of a search result such as a title, snippet, and URL in that result. Those classifications might be weighted somewhat by the position that those pages appear at within search results as well.

The original snippet for a page will often have the query terms within it, if available. And part of the decision as to whether or not Google might use a meta description from a page as the snippet for that page is going to depend upon whether or not that meta description contains the keyword terms used to search for the page. If a search engine decides to use content from a page, there may be a number of considerations involved as described in a Yahoo patent application from a couple of years ago, such as:

  • How readable the phrase is,
  • How relevant the phrase is to the query,
  • How relevant the phrase is to the page it appears upon,
  • How long the phrase is,
  • Combinations of the above

If the meta description doesn’t contain the query term or terms used to find the page, or it just doesn’t seem very relevant, then Google might choose to use something else. Google may follow a very similar path to that decision that the Yahoo patent does, at least for an initial snippet.

Google may then follow a classification approach described in the patent to classify the title, the meta description or a description from page text as a snippet, and the URL in the search result

In this newly granted patent, Google might decide to classify different sections of text within the HTML elements on pages to see if it can come up with the best match for the classifications of the query and classification of the different parts of the original search result.

The classifications of query and search result would then be matched up with text on a page within different HTML elements to see if the content on the page might make better choices.

Example

The home page from the Acme Manufacturing Company might be very well optimized for the term [roadrunner traps], and the page title and meta description for the page might include that term and Google might show the page title and meta description as written by the site owner on a search for [roadrunner traps].

Chances are that the home page will also rank very well for [Acme Manufacturing Company] as well, but if the title of the home page doesn’t contain the name of the company, and the meta description doesn’t either, Google might generate a description or snippet from the content found on the page.

It might take the HTML <title>, the first snippet Google initially creates, and the URL for the home page, and classify each of them, as well as classifying the query [Acme Manufacturing Company] as well.

It may then also classify different portions of text within HTML elements on the page, and classify them as well.

For instance, the alt text of the logo for the page might contain the text “Acme Manufacturing Company – Maker of fine explosive roadrunner traps”. Some of the other text on that page may also contain text that might be suitable for a description of the company, such as a short description of the company in a sidebar <p> element that says “Acme Manufacturing Company has been making custom traps for cartoon coyotes since 1948 for Looney Tunes Cartoons. Acme Manufacturing’s products have been ruggedly tested through many cartoon episodes viewed by millions.”

Google might take the logo alt text to use as a page title, and the sidebar content to use as part or all of an extended snippet.

I’ve provided a fairly high level look at this patent, and if you want a deeper look, I highly recommend drilling down into the patent.

There’s definitely some value in Google looking closely at classifications of queries and original search results, and trying to match them up with classifications of page elements to provide the best representation in a transformed search result of why a page might be relevant for a searcher’s query.

Another Approach to Expanded Snippets

The patent that I mentioned above about expanded snippets described how Google might show those longer snippets where it is presently showing Instant Previews, and may have been something that Google might have planned on doing before deciding to show screenshots of pages in those previews. It also describes a couple of ways that it might decide upon which text to show as larger snippets, but doesn’t include the classification processes that this page elements patent does. Here’s what it does tell us:

An expanded snippet may include the text excerpt of the snippet with additional text located in proximity to the text excerpt in the search result document, such as text before and/or after the text excerpt. In one implementation, the additional text may include a predetermined amount of text, such as a predetermined number of terms, before and/or after the text excerpt in the search result document.

In another implementation, the additional text may be more intelligently selected. For example, the additional text may include all (or less than all) of the text preceding the text excerpt to a beginning or end of a structural component (e.g., paragraph, table entry, section, etc.) in which the text excerpt occurs in the search result document. Alternatively, or additionally, the additional text may include a previous and/or following structural component based on the structural component in which the text excerpt occurs in the search result document.

Regardless, of whether Google is using the classification of search results and page elements, or is using the method described in this expanded snippets patent, I’m seeing lots of search results with longer snippets in them.

It’s definitely something to keep an eye upon, especially since the display of search results plays such an important role in whether or not searchers visit pages they see within those results.

Share

31 thoughts on “How Classification of Page Elements and Search Results May Influence Alternative Titles and Snippets Displayed in Google”

  1. Thanks for this! I’m really interesting in this stuff since it’s something I see a lot. Today I found that Google changed my title and description for a page in order to match the query. I was experimenting a bit and this was the case: the query in google.it was “scarpe”. I had a page in which the URL key is “scarpe” but in the title and in the metadescription I use a synonym “calzature”. Well, I saw Google changed the metatitle and description with “scarpe”, so that I have apparently all the metas with my query, ignoring completely my metas. Another case I saw is this: I have a page full of products (100): google in the description of the page ignored the one I put and put a summary of some products. It’s really bad to see actually, but I don’t understand why Google chose this way. In the page there is an optimized metadescription and in case there is even an optimized text.

    anyway, thanks for this. your article are always interesting!

  2. I’ve always thought the title tag has been an over rewarded element. Reading this post made me realise its probably because Google needs good titles for users to percieve the SERPs as being relevant (rather than just seeing it entirely as a ranking signal).

    Changing titles allows them to reward sites that they know are actually more relevant whilst also displaying SERPs that are percieved by users as relevant. Ultimately the power of titles for ranking will drop off even further.

    (stating the obvious I know but bear with those of us who need to think out loud occasionally)

  3. I’ve been seeing this a lot where G will cut off the end of the title tag and add “Laguna Beach Real Estate.” This has actually helped in the number of hits I’m getting so I’m not complaining. It’s interesting to see how G will take control your site based on the search queries. So far it’s been a good thing.

  4. They’ve definitely gotten better at choosing what to use for the title. I ran a test in December of 2009 to demonstrate to a developer I was working with that there was no point in using a meta title tag, and that it was certainly a mistake to use such a tag instead of an actual title element. I published a page with no title tag, and a meta title of “Is meta title the same as title?” I gave it a meta description of “Testing to see if a meta title element is treated the same way as a standard title element” and two heading tags: an of “Testing the “meta title” Tag” and it automatically had an of “SEO Articles and Tests” — the heading for the menu of that section of the site.

    Once the page had been indexed, I ran a site: search and it came up with a title of “SEO Articles and Tests” — the , which was far less relevant to the content of the page than the . I guessed that Google had picked the simply because it was the first heading in the page’s source code.

    Your post reminded me of this little unscientific test, so I’ve just run a new site: search, and the title Google is now using for the page is “Meta Title – Raise My Rank” — much more relevant, I’d say. I then ran a search for [title tag raise my rank] and the page comes up third in the results, with that same title, but this time, instead of using my meta description as the snippet, they’re grabbing text from the page: “Raise My Rank SEO Services … Testing the “meta title” Tag. I noticed recently that the meta data for a client’s pages contained two versions of the “title” tag: …”

  5. Sorry, I used tags to name the two headings, and of course they got removed. The first heading (Testing the “meta title” tag) is an h1 and the other is an h2

  6. That is very interesting Bill. I have never noticed Google showing a title in search results that didn’t match what was specified in the title tag.

    But now that you point it out, it does make sense from a conversion rate optimization standpoint.

    I can definitely see how an alternate and more relative title could and would benefit the author of the web page if it more accurately described what is on the page. That is never a bad thing in my opinion.

    I can’t tell you how many articles I have written the title for and then proceeded to write the article only to have the subject wander off into something else, something less relevant to the title.

    I am totally fine with Google picking what they consider a more relevant title if it benefits us both more on the back-end.

    Mark

  7. Bill,

    This is quite interesting. For years, people have focussed on keyworded page titles in order to get people to the site. Now with this, those could well go out of the window as one is not sure what big G will display now.

    Something to keep track of for sure to see how this comes about.

  8. Pingback: The Best Small Business Marketing News & Blog Posts of the Week — January 27, 2012 – DIYSEO
  9. Hi Alessio,

    Thank you very much.

    I’ve seen some changes to page titles over the past few months, and I’ve been seeing more and more extended snippets for search results as well. I was pretty excited to see this patent published on this topic, to see if it could give us some clues as to what Google might be doing.

    It sounds like you’re not very happy with some of the changes for your pages, which I understand completely.

    Was the information from the products included in the snippet related in some way to the query that you searched with? Did your meta description for that page include the query terms?

  10. Hi Tim,

    In the instances where people write titles that are actually unique, well written, and descriptive of the pages that they title, I think they should be given a lot of value by the search engines. In those cases, they do help make Google’s search results look better as well.

    I think the points you raise are pretty good. A page might be very relevant for a query, but the title might be a terrible match for the page it is a title to.

    I’ve seen a lot of bad titles for pages that are way too short, way too long, not focused specifically on the pages that they are titles for, or just completely wrong. It’s not a surprise that Google might change some titles that aren’t very good matches.

    I’m primarily concerned about the possibility that Google might take a good title and make it much worse. Fingers crossed that they won’t, but this is an automated process that site owners have no control over.

  11. Hi Sam,

    Is Google doing that on your pages in response to queries that contain some or all of the words “Laguna Beach Real Estate”?

    It is good to see that someone is actually seeing some benefit from these changes in titles.

  12. Hi Bob,

    Good to hear that you’ve been keeping a close eye on this phenomena, and are seeing improvements over Google’s choices in titles.

    Great test on the meta title element. I’m guessing that your right about Google picking that heading because it was the first one listed. Great to hear how their choices of what to display in response to different queries has changed since then as well. Very appreciative that you shared this with us.

  13. Hi Mark,

    It’s definitely something worth keeping an eye out for. I guess it’s more likely that you might notice changes to page titles when you know what’s actually in the titles for those pages. I’m not sure how frequently people look at the top of their browser window to check what’s up there to compare it with what Google might have shown in search results. But you notice right away when it’s a page from your site, and it’s somehow changed.

    Hopefully those title changes are improvements (fingers crossed). It just worries me to see this because I like to have control over changes like that.

  14. Hi Praveen,

    I suspect that in most instances, if you write titles that descriptive, and aren’t too short, and aren’t too long, and focus upon the keywords that a page is optimized for, that Google will show the title you’ve selected in response to the query you’re optimizing for. I don’t think that will change.

    It’s when titles aren’t very well crafted, aren’t very descriptive or focused, are missing completely, or just completely irrelevant that Google will probably replace them with something that might be unanticipated.

    And when a page title is changed so that it might be better for a specific query, that isn’t a bad thing. I’m pretty much concerned about those instances where Google might change a title for a page and it doesn’t improve the title, and make it more likely that it might be clicked upon in search results.

  15. So the stuff we’ve been doing to optimize all this time Google has now become smart enough to do for the people who aren’t, huh? I think in the near future there will be a lot of changes with the title tag in rankings/search displays. The Google tweaks to this are just beginning.

  16. Hi Bill,

    Tks a lot for your article i found useful to understand better the serp in google. As a french blogger , i found a bit more complicated as we have also for 2 days the “Search plus your world” algorythm on top of how google choose to display urls with snippets and/or Meta tags.

  17. Yep, good article. I actually haven’t seen this at all, but agree it makes sense, and as of now, I’m not opposed to it. But I’m wondering, when you look at Google Analytics and you compare metrics under the Content section – first dimension = Page Title second dimension = Landing Page, is that an accurate report to know what kind of titles Google has been generating for you? Looking at some of my clients’ analytics, I see 5 very different titles for the home page and they don’t contain keywords that we’re actively trying to rank for. Without going into much detail, some of the titles are “login for students and faculty”, “(college major 1)”, “(college major 2)” etc.

    Is this how you guys would investigate your automated titles?

  18. Hi Charles,

    I still believe that it’s in the best interest of site owners to do the best that they can in creating titles and descriptions for pages that do a great job of describing the content found on those pages, and follow some intelligent approach to choosing keywords to include within those and on those pages.

    Google may add sometimes their own titles when a page is relevant for a query, and will attempt to pick the best snippet they can from page content if necessary, but chances are that those titles and descriptions aren’t going to be as engaging or useful as titles or descriptions that someone who spends some time on those might create.

  19. Hi Clara,

    Google’s choosing of a different title for a page is something that might happen on many different sites, and not just WordPress sites.

    I don’t think it’s ever a good idea to have two different title elements for a page when you create that page. I haven’t seen a default WordPress feature that enables you to create multiple titles for a page, and if you see that happening, it’s possibly because of a flaw in a theme, or a plugin rather than an intentional effort to offer two different titles.

    Choose just one title for a page, and make it the best title that you can, using your specific choice of keyword terms or phrases. If the page happens to rank well for another query term that doesn’t include that optimized term, and Google decides to change the title for that page because it ranks well for that term, that might just be gravy on top of the traffic that you receive for the term that you chose to optimize for.

  20. Hi largeron,

    You’re welcome.

    When Google changes titles for pages like they have been, it’s not necessarily because of the Search Plus Your World feature that Google added. Google was sometimes doing this long before they launched their newest version of social search.

  21. Hi Mike,

    I guess the only way to try to determine that is to replicate the actual search to see if Google is changing the title for you as well for those pages using those specific queries.

  22. Hi Bill
    Interesting post here, especially to someone fairly new to all this. I must say that I have never known or seen Google changing the title but have not been looking. It’s an interesting development especially for those highly keyworded and targetted content titles.It will be intresting to see how this plays out. After the Panda update soem of my traffic increased so Google playing with titles willbe interesting and will now look out for this.
    Thanks Jon

  23. Hi Jon,

    They’ve actually been making changes to titles at times for a few years, but it looks like they are doing more of it, and being a little more transparent about it.

    If they limit those changes to scattered instances where the changes reflect the query term used in the search when the changed title appears in one of the search results, that shouldn’t necessarily be bad most of the time. I just would love to see when and where that happens when it does, and what Google is showing as a title.

  24. I have noticed recently that Google has been using the meta description tag for the SERPS title. It seems to be able to tell where the title is less relevant to the page. It’s important for me as I use Joomla and with its system of adding meta tags to the menu, and content somewhere else, it does happen sometimes that the title doesn’t accurately portray the content. Google is encouraging us all to up our game content-wise (in quality and relevance) which is only a good thing! Interesting stuff, thanks.

  25. I’m surprised to hear Google say programmatic meta descriptions are okay. I also noticed they didn’t say anything about the length of a page title. I’ve heard it should be < 70 characters? Although some of your titles are longer than that, Bill?

  26. Hi Thomas,

    I’m not sure if you’re reading the quote I included above as an indication that Google says “programmatic meta descriptions are okay,” or if you read that elsewhere.

    What the quote in my post is saying is that Google’s decision as to what snippets to show in search results is completely an algorithmic approach. So their decision as to whether they might use a meta description or some content from a page is one that follows a certain set of algorithms. If you do use programmatic meta descriptions, and that’s definitely something that you might do with a fairly large site, you should definitely make sure that what you end up with is different enough from one page to another and descriptive of the pages they describe.

    Google doesn’t say anything about the length of titles, and most of the statements that you might see from people writing about the lengths of titles is likely based upon how much a title Google might display in search results. That doesn’t mean that Google won’t look past the first 63 or so characters that they might display in search results. It’s usually not a bad idea not to make your titles too long, but I do sometimes write longer titles if they are descriptive, and I feel they are appropriate.

    If you look around at just about everything Google may have written about title elements, I’m not sure that you’ll see them provide a specific number of characters anywhere (I’ve done that search a number of times over the years).

Comments are closed.