Understanding how search engines work is an important component of SEO. The search engines are constantly tuning their algorithms. For that reason, the successful SEO professional is constantly studying search engine behavior and learning how search engines work.
- 1 Understanding Search Engine Results
- 1.1 Understanding the Layout of Search Results Pages
- 1.2 Understanding How Vertical Results Fit into the SERPs
- 1.3 Google’s Knowledge Graph
- 2 Algorithm-Based Ranking Systems: Crawling, Indexing, and Ranking
- 3 Understanding What Content Search Engines Can “See” on a Web Page
- 4 Determining Searcher Intent and Delivering Relevant, Fresh Content
- 4.1 Document Analysis and Semantic Connectivity
- 4.2 Content Quality and User Engagement
- 4.3 Problem Words, Disambiguation, and Diversity
- 4.4 Where freshness matters
- 4.5 Why These Algorithms Sometimes Fail?
- 4.6 The Knowledge Graph
- 5 Analyzing Ranking Factors
- 18.104.22.168 Domain-level link authority features
- 22.214.171.124 Page-level link metrics
- 126.96.36.199 Page-level keywords and content
- 188.8.131.52 Page-level features other than keywords
- 184.108.40.206 Domain-level brand metrics
- 220.127.116.11 Page-level traffic/query data
- 18.104.22.168 Page-level social metrics
- 22.214.171.124 Domain-level keyword usage
- 126.96.36.199 Domain-level keyword-agnostic features
- 5.1 Negative Ranking Factors
- 5.2 Other Ranking Factors
- 5.3 Using Advanced Search Techniques
- 5.4 Advanced Google Search Operators
- 6 More Advanced Search Operator Techniques
- 7 Vertical Search Engines
- 7.1 Vertical Search from the Major Search Engines
- 7.2 Universal Search/Blended Search
- 7.3 More specialized vertical search engines
- 7.4 Country-Specific Search Engines
- 7.5 Optimizing for Specific Countries
Understanding Search Engine Results
In the search marketing field, the pages the engines return to fulfill a query are referred to as search engine results pages (SERPs). Each engine returns results in a slightly different format, and these may include vertical results — results that can be derived from different data sources or presented on the results page in a different format, which we’ll illustrate shortly.
Understanding the Layout of Search Results Pages
The Figure shows the SERPs in Google for the query iphone 11 pro. The various sections outlined in the Google search results are as follows:
- Search query box (1)
- Vertical navigation (2)
- PPC advertising (3)
- Natural/organic/algorithmic results (4)
Each unique section represents a snippet of information provided by the engines. Here are the definitions of what each piece is meant to provide:
Search query box
All of the engines show the query you’ve performed and allow you to edit or reenter a new query from the search results page. If you begin typing, you may notice that Google gives you a list of suggested searches below. This is the Google autocomplete suggestions feature, and it can be incredibly useful for targeting keywords.
Each engine offers the option to search different verticals, such as images, news, video, or maps. Following these links will result in a query with a more limited index. In Figure 2-3, for example, you might be able to see news items about stuffed animals or videos featuring stuffed animals.
PPC (Paid Search) advertising
The text ads are purchased by companies that use either Google AdWords or Bing. The results are ordered by a variety of factors, including relevance (for which click-through rate, use of searched keywords in the ad, and relevance of the landing page are factors in Google) and bid amount (the ads require a maximum bid, which is then compared against other advertisers’ bids).
Natural /organic/ algorithmic results
These results are pulled from the search engines’ primary indices of the Web and ranked in order of relevance and importance according to their complex algorithms. This area of the results is the primary focus of this section of the book.
Query refinement suggestions
Query refinements are offered by Google, Bing, and Yahoo!. The goal of these links is to let users search with a more specific and possibly more relevant query that will satisfy their intent.
In March 2009, Google enhanced the refinements by implementing Orion Technology, based on technology Google acquired in 2006. The goal of this enhancement is to provide a wider array of refinement choices. For example, a search on principles of physics may display refinements for the Big Bang, angular momentum, quantum physics, and special relativity.
Understanding How Vertical Results Fit into the SERPs
These “standard” results, however, are certainly not all that the engines have to offer. For many types of queries, search engines show vertical results, or instant answers, and include more than just links to other sites to help answer a user’s questions. These types of results present many additional challenges and opportunities for the SEO practitioner.
Figure 2 shows an example of these types of results. The query in Figure 2 brings back a business listing showing an address and the option to get directions. This result attempts to provide the user with the answer he is seeking directly in the search results.
Figure 3 shows another example. The Google search in Figure 3 for weather plus a city name returns a direct answer. Once again, the user may not even need to click on a website if all she wanted to know was the temperature.
Figure 4 is an example of a search for a well-known painter. A Google search for the famous painter Edward Hopper returns image results of some of his most memorable works (shown in the lower-right of the screenshot). This example is a little different from the “instant answers” type of result shown in Figure 2 and Figure 3. If the user is interested in the first painting shown, he may well click on it to see the painting in a larger size or to get more information about it. For the SEO practitioner, getting placed in this vertical result could be a significant win.
As you can see, the vast variety of vertical integration into search results means that for many popular queries you can expect to receive significant amounts of information in the SERPs themselves. Engines are competing by providing more relevant results and more targeted responses to queries that they feel are best answered by vertical results, rather than web results.
Figure 5 is an example of a celebrity search on Bing. The results in Figure 5 include a series of images of the famous actor Charlie Chaplin.
As a direct consequence, site owners and web marketers must take into account how this incorporation of vertical search results may impact their rankings and traffic. For many of the searches shown in the previous figures, a high ranking — even in position #1 or #2 in the algorithmic/organic results — may not produce much traffic because of the presentation of the vertical results above them.
The vertical results also signify an opportunity, as listings are available in services from images to local search to news and products.
Google’s Knowledge Graph
The search engines are actively building structured databases of information that allow them to show answers to questions that are not simply linked to web pages. In Figure 4, the information on the upper right is an example of this. Google provides direct answers in the result, including Edward Hopper’s birth date, place of birth, the date and place of his death, his spouse, and more. In Figure 5, Bing provides similar information for Charlie Chaplin.
Not only is additional information shown, but it is not just a data dump: it shows that the search engines are working to develop their own knowledge of the relationships between people and things. In the case of Figure 4, we can see that Google understands that:
- Edward Hopper is the name of a person.
- People have dates and places of birth.
- People have dates and places of death.
- People might have spouses.
The search engines are actively mapping these types of relationships as part of their effort to offer more complete information directly in the search results themselves.
Advanced SERP Features
Here’s a list of the SERP features that we’ll be going over.
- Featured Snippet
- Knowledge Graph
- Knowledge Card
- Local Pack
- Local Teaser
- Top Stories
- People Also Asked
- Related Searches
- Searches Related to
- Site Links
- Google Ads (Paid Results)
- Shopping Results
Algorithm-Based Ranking Systems: Crawling, Indexing, and Ranking
Understanding how crawling, indexing, and ranking works is useful to SEO practitioners, as it helps them determine what actions to take to meet their goals. The search engines must execute many tasks very well to provide relevant search results. Put simplistically, you can think of these as:
Crawling and indexing trillions of documents (pages and files) on the Web (note that they ignore pages that they consider to be “insignificant,” perhaps because the pages are perceived as adding no new value or are not referenced at all on the Web). Responding to user queries by providing lists of relevant pages.
Crawling and Indexing
To offer the best possible results, search engines must attempt to discover all the public pages on the World Wide Web and then present the ones that best match up with the user’s search query. The first step in this process is crawling the Web. The search engines start with a seed set of sites that are known to be very high quality, and then visit the links on each page of those sites to discover other web pages.
The link structure of the Web serves to bind together all of the pages that were made public as a result of someone linking to them. Through links, search engines’ automated robots, called crawlers or spiders, can reach the many trillions of interconnected documents.
In Figure 2-10, you can see the home page of USA.gov, the official U.S. government website. The links on the page are outlined in red. Crawling this page would start with loading the page, analyzing the content, and then seeing what other pages USA.gov links to.
The search engine would then load those other pages and analyze that content as well. This process repeats over and over again until the crawling process is complete. This process is enormously complex, as the Web is a large and complex place.
Search engines do not attempt to crawl the entire Web every day. In fact, they may become aware of pages that they choose not to crawl because those pages are not likely to be important enough to return in a search result.
The first step in this process is to build an index of terms. This is a massive database that catalogs all the significant terms on each page crawled by the search engine.
A lot of other data is also recorded, such as a map of all the pages that each page links to, the clickable text of those links (known as the anchor text), whether or not those links are considered ads, and more.
To accomplish the monumental task of holding data on hundreds of trillions of pages that can be accessed in a fraction of a second, the search engines have constructed massive data centers to deal with all this data.
One key concept in building a search engine is deciding where to begin a crawl of the Web. Although you could theoretically start from many different places on the Web, you would ideally begin your crawl with a trusted seed set of websites.
Starting with a known trusted set of websites enables search engines to measure how much they trust the other websites that they find through the crawling process.
Retrieval and Ranking
For most searchers, the quest for an answer begins as shown in Figure 2-11.
The next step in this quest occurs when the search engine returns a list of relevant pages on the Web in the order it believes is most likely to satisfy the user. This process requires the search engines to scour their corpus of hundreds of billions of documents and do two things: first, return only the results that are related to the searcher’s query; and second, rank the results in order of perceived importance (taking into account the trust and authority associated with the site). It is both relevance and importance that the process of SEO is meant to influence.
Relevance is the degree to which the content of the documents returned in a search matches the user’s query intention and terms. The relevance of a document increases if the page contains terms relevant to the phrase queried by the user, or if links to the page come from relevant pages and use relevant anchor text.
You can think of relevance as the first step to being “in the game.” If you are not relevant to a query, the search engine does not consider you for inclusion in the search results for that query.
Importance refers to the relative importance, measured via citation (the act of one work referencing another, as often occurs in academic and business documents), of a given document that matches the user’s query. The importance of a given document increases with every other document that references it. In today’s online environment, citations can come in the form of links to the document or references to it on social media sites. Determining how to weight these signals is known as citation analysis.
You can think of importance as a way to determine which page, from a group of equally relevant pages, shows up first in the search results, which is second, and so forth. The relative authority of the site, and the trust the search engine has in it, are significant parts of this determination. Of course, the equation is a bit more complex than this, and not all pages are equally relevant. Ultimately, it is the combination of relevance and importance that determines the ranking order.
Evaluating Content on a Web Page
Search engines place a lot of weight on the content of each web page. After all, it is this content that defines what a page is about, and the search engines do a detailed analysis of each web page they find during their crawl to help make that determination.
You can think of this as the search engine performing a detailed analysis of all the words and phrases that appear on a web page, and then building a map of that data for it to consider showing your page in the results when a user enters a related search query. This map, often referred to as a semantic map, seeks to define the relationships between those concepts so that the search engine can better understand how to match the right web pages with user search queries.
If there is no semantic match of the content of a web page to the query, the page has a much lower possibility of showing up. Therefore, the words you put on the page, and the “theme” of that page, play a huge role in ranking.
Figure 2-13 shows how a search engine will break up a page when it looks at it, using a page on the Forbes website.
The navigational elements of a web page are likely similar across the many pages of a site. These navigational elements are not ignored, and they do play an important role, but they do not help a search engine determine what the unique content is on a page. To do that, the search engine focuses on the part of Figure 2-13 that is labeled “Unique Page Content.”
Determining the unique content on a page is an important part of what the search engine does. The search engine uses its understanding of unique content to determine the types of search queries for which the web page might be relevant. Because site navigation is generally not unique to a single web page, it does not help the search engine with that task.
This does not mean navigation links are not important — they most certainly are; however, they simply do not count when a search engine is trying to determine the unique content of a web page, as they are shared among many web pages.
One task the search engines face is judging the value of content. Although evaluating how the community responds to a piece of content using link analysis is part of the process, the search engines can also draw some conclusions based on what they see on the page.
For example, is the exact same content available on another website? Is the unique content the search engine can see two sentences long or 500 words long? Does the content repeat the same keywords excessively? These are a few examples of factors the search engine can evaluate when trying to determine the value of a piece of content.
Understanding What Content Search Engines Can “See” on a Web Page
Search engine crawlers and indexing programs are basically software programs. These programs are extraordinarily powerful. They crawl hundreds of trillions of web pages, analyze the content of all these pages, and analyze the way all these pages link to one another. Then they organize this into a series of databases that can respond to a user search query with a highly tuned set of results in a few tenths of a second.
This is an amazing accomplishment, but it has its limitations. Software is very mechanical, and it can understand only portions of most web pages. The search engine crawler analyzes the raw HTML form of a web page. If you want to see what this looks like, you can do so by using your browser to view the source.
Figure 2-14 shows how to do that in Chrome, and Figure 2-15 shows how to do that in Firefox. Typically you can access it most easily by right-clicking with your mouse on a web page to access a hidden menu.
it has nothing to do with the page’s content.
The search engine crawler is most interested in the HTML text on the page.
Figure 2-17 is an example of HTML text on a Semrush article page.
Although Figure 2-17 still shows some HTML encoding, you can see the “regular” text clearly in the code. This is the unique content that the crawler is looking to find.
In addition, search engines read a few other elements. One of these is the page title. The page title is one of the most important factors in ranking a given web page. It is the text that shows in the browser’s title bar (above the browser menu and the address bar).
Figure 2-18 shows the code that the crawler sees, using Trip Advisor as an example.
The first highlighted area in Figure 2-18 is for the <title> tag. The <title> tag is also often (but not always) used as the title of your listing in search engine results (see Figure 2-19).
In addition to page titles, search engines previously used the meta keywords tag. This is a list of keywords that you wish to have associated with the page. Spammers (people who attempt to manipulate search engine results in violation of the search engine guidelines) ruined the SEO value of this tag many years ago, so its value is now negligible, as search engines don’t use it anymore. Spending time on meta keywords is not recommended because of the lack of SEO benefit.
The second highlighted area in Figure 2-18 shows an example of a meta keywords tag.
Search engines also read the meta description tag (the third highlighted area in the HTML source in Figure 2-18). However, the content of a meta description tag is not directly used by search engines in their ranking algorithms.2
Nonetheless, the meta description tag plays a key role, as search engines often use it as a part or all of the description for your page in search results. Therefore, a well-written meta description can have a significant influence on how many clicks you get on your search listing, and the click-through rate on your search listing can impact your ranking. As a result, time spent on meta descriptions is quite valuable. Figure 2-20 uses a search on trip advisor to show an example of the meta description tag being used as a description in the search results.
The user’s keywords are typically shown in boldface when they appear in the search results (sometimes close synonyms are shown in boldface as well). As an example, in Figure 2-20, TripAdvisor is in boldface at the beginning of the description. This is called keywords in context (KWIC).
A fourth element that search engines read is the alt attribute for images. The alt attribute was originally intended to allow something to be rendered for audiences who cannot view the images, primarily:
Vision-impaired people who do not have the option of viewing the images.
People who turn off images for faster surfing. This is generally an issue only for those who do not have a broadband connection.
Search engines also read the text contained in the alt attribute of an image tag (<img>). An image tag is an element that is used to tell a web page to display an image.
What search engines cannot see?
It is also worthwhile to review the types of content that search engines cannot “see” in the human sense.
For instance, although search engines are able to detect that you are displaying an image, they have little idea what the image is a picture of, except for whatever information you provide in the alt attribute, as discussed earlier. They can recognize only some very basic types of information within images, such as the presence of a face, or whether images have pornographic content by how much flesh tone they contain. A search engine cannot easily tell whether an image is a picture of Bart Simpson, a. boat, a house, or a tornado. In addition, search engines typically don’t recognize any text rendered in the image.
The reality is that the search engines have the technology to handle these types of tasks to some degree. For example, you can take a picture of the Taj Mahal and drag it into the search box in Google image search, and the search engine will recognize it. However, because of the processing power required for image recognition, search engines do not currently try to recognize all of the images they encounter across the Web.
Search engines are also experimenting with technology to use optical character recognition (OCR) to extract text from images, but it is not yet in general use within search. The main problem with applying OCR and image processing technology is that it’s very computationally intensive, and not practical to apply at the scale of the Web.
Audio and video files are also not easy for search engines to read. As with images, the data is not easy to parse. There are a few exceptions where the search engines can extract some limited data, such as ID3 tags within MP3 files, or enhanced podcasts in AAC format with textual “show notes,” images, and chapter markers embedded. Ultimately, though, search engines cannot distinguish a video of a soccer game from a video of a forest fire.
Search engines also cannot read any content contained within a program. The search engine really needs to find text that is readable by human eyes looking at the source code of a web page, as outlined earlier. It does not help if you can see it when the browser loads a web page — it has to be visible and readable in the source code for that page.
The problem arises because the content is retrieved by a script running on the client computer (the user’s machine) only after receiving some input from the user. This can result in many potentially different outputs. In addition, until that input is received, the content is not present in the HTML of the page, so the search engines cannot easily see it.
As of HTML 5, a construct known as the embed tag (<embed>) was created to allow the incorporation of plug-ins into an HTML page. Plug-ins are programs located on the user’s computer, not on the web server of your website. The embed tag is often used to incorporate movies or audio files into a web page; it tells the plug-in where it should look to find the data file to use. Content included through plug-ins may or may not be invisible to search engines.
Frames and iframes are methods for incorporating the content from another web page into your web page. Iframes are more commonly used than frames to incorporate content from another website. You can execute an iframe quite simply with code that looks like this:
<iframe src =”http://accounting.careerbuilder.com” width=”100%” height=”300″>
<p>Your browser does not support iframes.</p> </iframe>
Frames are typically used to subdivide the content of a publisher’s website, but they can be used to bring in content from other websites, as in http://accounting.careerbuilder.com on the Chicago Tribune website, shown in Figure 2-21.
Figure 2-21 is an example of something that works well to pull in content (provided you have permission to do so) from another site and place it on your own. However, the search engines recognize an iframe or a frame used to pull in another site’s content for what it is, and therefore may ignore that content. In other words, they don’t consider content pulled in from another site as part of the unique content of your web page.
Determining Searcher Intent and Delivering Relevant, Fresh Content
Modern commercial search engines rely on the science of information retrieval (IR). This science has existed since the middle of the 20th century, when retrieval systems powered computers in libraries, research facilities, and government labs. Early in the development of search systems, IR scientists realized that two critical components comprised the majority of search functionality: relevance and importance (which we defined earlier in this chapter). To measure these factors, search engines perform document analysis (including semantic analysis of concepts across documents) and link (or citation) analysis.
Document Analysis and Semantic Connectivity
In document analysis, search engines look at whether they find the search terms in important areas of the document — the title, the metadata, the heading tags, and the body of the text. They also attempt to automatically measure the quality of the document based on document analysis, as well as many other factors.
Reliance on document analysis alone is not enough for today’s search engines, so they also look at semantic connectivity. Semantic connectivity refers to words or phrases that are commonly associated with one another. For example, if you see the word aloha, you associate it with Hawaii, not Florida. Search engines actively build their own thesaurus and dictionary to help them determine how certain terms and topics are related. By simply scanning their massive databases of content on the Web, they can use fuzzy set theory and certain equations to connect terms and start to understand web pages and sites more like a human does.
The professional SEO practitioner does not necessarily need to use semantic connectivity measurement tools to optimize websites, but for those advanced practitioners who seek every advantage, semantic connectivity measurements can help in each of the following sectors:
- Measuring which keyword phrases to target
- Measuring which keyword phrases to include on a page about a certain topic
- Measuring the relationships of text on other high-ranking sites and pages
- Finding pages that provide “relevant” themed links
Although the source for this material is highly technical, SEO specialists need only know the principles to obtain valuable information. It is important to keep in mind that although the world of IR has hundreds of technical and often difficult-to-comprehend terms, these can be broken down and understood even by an SEO novice.
Common types of searches in the IR field include:
A proximity search uses the order of the search phrase to find related documents. For example, when you search for “sweet German mustard” you are specifying only a precise proximity match. If the quotes are removed, the proximity of the search terms still matters to the search engine, but it will now show documents that don’t exactly match the order of the search phrase, such as Sweet Mustard — German.
Fuzzy logic technically refers to logic that is not categorically true or false. A common example is whether a day is sunny (i.e., is 50% cloud cover a sunny day?). In search, fuzzy logic is often used for misspellings.
These are searches that use Boolean terms such as AND, OR, and NOT. This type of logic is used to expand or restrict which documents are returned in a search.
Term weighting refers to the importance of a particular search term to the query. The idea is to weight particular terms more heavily than others to produce superior search results. For example, the appearance of the word the in a query will receive very little weight in selecting the results because it appears in nearly all English language documents. There is nothing unique about it, and it does not help in document selection.
IR models (search engines) use fuzzy set theory (an offshoot of fuzzy logic created by Dr. Lotfi Zadeh in 1969) to discover the semantic connectivity between two words. Rather than using a thesaurus or dictionary to try to reason whether two words are related to each other, an IR system can use its massive database of content to puzzle out the relationships.
Although this process may sound complicated, the foundations are simple. Search engines need to rely on machine logic (true/false, yes/no, etc.). Machine logic has some advantages over humans, but it doesn’t have a way of thinking like humans, and concepts that are intuitive to humans can be quite hard for a computer to understand. For example, oranges and bananas are both fruits, but oranges and bananas are not both round. To a human this is intuitive.
For a machine to understand this concept and pick up on others like it, semantic connectivity can be the key. The massive human knowledge on the Web can be captured in the system’s index and analyzed to artificially create the relationships humans have made. Thus, a machine knows an orange is round and a banana is not by scanning thousands of occurrences of the words banana and orange in its index and noting that round and banana do not have great concurrence, while orange and round do.
This is how the use of fuzzy logic comes into play, and the use of fuzzy set theory helps the computer to understand how terms are related simply by measuring how often and in what context they are used together.
For example, a search engine would recognize that trips to the zoo often include viewing wildlife and animals, possibly as part of a tour.
To see this in action, conduct a search on Google for zoo trips. Note that the boldface words that are returned match the terms that are italicized in the preceding paragraph. Google is setting “related” terms in boldface and recognizing which terms frequently occur concurrently (together, on the same page, or in close proximity) in their indexes.
Search companies have been investing in these types of technologies for many years. In September 2013, Google quietly let the world know that it had rewritten its search engine and given it the name “Hummingbird”. This rewrite was in large part done to enable a whole new set of capabilities for recognizing the relationships between things.
For example, if you use Google’s voice search (click on the microphone icon at the right of the search box on Google.com) and ask it “Who is Tom Brady?” it will answer that question for you with a search result, but then use audio to tell you that he is an “American football quarterback for the New England Patriots of the National Football League.”
This shows that Google understands many aspects of Tom Brady. For example:
- He has an occupation: quarterback, playing American football (as distinct from the way the term football is used outside of the United States and Canada).
- He plays on a team: the New England Patriots.
- The New England Patriots belong to a league: the NFL.
Content Quality and User Engagement
Search engines also attempt to measure the quality and uniqueness of a website’s content. One method they may use for doing this is evaluating the document itself. For example, if a web page has lots of spelling and grammatical errors, that can be taken as a sign that little editorial effort was put into that page.
They can also analyze the reading level of the document. One popular formula for doing this is the Flesch-Kincaid Grade Level Readability Formula, which considers factors like the average word length and the words per sentence to determine the level of education needed to be able to understand the sentence. Imagine a scenario where the product being sold on a page is children’s toys and the search engine calculates a reading level of a college senior. This could be another indicator of a poor editorial effort.
The other method that search engines can use to evaluate the quality of a web page is measuring actual user interaction. For example, if a large number of users who visit the web page after clicking on a search result immediately return to the search engine and click on the next result, that would be a strong indicator of poor quality.
Engagement with a website began to publicly emerge as a ranking factor with the release of the Panda update by Google on February 23, 2011.6 Google has access to a large number of data sources that it can use to measure how visitors interact with your website. Just because Google has access to this data, however, does not mean that it’s definitely using the data as a ranking factor. That noted, some of those sources include:
Interaction with web search results
For example, if a user clicks through on a SERP listing and comes to your site, clicks the back button, and then clicks on another result in the same set of search results, that could be seen as a negative ranking signal. Or if the results below you in the SERPs are getting clicked on more than you are, that could be seen as a negative ranking signal for you and a positive ranking signal for them. Whether search engines use this signal or not, or how much weight they might put on it, is not known.
It is hard to get a firm handle on just what percentage of websites run Google Analytics. A 2008 survey of websites by Immeria.net showed their share at 59%,7 and the Metric Mail Blog checked the top 1 million sites in Alexa and found that about 50% of those had Google Analytics.8 Suffice it to say that Google is able to collect detailed data about what is taking place on a large percentage of the world’s websites.
This provides Google with a rich array of data on that site, including:
- Bounce rate
- The percentage of visitors who visit only one page on your website.
- Time on site
The time spent by the user on the site. Note that Google Analytics receives information only when each page is loaded, so if you view only one page it does not know how much time you spent on that page. More precisely, then, this metric tells you the average time between the loading of the first page and the loading of the last page but does not take into account how long visitors spent on the last page loaded.
Page views per visitor
The average number of pages viewed per visitor on your site.
It is not known how many users out there use the Google Toolbar, but we believe that it numbers in the millions. For these users, Google can track their entire web surfing behavior. Unlike Google Analytics, the Google Toolbar can measure the time from when a user first arrives on a site to the time when she loads a page from a different website. It can also get measurements of bounce rate and page views per visitor.
This enables users to vote for a page on the page itself. There is currently no evidence that Google uses this as a ranking factor, but in theory, it could.
Chrome Personal Blocklist Extension
Google offers a Chrome add-on called the Personal Blocklist Extension. This enables users of the Chrome browser to indicate a search result they don’t like. This was first used by Google as a part of its Panda algorithm, which attempts to measure the quality of a piece of content. You can read more about this algorithm in Chapter 9.
Google has its own URL shortener. This tool allows Google to see what content is being shared, and which content is being clicked on, even in closed environments where Google web crawlers are not allowed to go.
What matters most is how your site compares to that of your competition. If your site has better engagement metrics, this is likely to be seen as an indication of the quality and can potentially boost your rankings with respect to your competitors. Little has been made public about the way search engines use these types of signals, so the preceding comments are our speculation on what Google may be doing in this area.
In link analysis, search engines measure who is linking to a site or page and what they are saying about that site/page. They also have a good grasp on who is affiliated with whom (through historical link data, the site’s registration records, and other sources), who is worthy of being trusted based on the authority of sites linking to them, and contextual data about the site on which the page is hosted (who links to that site, what they say about the site, etc.).
Link analysis goes much deeper than counting the number of links a web page or website has, as all links are not created equal (one link can be worth 10 million times more than another one). Links from a highly authoritative page on a highly authoritative site will count more than other links of lesser authority. A search engine can determine a website or page to be authoritative by combining an analysis of the linking patterns and semantic analysis.
For example, perhaps you are interested in sites about dog grooming. Search engines can use semantic analysis to identify the collection of web pages that focus on the topic of dog grooming. The search engines can then determine which of these pages about dog grooming have the most links from the set of websites relevant to the topic of dog grooming. These pages are most likely more authoritative on the topic than the others.
The actual analysis is a bit more complicated than that. For example, imagine that there are five pages about dog grooming with a lot of links from pages across the Web on the topic, as follows:
- Page A has 213 topically related links.
- Page B has 192 topically related links.
- Page C has 203 topically related links.
- Page D has 113 topically related links.
- Page E has 122 topically related links.
Further, it may be that Pages A, B, D, and E all link to one another, but none of them links to Page C. In fact, Page C appears to have the great majority of its relevant links from other pages that are topically relevant but have few links to them. In this scenario, Page C may not be considered authoritative because it is not linked to by the right sites.
The concept of grouping sites based on who links to them, and whom they link to, is referred to as grouping sites by link neighborhood. The neighborhood you are in says something about the subject matter of your site, and the number and quality of the links you get from sites in that neighborhood say something about how important your site is to that topic.
The degree to which search engines rely on evaluating link neighborhoods is not clear, and links from irrelevant pages can still help the rankings of the target pages. Nonetheless, the basic idea remains that a link from a relevant page or site should be more valuable than a link from a nonrelevant page or site.
Another factor in determining the value of a link is the way the link is implemented and where it is placed. For example, the text used in the link itself (i.e., the actual text that will go to your web page when the user clicks on it) is also a strong signal to the search engines.
This is referred to as anchor text, and if that text is keyword-rich (with keywords relevant to your targeted search terms), it can potentially do more for your rankings in the search engines than if the link is not keyword-rich. For example, anchor text of “Dog Grooming Salon” may bring more value to a dog grooming salon’s website than anchor text of “Click here.” However, take care. If you get 10,000 links using the anchor text “Dog Grooming Salon” and you have few other links to your site, this definitely does not look natural and could lead to a penalty.
The semantic analysis of a link’s value goes deeper than just the anchor text.
For example, if you have that “Dog Grooming Salon” anchor text on a web page that is not really about dog grooming at all, the value of the link is lower than if the page is about dog grooming. Search engines also look at the content on the page immediately surrounding the link, as well as the overall context and authority of the website that is providing the link.
Evaluating Social Media Signals
Sites such as Facebook and Twitter have created whole new ways for users to share content or indicate that they value it. This has led many to speculate that search engines could be using these signals as a ranking factor. Fueling that speculation, in August 2013, Moz released the data from its latest correlation study, and it showed a very strong correlation between +1s and ranking in Google.
Problem Words, Disambiguation, and Diversity
On the opposite side of the coin are words that present an ongoing challenge for the search engines. One of the greatest challenges comes in the form of disambiguation. For example, when someone types in boxers, does he mean the prize fighter, the breed of dog, or the type of underwear? Another example is jaguar, which is at once a jungle cat, a car, a football team, an operating system, and a guitar. Which does the user mean?
Search engines deal with these types of ambiguous queries all the time. The two examples offered here have inherent problems built into them, but the problem is much bigger than that. For example, if someone types in a query such as cars, does he:
- Want to read reviews?
- Want to go to a car show?
- Want to buy one?
- Want to read about new car technologies?
The query cars is so general that there is no real way to get to the bottom of the searcher’s intent based on this query alone. One way that search engines deal with this is by looking at prior queries by the same searcher to find additional clues to his intent.
Another solution they use is to offer diverse results. As an example, Figure 2-24 shows a generic search, this time for GDP.
This brings up an important ranking concept. It is possible that a strict analysis of the relevance and link-driven importance scores in Figure 2-24 would not have resulted by itself in the Investopedia.com result being on the first page, but the need for diversity elevated the page’s ranking. This concept of altering the results in this manner is sometimes referred to as query deserves diversity (QDD).
A strict relevance- and importance-based ranking system might have shown a variety of additional government pages discussing the GDP of the United States. However, a large percentage of users will likely be satisfied by the government pages already shown, but for those users who are not, showing more of the same types of pages is not likely to raise their level of satisfaction with the results.
Introducing a bit of variety allows Google to also provide a satisfactory answer to those who are looking for something different from the government pages. Google’s testing has shown that this diversity-based approach has
resulted in a higher level of satisfaction among its users. For example, the testing data for the nondiversified results may have shown lower click-through rates in the SERPs, greater numbers of query refinements, and even a high percentage of related searches performed subsequently.
The idea to deliberately introduce diversity into the result algorithm makes sense and can enhance searcher satisfaction for queries such as:
Company names (where searchers might want to get positive and negative press, as well as official company domains)
Product searches (where ecommerce-style results might ordinarily fill up the SERPs, but Google tries to provide some reviews and noncommercial, relevant content)
News and political searches (where it might be prudent to display “all sides” of an issue, rather than just the left- or right-wing blogs that did the best job of obtaining links)
Search engines also personalize results for users based on their search history or past patterns of behavior. For example, if a searcher has a history of searching on card games, and then does a search for dominion, the search engine may choose to push some of the results related to the Dominion card game higher in the results, instead of emphasizing the power company.
Where freshness matters
Much of the time, it makes sense for the search engines to deliver results from older sources that have stood the test of time. However, other times the response should be from newer sources of information.
For example, when there is breaking news, such as an earthquake, the search engines begin to receive queries within seconds, and the first articles begin to appear on the Web within 15 minutes.
In these types of scenarios, there is a need to discover and index new information in near real time. Google refers to this concept as query deserves freshness (QDF). According to the New York Times, QDF takes several factors into account,10 such as:
- Search volume
- News coverage
- Blog coverage
QDF applies to up-to-the-minute news coverage, but also to other scenarios such as hot, new discount deals or new product releases that get strong search volume and media coverage. There has also been speculation that Google will apply QDF more to sites that have higher PageRank.11
Why These Algorithms Sometimes Fail?
Search engines do some amazing stuff. Nonetheless, there are times when the process does not work as well as you would like to think. Part of this is because users often type in search phrases that provide very little information about their intent (e.g., if they search on car, do they want to buy one, read reviews, learn how to drive one, learn how to design one, or something else?). Another reason is that some words have multiple meanings, such as the jaguar example we used previously in this section.
The Knowledge Graph
Traditional search results are derived by search engines crawling and analyzing web pages and then presenting that information in the search results. However, Google’s mission “is to organize the world’s information and make it universally accessible and useful”. Google is actively pursuing initiatives to build databases of information that go far beyond traditional web-based search.
Note that earlier in this chapter we wrote about vertical search. Vertical search relates to breaking search into different categories, such as a search for images, videos, or local business information. The Knowledge Graph is more about providing richer answers directly in the search results, often answering the user’s question directly without her having to click through to a website.
In May 2012, Google announced the Knowledge Graph. Initially, this was a set of structured databases of information that allows Google to access information without deriving it from the Web. You can see an example of the type of data that Google might extract from its Knowledge Graph database in Figure 2-25.
Google initially built the Knowledge Graph using data from Freebase, Wikipedia, and the CIA Fact Book. This allowed Google to answer many questions, but really only satisfied a very small number of search queries. For that reason, Google is constantly working on expanding the information in the Knowledge Graph.
In addition, Google is investing in ways to more reliably extract information from other sources, including websites, to present as direct answers in search. Google refers to these as “featured snippets.” Figure 2-26 shows the search result for buying a car.
In this result, Google provides a set of step-by-step instructions extracted from the CNN Money website. Note that two steps are omitted, so to get the complete procedure or additional details on each step, the user must click through to the CNN Money website.
In some cases, Google does provide the complete instructions in the search results, but most of the time it does not. A study performed by Stone Temple Consulting examimed 276 examples of step-by-step instructions, and found that 217 of these (79%) did not provide the complete instructions.
A related concept is semantic search, which overlaps the Knowledge Graph to some degree, but also takes into account many other factors to personalize results for the searcher. You can see a depiction of some of these factors in Figure 2-27.
Analyzing Ranking Factors
Moz periodically conducts surveys of leading SEOs to determine what they think are the most important ranking factors.12 Here is a high-level summary of the top nine results, in priority order (as suggested by the referenced study):
- Domain-level link authority features
- Page-level link metrics
- Page-level keywords and content
- Page-level keyword-agnostic features
- Domain-level brand metrics
- Usage and traffic/query data
- Page-level social metrics
- Domain-level keyword usage
- Domain-level keyword-agnostic features
Here is a brief look at each of these:
Domain-level link authority is based on a cumulative link analysis of all the links to the domain. This includes factors such as the number of different domains linking to the site, the trust/authority of those domains, the rate at which new inbound links are added, the relevance of the linking domains, and more.
This refers to the links as related to the specific page, such as the number of links, the relevance of the links, and the trust and authority of the links received by the page.
Page-level keywords and content
This describes the use of the keyword term/phrase in particular parts of the HTML code on the page (<title> tag, <h1>, alt attributes, etc.).
Page-level features other than keywords
Factors included here are page elements such as the number of links on the page, number of internal links, number of followed links, number of “nofollow” links, and other similar factors.
Domain-level brand metrics
This factor includes search volume on the website’s brand name, mentions, whether it has a presence in social media, and other brand-related metrics.
Page-level traffic/query data
Elements of this factor are click-through rate to the page in the search results, bounce rate of visitors to the page, and other similar measurements.
Social metrics considered include mentions, links, shares, likes, and other social media site–based metrics. It should be emphasized that many SEO practitioners believe that this is a ranking factor even though studies have since shown otherwise, and representatives from Google clearly state that social signals are not part of their algorithm.
Domain-level keyword usage
This refers to how keywords are used in the root or subdomain name, and how impactful that might be on search engine rankings.
Domain-level keyword-agnostic features
Major elements of this factor in the survey include the number of hyphens in the domain name, number of characters in the domain name, and domain name length.
Negative Ranking Factors
It’s also possible to have negative ranking factors. For example, if a site has a large number of low-quality inbound links that appear to be the result of artificial efforts by the publisher to influence search rankings, the site’s rankings can be lowered. This is, in fact, exactly what Google’s Penguin algorithm does. Some other potential negative ranking factors include:
Malware being hosted on the site
The search engines will act rapidly to penalize sites that contain viruses or Trojans.
Search engines want publishers to show the same content to the search engine as is shown to users.
Google has a strong policy against paid links, and sites that sell links may be penalized.
As an extension of the prior negative ranking factor, promoting the sale of paid links may be a negative ranking factor.
Back in 2010, Google’s Matt Cutts announced that Google was making page speed a ranking factor. In general, it is believed that this is a negative factor for pages that are exceptionally slow.
Other Ranking Factors
The ranking factors we’ve discussed so far are really just the basics. Search engines potentially factor in many more signals. Some of these include:
If, over time, your site has acquired an average of 5 links per day, and then the links suddenly start to come in at a rate of 10 per day, that could be seen as a positive ranking signal. On the other hand, if the rate of new links drops to 2 per day, that could be a signal that your site has become less relevant.
However, it gets more complicated than that. If your site suddenly starts to get 300 new links per day, you have either become a lot more relevant or started to acquire links in a spammy way. The devil is in the details here, with one of the most important details being the origin of those new links. The concept of considering temporal factors in link analysis is documented in a U.S. patent held by Google, which you can look up by searching for patent number 20050071741.
Personalization is one of the most talked-about frontiers in search. There are a few ways in which personalization can take place. For one, a search engine can perform a geolocation lookup to figure out where a user is approximately located. Based on this, the search engine can show results tailored to a user’s current location. This is very helpful, for example, if the user is looking for a local restaurant.
Another way a search engine can get some data on a user is if he creates a profile with the search engine and voluntarily provides some information. A simple example would be a language preference. If the user indicates he prefers Portuguese, the search engine can tailor the results to that preference.
Search engines can also look at the search history for a given user. Basically, the search engine maintains a log of all the searches the user has performed when he is logged in. Based on this, it can see that he has been checking out luxury cars recently, and can use that knowledge to tweak the results he sees after he searches on jaguar. This is sometimes referred to as adaptive search.
To reduce the level of personalization, users can log out of their Google account. However, this does not disable all personalization, as Google may still tie some history to the person’s computer. A user can disable all personalization by using Google’s Chrome browser in Incognito mode. This will allow her to see Google results that are not personalized based on search history. However, the results will still be personalized to her location.
A user can also depersonalize search results by performing her search query, and then appending &pws=0 to the end of the search page URL and reloading the page. Note, this works only if she has turned off Google Instant (Google’s feature of showing results instantly as the user types). Or, the user can choose the option “Disable customizations based on web history” under “webhistory” under the gear icon in the SERPs.
Using Advanced Search Techniques
One of the basic tools of the trade for an SEO practitioner is the search engines themselves. They provide a rich array of search operators that can be used to perform advanced research, diagnosis, and competitive analysis. The following are some of the more basic operators:
Excludes the keyword from the search results. For example, loans – student shows results for all types of loans except student loans.
Shows search results for the exact phrase — for example, “seo company”. You can also use “” to force the inclusion of a specific word. This is particularly useful for including stopwords (keywords that are normally stripped from a search query because they usually do not add value, such as the word the) in a query, or if your keyword is getting converted into multiple keywords through automatic stemming. For example, if you mean to search for the TV show The Office, you would want the word The to be part of the query. As another example, if you are looking for Patrick Powers, who was from Ireland, you would search for “patrick powers” Ireland to avoid irrelevant results for Patrick Powers.
keyword1 OR keyword2
Shows results for at least one of the keywords — for example, google OR Yahoo!.
These are the basics, but for those who want more information, what follows is an outline of the more advanced search operators available from the search engines.
Advanced Google Search Operators
Google supports a number of advanced search operators that you can use to help diagnose SEO issues.
Force an exact-match search. Use this to refine results for ambiguous searches, or to exclude synonyms when searching for single words.
Example: “steve jobs”
Search for X or Y. This will return results related to X or Y, or both. Note: The pipe (|) operator can also be used in place of “OR.”
Search for X and Y. This will return only results related to both X and Y. Note: It doesn’t really make much difference for regular searches, as Google defaults to “AND” anyway. But it’s very useful when paired with other operators.
Example: jobs AND gates
Exclude a term or phrase. In our example, any pages returned will be related to jobs but not Apple (the company).
Example: jobs ‑apple
Acts as a wildcard and will match any word or phrase.
Example: steve * apple
Group multiple terms or search operators to control how the search is executed.
Example: (ipad OR iphone) apple
Search for prices. Also works for Euro (€), but not GBP (£) ?
Example: ipad $329
A dictionary built into Google, basically. This will display the meaning of a word in a card-like result in the SERPs.
Returns the most recent cached version of a web page (providing the page is indexed, of course).
Restrict results to those of a certain filetype. E.g., PDF, DOCX, TXT, PPT, etc. Note: The “ext:” operator can also be used—the results are identical.
Limit results to those from a specific website.
Find sites related to a given domain.
Find pages with a certain word (or words) in the title. In our example, any results containing the word “apple” in the title tag will be returned.
Similar to “intitle,” but only results containing all of the specified words in the title tag will be returned.
Example: allintitle:apple iphone
Find pages with a certain word (or words) in the URL. For this example, any results containing the word “apple” in the URL will be returned.
Similar to “inurl,” but only results containing all of the specified words in the URL will be returned.
Example: allinurl:apple iphone
Find pages containing a certain word (or words) somewhere in the content. For this example, any results containing the word “apple” in the page content will be returned.
Similar to “intext,” but only results containing all of the specified words somewhere on the page will be returned.
Example: allintext:apple iphone
Proximity search. Find pages containing two words or phrases within X words of each other. For this example, the words “apple” and “iphone” must be present in the content and no further than four words apart.
Example: apple AROUND(4) iphone
Find the weather for a specific location. This is displayed in a weather snippet, but it also returns results from other “weather” websites.
Example: weather:san francisco
See stock information (i.e., price, etc.) for a specific ticker.
Force Google to show map results for a locational search.
Example: map:silicon valley
Find information about a specific movie. Also finds movie showtimes if the movie is currently showing near you.
Example: movie:steve jobs
Convert one unit to another. Works with currencies, weights, temperatures, etc.
Example: $329 in GBP
Find news results from a certain source in Google News.
Example: apple source:the_verge
Not exactly a search operator, but acts as a wildcard for Google Autocomplete.
Example: apple CEO _ jobs
Here are the ones that are hit and miss, according to my testing:
Search for a range of numbers. In the example below, searches related to “WWDC videos” are returned for the years 2010–2014, but not for 2015 and beyond.
Example: wwdc video 2010..2014
Find pages that are being linked to with specific anchor text. For this example, any results with inbound links containing either “apple” or “iphone” in the anchor text will be returned.
Example: inanchor:apple iphone
Similar to “inanchor,” but only results containing all of the specified words in the inbound anchor text will be returned.
Example: allinanchor:apple iphone
Find blog URLs under a specific domain. This was used in Google blog search, but I’ve found it does return some results in regular search.
Example: blogurl:microsoft.comSIDENOTE. Google blog search discontinued in 2011
Find results from a given area.
Example: loc:”san francisco” appleSIDENOTE. Not officially deprecated, but results are inconsistent.
Find news from a certain location in Google News.
Example: loc:”san francisco” appleSIDENOTE. Not officially deprecated, but results are inconsistent.
Here are the Google search operators that have been discontinued and no longer work. ?
Force an exact-match search on a single word or phrase.
Example: jobs +appleSIDENOTE. You can do the same thing by using double quotes around your search.
Include synonyms. Doesn’t work, because Google now includes synonyms by default. (Hint: Use double quotes to exclude synonyms.)
Find blog posts written by a specific author. This only worked in Google Blog search, not regular Google search.
Example: inpostauthor:”steve jobs”SIDENOTE. Google blog search was discontinued in 2011.
Similar to “inpostauthor,” but removes the need for quotes (if you want to search for a specific author, including surname.)
Example: allinpostauthor:steve jobs
Find blog posts with specific words in the title. No longer works, as this operator was unique to the discontinued Google blog search.
Example: intitle:apple iphone
Find pages linking to a specific domain or URL. Google killed this operator in 2017, but it does still show some results—they likely aren’t particularly accurate though. (Deprecated in 2017)
Find information about a specific page, including the most recent cache, similar pages, etc. (Deprecated in 2017). Note: The
id: operator can also be used—the results are identical.SIDENOTE. Although the original functionality of this operator is deprecated, it is still useful for finding the canonical, indexed version of a URL. Thanks to @glenngabe for pointing this one one!
Find results from a certain date range. Uses the Julian date format, for some reason.
Example: daterange:11278–13278SIDENOTE. Not officially deprecated, but doesn’t seem to work.
Find someone’s phone number. (Deprecated in 2010)
Example: phonebook:tim cook
Searches #hashtags. Introduced for Google+; now deprecated.
More Advanced Search Operator Techniques
You can also use more advanced SEO techniques to extract more information.
Determining keyword difficulty
When you are building a web page, it can be useful to know how competitive the keyword is that you are going after, yet this information can be difficult to obtain. However, there are steps you can take to get some idea of how difficult it is to rank for a keyword. For example, the intitle: operator shows pages that are more focused on your search term than the pages returned without that operator (e.g., intitle:”dress boots”).
You can use different ratios to give you a sense of how competitive a keyword market is (higher results mean that it is more competitive). For example:
dress boots (108,000,000) versus “dress boots” (2,020,000) versus
intitle:”dress boots” (375,000)
Ratio: 108,000/375 = 290:1
Exact phrase ratio: 2,020/37 = 5.4:1
Another significant parameter you can look at is the inanchor: operator — for example, inanchor:”dress boots”. You can use this operator in the preceding equation instead of the intitle: operator.
Using number ranges
The number range operator can help restrict the results set to a set of model numbers, product numbers, price ranges, and so forth. For example:
Unfortunately, the number range combined with inurl: is not supported, so the product number must be on the page. The number range operator is also
great for copyright year searches (to find abandoned sites to acquire). Combine it with the intext: operator to improve the signal-to-noise ratio — for example, intext:”copyright 1993..2011” -2014 blog.
Using advanced doc type search
The filetype: operator is useful for looking for needles in haystacks. Here are a couple of examples:
confidential business plan -template filetype:doc forrester research grapevine filetype:pdf
Determining listing age
You can label results with dates that give a quick sense of how old (and thus trusted) each listing is; for example, by appending the &as_qdr=m199 parameter to the end of a Google SERP URL, you can restrict results to those within the past 199 months.
Uncovering subscriber-only or deleted content
You can sometimes get to subscriber-only or deleted content from the Cached link in the listing in the SERPs (found under the down arrow after the URL in the search listing) or by using the cache: operator. Don’t want to leave a footprint? Add &strip=1 to the end of the Google cached URL. Images on the page won’t load.
If no Cached link is available, use Google Translate to take your English document and translate it from Spanish to English (this will reveal the content even though no Cached link is available):
The related: operator will look at the sites linking (the linking sites) to the specified site, and then see which other sites are commonly linked to by the linking sites. These are commonly referred to as neighborhoods, as there is clearly a strong relationship between sites that share similar link graphs.
Finding Creative Commons (CC) licensed content
Use the as_rights parameter in the URL to find Creative Commons licensed content. Here are some example scenarios to find CC-licensed material on the Web:
Permit commercial use http://google.com/search?as_rights=
Permit derivative works http://google.com/search?as_rights=
Permit commercial and derivative use
Make sure you replace <KEYWORDS> with the keywords that will help you find content that is relevant to your site. The value of this to SEO is indirect. Creative Commons content can potentially be a good source of content for a website. An easier option if you don’t need this same freedom in your Creative Commons searches is to use Google’s Advanced Search page, where you can specify your Creative Commons license type.
Vertical Search Engines
Vertical search is a term sometimes used for specialty or niche search engines that focus on a limited data set. Examples of vertical search solutions provided by the major search engines are image, video, news, and blog searches. These may be standard offerings from these vendors, but they are distinct from the engines’ general web search functions.
Vertical search engines sometimes come in the form of specialty websites, such as travel sites (such as TripAdvisor), and local business listing sites (such as YellowPages.com). Any site that focuses on vertically oriented niche markets could be considered a vertical search engine.
Vertical search results can provide significant opportunities for the SEO practitioner. High placement in these vertical search results can equate to high placement in the web search results, often above the traditional 10 blue links presented by the search engines.
Vertical Search from the Major Search Engines
The big three search engines offer a wide variety of vertical search products.
Here is a partial list:
Google Maps, Google Images, Google Shopping, Google Blog Search, Google Video, Google News, Google Custom Search Engine, Google Book Search
Yahoo! News, Yahoo! Local, Yahoo! Images, Yahoo! Video, Yahoo! Shopping, Yahoo! Autos
Bing Images, Bing Videos, Bing News, Bing Maps
All three search engines offer image search capability. Basically, image search engines limit the data that they crawl, search, and return in results to images. This means files that are in GIF, TIF, JPG, PNG, and other similar formats. Figure 2-30 shows the image search engine from Bing.
Image search engines get a surprisingly large number of searches performed on them. Unfortunately, market data on these volumes is not often published, but according to comScore, more than 1 billion image searches are performed on Google Images per month. However, it is likely that at least that many image-related search queries occur within Google web search in the same timeframe. However, because an image is a binary file, it cannot be readily interpreted by a search engine crawler.
Search engines have had to historically rely on text surrounding the image, the alt attribute within the <img> tag, and the image filename. However, Google now offers a search by image feature that allows users to drag an image file into the Google Images search box and it will attempt to identify the subject matter of the image and show relevant results.
As with image search, video search engines focus on searching specific types of files on the Web — in this case, video files such as MPEG, AVI, and others. Figure 2-31 shows a quick peek at video search results from YouTube.
A very large number of searches are also performed in video search engines. YouTube is the dominant video search engine. Current data on total monthly searches is not readily available, but in June 2011, over 3.8 billion searches were performed on YouTube. This suggests that YouTube is the third largest search engine on the Web (Bing is larger when you consider the cumulative search volume of Bing and Yahoo!). As with image search, many video searches are also performed directly within Google web search.
You can gain significant traffic by optimizing for video search engines and
participating in them. Once again, these are binary files and the search engine cannot easily tell what is inside them.
This means optimization is constrained to data in the header of the video and on the surrounding web page.
However, each search engine is investing in technology to analyze images and videos to extract as much information as possible. For example, the search engines are experimenting with OCR technology to look for text within images, and transcription and other advanced technologies are being used to analyze video content. In addition, flesh-tone analysis is being used to detect pornography or recognize facial features. The application of these technologies is in its infancy and is likely to evolve rapidly over time.
Next up in our hit parade of major search verticals is local search (a.k.a. map search). Local search results are now heavily integrated into the traditional web search results, so a presence in local search can have a large impact on organizations that have one or more brick-and-mortar locations. Local search engines search through databases of locally oriented information, such as the
name, phone number, and location of local businesses around the world, or just provide a service, such as offering directions from one location to another. Figure 2-33 shows Google Maps local search results.
The integration of local search results into regular web search results has dramatically increased the potential traffic that can be obtained through local search. We will cover local search optimization in detail in “Optimizing for Local Search”.
The major search engines also offer a number of specialized offerings. One highly vertical search engine is Google Books search, which specifically searches only content found within books, as shown in Figure 2-35.
Bing also has some unique vertical search features. One of the more interesting ones is its product search solution. Instead of having a separate shopping search engine, Bing has integrated the product results into the main body of its search results, as shown on the right side of Figure 2-36.
Universal Search/Blended Search
Google made a big splash in 2007 when it announced Universal Search, the notion of integrating images, videos, and results from other vertical search properties directly into the main web search results. Prior to this announcement, all the search engines showed their search results in separate vertical search engines. You have already seen an example of this in Figure 2-36, which shows Bing’s way of integrating product search features directly into the main search results.
After Google’s announcement, both Bing and Yahoo! quickly followed with their own implementations. Each type of result you see on a search results page offers different opportunities for obtaining traffic from search engines. People now refer to this general concept as a blended search (because Universal Search is specifically associated with Google).
Figure 2-37 shows an example of blended search results from a Google search.
More specialized vertical search engines
Vertical search can also come from third parties. Here are some examples:
- Comparison shopping engines (e.g., PriceGrabber, Shopzilla, and Nextag)
- Travel search engines (e.g., Expedia, Travelocity, and Kayak)
- Real estate search engines (e.g., Trulia and Zillow)
- Job search engines (e.g., Indeed, CareerBuilder, and SimplyHired)
- Music search engines (e.g., iTunes Music Store)
- B2B search engines (e.g., KnowledgeStorm and ThomasNet)
There is an enormous array of different vertical search offerings from the major search engines, and from other companies as well. We can expect that this explosion of different vertical search properties will continue.
Effective search functionality on the Web is riddled with complexity and challenging problems. Being able to constrain the data types (to a specific type of file, a specific area of interest, a specific geography, etc.) can significantly improve the quality of the results for users.
Country-Specific Search Engines
At this stage, search is truly global in its reach. Google is the dominant search engine in many countries, but not all of them. How you optimize your website depends heavily on the target market for that site, and the one or more search engines that are the most important in that market.
According to comScore, Google receives 54.3% of all searches performed worldwide as of April 2014. In many countries, that market share is 80% or more.
Here is some data on countries where other search engines are major players:
China Internet Watch reported in September 2014 that Baidu had about 70% market share. This is significant because China has the largest Internet usage in the world, with 618 million users in 2010 according to China Internet Network Information Center.
According to figures reported by Yandex, the company’s market share in Russia comprised about 62% of all searches in April 2014.13
Naver was estimated to have about 70% market share in South Korea in March 2014.14
In January 2014, the Startup Yard blog reported that Seznam had more than 60% market share in the Czech Republic.15
Optimizing for Specific Countries
One of the problems international businesses continuously need to address with search engines is identifying themselves as “local” in the eyes of the search engines. In other words, if a search engine user is located in France and wants to see where the wine shops are in Lyon, how does the search engine know which results to show?
Here are a few of the top factors that contribute to international ranking success:
- Owning the proper domain extension (e.g., .com.au, .uk, .fr, .de, .nl) for the country that your business is targeting
- Hosting your website in the country you are targeting (with a country-specific IP address)
- Registering with local search engines:
- Google My Business
- Yahoo! Small Business
- Bing Places
- Placing your relevant local address data on major pages of the site
- Setting your geographic target in Google Search Console (you can read more about this at http://bit.ly/country_targeting); note that Google does not really need you to do this if your site is on a country code top-level domain (ccTLD), such as .de or .co.uk, as the preferred regional target is assumed.
All of these factors act as strong signals to the search engines regarding the country you are targeting, and will make them more likely to show your site as a relevant local result.