The search acquisition funnel begins with understanding the searcher. Searchers themselves provide very little that indicates their intent. Search queries are an average of three words long and nearly 25 percent are only one word.
Fortunately, we have a great deal of data beyond the queries themselves to help us derive potential intent. Search engines track searcher behavior and by looking at what searchers click on and what they search for next, search engines can learn a great deal about what someone might mean by those one to three words.
We can see much of what the search engines have learned just by looking at what content search engines rank highly for a particular query.
As we’ve learned, search engines also make a great deal of search data, demographic data, and psychographic data available that can provide detailed insight into the intent and motivation of potential customers.
One way search engines derive intent is by classifying queries into intention categories. Each search engine group queries slightly differently, but for our purposes, a good classification system is:
- Navigational—With this type of query, the searcher is looking for something specific. These are often one word queries that have a high likelihood of a single meaning. Many, but not all, navigational queries are branded. Some examples are [Volvo] [amazon.com] and [Twitter Vanessa fox]. Approximately 18 percent of search queries are navigational.
- Commercial (also called transactional)—These searches are likely purchase-related and the search engines tend to favor results that enable the searcher to buy something, such as e-commerce sites. Some examples of this type of search include [buy shoes online] and [book vacation rental in Greece].
- Informational (also called research)—This query tends to be more generic and non-commercial. Search engines tend to favor results that are not commercial in nature. Examples include ‘‘when was Abraham Lincoln born’’ and ‘‘highest mountain in the world.’’
- Prepurchase research—For our purposes, it’s useful to consider this subset of informational queries in which the searcher is doing research with a high likelihood of a later purchase. Some examples of this type of search include ‘‘digital camera reviews’’ and ‘‘what cars get the best gas mileage.’’ This category of searchers is useful to seek out because it’s an audience who is not yet ready to purchase but will be soon and is currently gathering data to make decisions about what to buy. You want to provide information to these searchers so they later include you in their list of purchase choices.
- Action—This type of query signals that the searcher wants to do something, such as download an application or watch a video. I performed an action query just yesterday when riding in the car with my five-year-old niece. Results for [funny cat videos] on a mobile phone browser can provide hours of entertainment. This type of search can be considered a subset of informational queries.
As noted, search engines determine intent in part based on past search behavior. Search engines see millions of searches a day, and, over a period of years, this data can be very valuable in determining likely intent.
For instance, if one million people a day search for [Britney Spears] and 80 percent of those searchers clicked on a search result that linked to a video or did a second search for [Britney Spears video] (called query refinement), then the search engine looking at this data might conclude that generally, searchers looking for Britney Spears are actually looking for Britney Spears videos. To provide the most useful results for those searchers, the search engine might start surfacing videos in the search results even though the query alone didn’t indicate that desire.
How is this useful to you? If you run a Britney Spears site, you can increase your coverage by adding videos to your site so your site has the opportunity to rank not only for the Web page, but for videos as well.
Most searches are short because of what economist and psychologist Herbert Simon called ‘‘bounded rationality.’’3 We exert the smallest effort (short queries) to get an adequate result. We only exert more effort (query refinement) if the initial work didn’t get us what we needed.
Google’s quality guidelines give an example of how difficult it can be to determine intent from a single-word query:
Someone searching for [calendar] could be looking to find and print a calendar for the current month. Or they could be looking for a calendar of holidays. Or they could be looking for an online calendar to use for appointments.
Google breaks the query interpretation into three categories:
- Dominant interpretation
- Common interpretations
- Minor interpretations
Some of the factors that Google uses to determine intent include:
- Current events—the searcher likely wants the latest of something (such as latest sports scores versus historical ones).
- Location—if the query is regional in nature but doesn’t include a location, the searcher likely is interested in their current location (such as local pizza restaurants versus national ones).
- Previous behavior—Google uses both overall searcher behavior in aggregate and individual searcher behavior to predict current intent.
Intent factors heavily into how search results are generated globally and in personalized results
The easiest way to find out what search engines have learned about searcher intent is to conduct searches and evaluate the results. Take the query [iPod Touch], for instance. The results include not only the official Apple site and sites that describe the iPod Touch, but also video results, news results, and shopping results
These results are a clear indication that many searching for [iPod Touch] went on to view videos, news, and e-commerce sites. The suggested related searches tell us that many did subsequent searches for the third generation iPod Touch and the model with a camera.
Query refinements are the queries searchers type in after doing an initial search. They might do these refinements right away (without clicking on results) or after visiting a few pages and then returning to the search engine. Searchers often enter refinements when the initial search doesn’t provide satisfactory results, but the results do provide clues on what to search for.
For instance, if someone does a search for a high school name and gets a set of results with that high school name, but not the correct one, the searcher might notice that the results that appear have geographic information and add that to the query. [South High School] (in Bakersfield, CA), might refine to [South High School Bakersfield CA].
Why do searchers refine their queries? An August, 2009 Hitwise study about Canadian searchers found that searches were successfullyonly 70 percent of the time and the rest of the time searchers had to ‘‘re-search to find relevant results.’’
In this particular study, Hitwise found that the lack of relevance might stem from the international nature of the searches. Since the searches were conducted in North America in English, search engines seemed to weigh U.S. sites more heavily. So, for instance, a search for a particular retail store in Canada might produce results for U.S. store locations.
A Penn State researcher found that 22 percent of queries are refined. At the beginning of a search, searchers narrow their queries and then as the search progresses, searchers reformulate queries (often based on what results appear).
A Microsoft Bing internal study found that searchers refine their queries, bounce back to the search results, or abandon the search 50 percent of the time. The breakdown of that 50 percent is shown in Figure 3.2.
Microsoft also found that searchers repeated 24 percent of their queries during a session. Forty-one percent of searchers change their search term (or search engine) if they don’t find what they’re looking for on the first page of results, and 88 percent do so after three pages.
During information retrieval, searchers often change focus to different aspects of their tasks.7 This change may cause searchers to refine
their queries to broader or narrower concepts, from topical to non-topical searches, and to unexpected shifts.
The search engines try to help searchers narrow in on what they’re looking for and to provide refinement suggestions in a number of ways.
It can be valuable to make note of what refinements the engines suggest, both because it sheds light on what searchers are likely looking for (since refinements tend to be based on past search behavior) and because searchers will often choose the refinement in the initial search. It may be difficult to rank for the initial term due to a lot of competition, but you might be able to more easily rank for the longer term.
- Google search suggest: Google provides refinement suggestions as the searcher is typing. For a company that sells used cars online, it may be helpful to know that when someone searches for [cars], Google suggests sports cars, old cars, used cars, and cars for sale. The company may have included information on their site about used cars, but perhaps didn’t think about mentioning ‘‘old’’ cars (see Figure 3.3).
- Google related searches: Google’s related searches can also help businesses understand their audience and ensure they can be found for relevant searches. Again, with [cars], we see that Google is suggesting [old cars] as a follow up search, so chances are good that a lot of people looking to buy a used car will click on the old cars link, so you want to make sure that your used car site ranks highly for that search (see Figure 3.4).
- Yahoo! search suggestions: Yahoo! provides a similar set of suggestions in their search box (see Figure 3.5).
- Yahoo! search assist: Yahoo! also provides a way for searchers to browse related concepts. This ‘‘search assist’’ feature combines search and browse to help searchers who might otherwise type in just one or two words. The related concepts listed here can provide great insight into searcher intent (see Figure 3.6).
- Bing categorized search: Microsoft Bing’s categorized search results go one step farther. Not only do the search results suggest refinements, but the page is also broken up into listings for each of them. So a search for [used cars] will not only produce results
Using Demographic Data
Microsoft adCenter provides demographic information about searchers based on the search patterns of MSN users. For instance, when looking at shoe searches, we see that 75 percent of searches for [Payless shoes] are from women, whereas men and women search for [running shoes] in nearly equal numbers (with the 25–35 age group searching the most at over 28 percent).
If we combine search volume and gender data, we can generate a shoe-related top 10 keyword list for both men and women (see Figures 3.8 and 3.9). By volume, the top three shoe-related terms men search for are [running shoes], [Jordan shoes], and [Nike shoes]; the top three shoerelated terms women search for are [Payless shoes], [running shoes] and [DSW shoes].
This information could be highly useful to Nike. Their site currently isn’t on the first page of Google search results for [running shoes] (see Figure 3.10). The top results are runningshoes.com, roadrunner sports.com, runningwarehouse.com, runnersworld.com, and holabird sports.com. Their competitors Saucony, Brooks, and New Balance do rank on the first page, and the first Nike-related result is a site called bestbuynike.com.
In July 2009, 1,830,000 people searched Google for running shoes. Nike would probably like a share of those searchers.
If searches for your product are done primarily by one gender, consider how they process language and shop differently. Researchers from Northwestern University and the University of Haifa found that the areas of the brain associated with language work harder in girls than in boys. So, perhaps copy aimed at men should be succinct and to the point.
In Why We Buy: The Science of Shopping, Paco Underhill noted that fewer men look for price tags.8 And some research has found that men are more likely tomake impulse purchases online. You can incorporate this data into other audience analysis and market research that you have and may decide that your product pages aimed at women should have highly visible prices and ensure information that answers questions is easily available.Studies have also shown that online, women are more engaged with images and men are less patient. This means that men are more likely to abandon your home page if it starts loading a Flash module. Men are also more impatient with poor navigational structure.
A study that focused on differences in how men and women search found that ‘‘on average, men make decisions quicker, spend less time on sites, are more likely to have preestablished ‘favored’ vendor sites that they use in the search process and show less resistance to sponsored listings. Women tended to be more deliberate in reading search results, spend more time with their searches and spend more time on sites before making decisions.’’
Consider the age of your target audience as well. If you cater to an older audience, make sure that when they click the search results to your site, the type is larger and clear and easy to read.
How Searchers Interact with Results
Gord Hotchkiss, CEO of Enquiro Research, has given us fascinating insight into the way searchers view and click on results.
He describes how our brains use cognitive shortcuts when performing mental tasks. As we learn a behavior, we internalize it and perform that behavior on autopilot so our brains are free to concentrate on other tasks. Up to 45 percent of our daily actions are done by habit, without conscious thought.11 Our brains take similar shortcuts when scanning the search results page. We don’t initially read the text on the page; we look for matches in the shape of the characters in the query. Hotchkiss notes:
In cognitive psychology, this is called the ‘‘pop out’’ effect. We can recognize shapes much faster than we can read words. The shapes of our query literally ‘‘pop out’’ from the page as a first step toward matching relevance. The effect is enhanced by query (or hit) bolding. This matching game is done at the sub-cortical level.
How Working Memory Plays into Evaluating Search Results
Enquiro’s eye tracking studies have shown that we scan the page in an F-shaped pattern: we start in the upper left and then move down the left margin (see Figure 3.11).
Because of the capacity of our working memory, we break the page into chunks of three to four results. This corresponds to cognitive psychologist George Miller’s rule of ‘‘7 þ/ 2,’’ which posits that our working memory can hold around seven chunks of information.13 We can hold more digits than words in our working memory, which may partially explain how we evaluate search results. A 2008 study at UC Davis found that our working memory only allows us to consider three to four things at one time.
Rather than evaluate all results on the page at once, we first evaluate a single three to four item chunk and move on to the next if we don’t find a match. Fifty percent of the time, searchers click on results in the first chunk.
This is one reason why a number one ranking is less important than people think. If you rank in the top three to four results and have the most compelling title and description, you may win the click over the sites ranking above you.
Why don’t we feel compelled to review the results in all of the chunks before making a decision? It likely has to do, once again, with
bounded rationality. If we reach an adequate solution within the first chunk, we have no reason to continue our evaluation.
All of this chunking and scanning takes place in milliseconds. Once we see a positive pattern match in a result, our eyes rest on that result long enough to read the words. When we see words that reinforce that this result is relevant for the query (we may recognize the brand) or see a description includes positive attributes we were looking for but didn’t include in the query (such as a regional location or price attribute), we may click.
One Enquiro study involved finding out if searchers with different intents (commercial versus information gathering) scanned results differently. They showed both groups the same set of algorithmic and sponsored results. They found that both groups scanned the results in nearly the same way. The difference happened at the click. In the commercial intent group, half of the searchers clicked the algorithmic results and half clicked the paid search results. In the information gathering group, everyone clicked the algorithmic results. This happened even though both groups looked at the algorithmic and paid search results in exactly the same way and for the same amount of time.
Our search autopilot only switches off once we are ready to click on a result. The entire interaction generally takes less than ten seconds. We spend less than two seconds evaluating each result before the click.
So if we scan results a chunk at a time and evaluate each cluster of results, what are we evaluating? What is displayed in that result can make a big difference in whether that listing gets a click.
What Compels a Click?
We subconsciously look for relevancy clues in the title and snippet once we’ve paused on a result. For instance, if we’re looking for a car, we might respond positively to words like ‘‘fuel efficient’’ and ‘‘5 star crash rating.’’
Some businesses don’t like seeing their competition rank highly in the results along with them. But this might not be such a bad thing. One study found that when a well-known competitor was missing from results, searchers found the results to be less trustworthy and didn’t feel as comfortable clicking
How Important Is a Number One Ranking?
As we discovered earlier, if your search result is more compelling than those around it, searchers may very well click on it if it’s within the cluster they’re evaluating, even if it isn’t ranked highest. But ranking well does make a difference. Sixty-two percent of searchers click a result on the first page of results and 90 percent click within the first three pages.17 Data leaked from AOL shows that first page rankings are even more important, as 90 percent of clicks within that data set were on the first page of results.
Breakdown of the Search Results Page
According to a study by Bernard J. Jansen at Penn State University, 60 percent of searchers rely primarily on the title of the result when evaluating whether to click.
And as we saw earlier in the eye tracking heat maps, searchers don’t even evaluate the entire title—they look primarily at the left half of it. The description displayed under the title and the URL also factor into what a searcher clicks on. Search engines bold words in all three places that match the query, which can draw a searcher’s attention.
In the search results below for the query [rheumatoid arthritis], searchers may skip the second result when scanning, since the words that match the query don’t appear in the left side of the title (see Figure 3.12).
Many businesses don’t think about optimizing the description listed under the title of their search results, but that is your one opportunity to provide a compelling marketing message and entice potential customers to click.
Even worse than not optimizing the title and description of your listing is not providing them at all. For instance, see the search results display of some high profile brands that have spent considerable ad dollars on offline campaigns that have spurred search volume.
With this Sprint listing, for instance, no description appears and the title is less than helpful (see Figure 3.13).
And here, the text that compels searchers to click is from an Atlas tracking code (see Figure 3.14).
Verizon doesn’t do much better (see Figure 3.15).
Enhancing Your Results
We can see that descriptive, compelling search results can help motivate a click, and you can take further advantage of how the search engines are evolving the way these results are displayed.
Yahoo!, for instance, has a program called SearchMonkey that enables Web developers to include additional details such as photos, links, ratings, and store hours to the listing. Google has begun extracting metadata from pages to display similar information with a program it
calls Rich Snippets. Bing extracts details from the Web page and displays them in a pop up that appears when the searcher hovers the mouse over the listing.
Amit Kumar, of Yahoo!’s SearchMonkey, states, ‘‘Our tests uncovered that users found these [SearchMonkey] apps useful; in fact, in some cases, we saw a lift in click-through rate of as high as 15 percent.’’
How Searchers Evaluate the Page They’ve Clicked On
We can judge a site visually in as little as 50 milliseconds.21 And when performing a task, we focus on that and become blind to anything else. Even if visitors to a site look directly at something, they may not see it at all if it doesn’t obviously apply to their task.
This behavior correlates well with customer behavior in a store. In Why We Buy, Paco Underhill describes how shoppers enter stores. They don’t stop at the front door and look around. They walk in and scan as they go. Much of what they see during this transition is lost on them.
Even in physical stores, shoppers scan words quickly. ‘‘Putting a sign that requires 12 seconds to read in a place where customers spend four seconds is just slightly more effective than putting it in your garage,’’ says Underhill.
Product placement can help maximize sales both in stores and on your site. The salsa will sell better next to the chips than next to the mustard.
For the purpose of search strategy, it’s important to understand that what page the searcher lands on and how easily the searcher can scan the page and recognize that it answers their query is a vital part of the search acquisition funnel.
Consider the query [what do Peace Corps volunteers do]. It leads searchers to a page that clearly answers the searcher’s question. The context and purpose of the page are clear on even a quick scan (see Figure 3.16).
From Page to Conversion
Once the searcher lands on your page and confirms the content will satisfy the search, make sure the page will also fulfill the business need and not simply lead to a dead end. In the next chapter, we’ll dive into the details of developing searcher personas, which include pinpointing company goals and ensuring that those are presented as clear calls to action on the page. In the case of the Peace Corps example in Figure 3.16, we can see that the page has a clear and obvious call to action with the Apply Now button.
Consider this PEAK6 recruiting page, on the other hand. It not only doesn’t provide a clear heading to anchor the visitor about the context of the page, but the primary business purpose for creating the page at all (gathering job applicants) is hidden in a tiny link at the bottom of the page that is easily overlooked (see Figure 3.17).
An effective way to intersect searcher interest and business needs to identify key audience segments is by building searcher personas. You can then map out a compelling conversion path for these personas by way of searcher conversion workflows. You can use these workflows to determine key points of collaboration between departments of your organization.