Three Ways to Improve External Search Engine Usability
Published on July 9, 2002
Most diligent webmasters spend time on log analysis, analyzing server logs to determine where visitors are coming from, which pages they're viewing, how long they're staying, and other significant demographic and technical information. One of the most important details that logs show is which search engines visitors use to get to the site, and which search terms visitors enter. Checking this information is an extremely useful, fascinating, and almost voyeuristic endeavor that has become a hobby in and of itself (see Disturbing Search Requests).
The usual procedure is to view the search logs, see which search engines are most popular, and find out which general subjects people are searching for. This is helpful but rarely is the analysis carried any further than generalizations about the most popular engines and words used in searches: "Of the people who come to our site via a search engine, 70% are coming from Google, and most use phrases like 'free software download' or 'download shareware.'"
This is great information, but two things are often missed. First, how the link to your site is presented in the results is critical. Next, what are visitors experiencing once they select that link? Many sites spend significant resources ensuring high placement in search engines, but usually little effort is spent on designing how those results are displayed and whether the pages they point to will help the visitor achieve their goal. With on-site search engines, it is easy to control the way results are presented to visitors, provide similar search suggestions, and ensure that the results are relevant and helpful. This is more difficult with external search engines, but not impossible. Here are some thoughts as to how to evaluate the usability of results on search engines and improve their effectiveness and relevance.
The Three Ways
There are three methods that can be used in improving how links to your site appears on external search engines, and how relevant and useful the resulting pages those links point to are:
- Recreating search logs
- Cognitive walkthrough
- Usability testing
Now for these concerns in detail.
Re-creating Search Logs
Search log re-creation is a simple process that sounds much like its name: Look at your search logs and re-create those same queries on the same search engines that visitors used to arrive at your site. So, if last month 30 people searched for "e-commerce white papers" at Yahoo and arrived at your site, you should now go to Yahoo, enter the exact same query, and look at the results. Then, ask yourself:
- Where is the link to your site?
- Is there more than one link?
- Do the links go to different pages?
- How does the link appear--is the <title> element used? Is the URL printed? Is there an "updated" date? Does it include a description of the site or page? Does this description come from meta tags or is it taken from the text on the page? Does it highlight your search terms?
- Based on the search query, is your link more or less relevant than other links in the results?
- When you click on the link to your site, what page appears? Does it bring you to the home page, to a product page, or to another page? Is it easy to tell where you are in the site? Can you easily navigate to other important pages or sections?
Perhaps you'll find that the search engine is using the <title> from your page as the link text, but it cuts off the text after 30 characters, and your <title> tag is 50 characters long. Maybe there are four links to your site but, from the information provided in the results, it's hard to tell which is the best link to choose. Or, it could be that the top result is a link to your site, but it's the "Contact Us" page, which doesn't have a description of the site, nor does it have any products or links to major pages listed.
By looking through your search logs, you'll probably find that people are arriving at your site by using search terms that you hadn't thought of. Maybe they're looking for MP3s of a certain band and finding a profile page of one of your employees, whose favorite bands are listed. Possibly they have a complaint about a company and are looking for a place to gripe, and it just so happens that you have a case study about the work you did for that company. The search terms people use and which pages they arrive on can help you understand what goals people are trying to accomplish (such as downloading MP3s or complaining about a company) and whether your site can help them reach their goal. In some cases your site can do so, and then you can make sure your design supports the task they are trying to accomplish. In other cases, there might not be anything you can do to help them, nor are you interested in catering to those customers or markets.
From this exercise, you should begin to understand what it's like for a visitor to come to your site for the first time through a result listed in a search engine. Search log analysis is useful because it allows you to quickly go through a large amount of actual user data to see what people are looking for that makes them visit your site. Of course, it only provides data about actual visitors, which could be very different from desired visitors, and not everyone coming to your site really wanted to come to your site. As the above examples show, people might be looking for something that can't be accomplished on your site, but because of a few key words on some pages, they may end up visiting. Realize that this traffic may not be beneficial to your site's goals, and in some cases could be a problem.
A cognitive walkthrough is the process of evaluating a user interface based on a specific set of tasks to determine how easy it is for someone to learn the interface. Usability specialists who are evaluating software interfaces most often use this technique, but the basic idea can be applied in many different situations. You need to go through the steps that a visitor would take to get to your site--go to a search engine, enter in a term that they would probably use, and see what results you get. Ask the same questions as when you recreated search logs: Is there a link to your site? Is it relevant to the search terms you entered? What information is provided along with the link? What page do you end up on when you click the link? Does it provide relevant information?
You will get the best results if your actions accurately reflect those of your users. If you've used personas as part of the development process (and you should), you can walk through the process in the shoes of one of your personas. Danielle, a 37-year-old stay-at-home mother new to the Internet might enter one or two words in the search box on MSN.com, because that was the preset homepage when she bought her computer. Alex, a 25-year-old computer programmer might enter more advanced searches using +/- modifiers on his tweaked Google toolbar.
Using personas is beneficial because you are not the user. Personas help remind you who the users actually are. Understand your users and know what search engines they use, and focus your efforts there. Also keep in mind that your search logs might not be telling you the whole story. You might believe that your users use Altavista more than any other search engine because, after analyzing your search logs, you found that most of the referrals are coming from Altavista. Well, it could be the case that your site isn't listed in any other search engines, so the only place people can find your site is through Altavista. Maybe your target audience overwhelmingly uses Yahoo and you're not listed there, so there is a large group of potential visitors who have never discovered your site.
The bottom line is that you need to identify your target audiences and use the techniques that they would use to try to get to your site. Use their language, their search engines, and their browsing habits. See if they'd be likely to find your site, and evaluate the impression they would have of it when they arrived there, presumably for the first time.
Recreating search logs and using personas to guide cognitive walkthroughs are useful techniques, but the best way to see not only if people are able to find your site through search engines but also what happens once they arrive on your site is to do usability testing. The basics and benefits of usability testing have been covered enough, so I won't bother going into details, but I will explain a technique I call the External Search Usability Test.
The process is almost the same as the two previous methods. In this case, instead of analyzing logs or walking through the process yourself, you're going to watch a user go through the process.
Have the tester go to their preferred search engine. Give them a generic task or topic to search for that should yield your site in the results. For example, if you sell typefaces: "You are putting together a family newsletter and want to find some new fonts to use." Or, if you have a site that features reviews of rock albums: "You want to get some information about the new They Might Be Giants album to see if you'd be interested in buying it." Now, of course, this is biased, since we're assuming that the person would choose to search for this information or even choose to use the Internet to help them complete their task. Keep this in mind, but understand that the purpose of this is only to observe their search behavior and determine how well your site performs in this context. Encourage the tester to talk through their thought process out loud. What search engine do they use? Why? What words or phrases do they type in? What are the results? Ask them to go to whichever site they think would help them complete their task. It's most helpful if they don't know which site is yours, so they don't feel pressured to choose it if it is in the results. Maybe they choose a link to your site, maybe to a competitor, or maybe they aren't happy with the results.
Don't force the tester, but make sure they know that you're more interested in the search results they get and what they think of them than in going deep into the sites that come up in the results. If they don't manage to get to your site on their own after they've tried a few different search terms, ask them to run a search that you know returns your site in the first page of results. Again, see if they choose the link to your site. If not, ask them why they chose the links that they did try, and if there was any reason they didn't choose to try a link to your site. Then, have them click the link to go to your site and ask what their impression is of the resulting page. Does it help them complete their task? Is it clear what the purpose of the site is? Is it clear where they are within the site and where else within the site they can go?
This technique should be repeated on several users. While there are differences of opinion within the usability community as to how many testers are needed for accurate results, in this case, 3-5 testers should give you enough data to expose any potentially major problems. However, search listings and content change, business strategies take on new directions, and users' preferred engines change, so it would be best to run a test on a regular basis, perhaps once every six months. This will allow you to allow for changes in your site and the search engines while keeping your site performing well in the results.
Gaining a Wider Audience
The three techniques above describe ways of analyzing the usability and effectiveness of external search engines. Once you have evaluated the results, the next step obviously would be to make changes to improve the quality and context of the results. This could mean submitting your site to search engines where it currently isn't listed; changing the <title> element content to more accurately reflect the content of the pages; altering copy on high-traffic pages to produce better results in search engines that quote text from your pages; improving the information architecture to make it easier to navigate from a low-level page to the homepage or a main section head. The analysis will point out problems, and it should be apparent as to how you can go about fixing those problems.
It is no longer enough to get visitors to your site. You must get the right visitors to your site for the right reasons! Using these three techniques to improve how your site appears in search engine results can help. They require little time and effort, and when combined with a comprehensive promotion strategy, including search engine placement, can increase traffic and make a site more efficient and effective at reaching the target audience.
Jeff Lash is a User Experience Designer in the Health Sciences division of Elsevier. He is a co-founder and Advisory Board member of the Asilomar Institute for Information Architecture (AIfIA) and has also written articles and tutorials for Boxes and Arrows and WebWord. His personal website is jefflash.com.