When Good SEO Becomes Bad Information Architecture
Have you ever heard the phrase search engine optimization (SEO) architecture? At first glance, it might seem like a good idea because the goal of SEO and information architecture (IA) is to make the products, services and information on your website easy to find. However, information architects often have greater insights into user/searcher mental models because they utilize two specific usability tests to determine these mental models: open and closed card sort tests.
Let’s examine both of these usability tests and how SEO professionals often misinterpret—or even ruin—the results of these tests.
Open card sort test
An open card sort test is a formative usability test in which users/searchers are presented with blank index cards (either the physical kind or the online version) and a list of items. Test participants are then asked to: (a) place items into groups, and (b) come up with labels for each of the groups. Here is a simple example.
Suppose the item on the first card is blue. Immediately, a test participant might think, “blue is a color.” The item on the second card is green. The user might think, “blue and green are both colors,” and will immediately place the blue and green cards next to each other as a group. The item on the third card is orange—another color. If the test participant is thinking out loud, you might hear them say the word “color.” You can immediately see an initial mental model and grouping: colors.
The item on the fourth card is peach, and a test participant might think, “A peach is a fruit.” On the blank cards, the test participant might write two groups: colors and fruit:
However, the test participant might realize after creating two possible ways of organizing the items, “an orange is a fruit, too.” And he might move the orange item into the fruit category label.
Which way of organizing these items is correct: 2 colors and 2 pieces of fruit, or 3 colors and one piece of fruit? The item on the next card is cherry, which the test participant immediately puts into the fruit category, but then the test participant might move orange back into the colors category.
But on second thought (or third or fourth), the test participant moves orange back into the fruit category.
“No no no!” the test participant says out loud. “They are all colors.” And he moves all of the items underneath the colors category and removes the fruit category altogether.
Please understand that this is a grossly oversimplified version of an open card sort test for organizing website content. But I hope this example illustrates that people organize and label information in multiple ways—and they often change their minds many times during the test. Now imagine a website with a very complex architecture: multiple taxonomies, cross-referencing and so forth. It takes a seasoned professional to analyze the complex data from this usability test and to truly construct a findable architecture based on analysis of that data. An SEO professional might not have these skills.
Where SEO professionals make open card sort mistakes: During this usability test, it can be very difficult to not ask “leading” questions that put keywords into a test participants’ head when that person might otherwise not have thought of that word. Furthermore, the items that are needed to categorize should be carefully selected (i.e. not necessarily from data from keyword research tools) so as not to lead users/searchers into validating the SEO professional’s mental model. During these types of usability tests, we want the mental models of users/searchers, not the mental models of a keyword research tool or an SEO professional.
Finally, and my esteemed colleague and fellow Search Engine Land writer Andy Atkins-Krüger can attest to this, users/searchers from different countries will label, group and prioritize information differently. So blindly keeping the same information architecture and corresponding navigation schemes from country to country could be, quite simply, a mistake.
I have found that the best information architectures and corresponding navigation designs emerge after both open and closed card sort usability tests are performed. Now, let’s look at the second type of card sort test.
Closed card sort test
A closed card sort test is a validative usability test in which users/searchers are presented with pre-labeled categories and a list of items. Test participants are then asked to place each item underneath one of the categories, the one that they feel is the most appropriate. There should always be a blank index card in the event that test participants come up with their preferred category label, and there should also be an “I don’t know” pile in the event that participants honestly do not believe that any of the pre-determined labels make sense to them.
Below is a very simplified version of a closed card sort test. Test participants are asked to move items in the far-left column into one of the two pre-determined categories.
Where SEO professionals make closed card sort mistakes: This is the usability test where I see SEO professionals making the most profound mistakes. Believe it or not, I observe keyword-stuffed information architectures all of the time. I see architectures with keywords that would never transform into usable site navigation. I frequently see undistinguishable labels on items. Test results are faulty because an SEO professional puts “leading” categories and “leading” item names to be sorted, all in the effort to validate his or her mental model.
In addition, no matter how many users/searchers participate in a closed card sort test, you might not see any trends in how users/searchers organize the content. This often occurs with websites that have poorly labeled—and poorly prioritized—primary navigation. If the closed card sort isn’t yielding meaningful results, chances are you will need to reexamine your information architecture and not allow the SEO professional’s input at this particular time.
One of the key components of both of these usability tests is objectivity. The goal of usability testing is not to validate the SEO professional’s mental model or the web designer/developer’s mental model. The goal is not to prove that SEO professionals or web designers/developers are right (or wrong). The goal is to observe, listen and learn user/searcher mental models and create a website that: (a) validates user mental models, and (b) accommodates searcher goals and behaviors.
IA testing tips for SEO professionals
Here is my short list of tips when determining effective website information architectures:
- With both the open and closed card sort tests, it is imperative that SEO professionals resist the urge to put words in users’ mouths. The point of usability tests for information architecture is to listen to users, and to objectively observe their behaviors and actions. You want to determine the best labels based on user/searcher mental models, not your own.
- The content that you put on the index cards (whether it’s physical index cards or the online version) is critical to the success of both usability tests. If you find that you cannot be objective in using keywords, then hire an information architect to come up with the item names.
- Sometimes, a navigation label is clear without a keyword. Resist the urge to add one for ranking reasons when the label is clear. Do not listen to the SEO Borg and think, “resistance is futile.” There are other important places on a web page that you can implement keywords… and the page can still get search engine visibility and conversions.
- Don’t imitate competitor architectures and site navigation because it makes sense to you and the competitor site ranks well. Kim Krause-Berg warned against this in her excellent article, It’s A Fatal Mistake To Copy Successful Websites. Sure it makes sense to you. And it appears to make sense to a commercial web search engine because it ranks. But it doesn’t make sense to users, and they are the ones who will ultimately purchase your products and services.
- Take off your search (querying) blinders. People who work in a “search” environment often see the “search” part of searcher behavior and not the “browse” part. Web searchers browse. They do not just type a keyword, click on a search listing, view a web page, and convert on the exact landing page. If their information scent is validated on the landing page, searchers will continue to browse a site to complete their tasks. As I said in Information Architects Are From Venus, SEOs Are From Mars, browsing and retrieval (searching) are equally important finding behaviors.
I would love to hear opinions of “SEO architectures” from professional information architects. What do SEO professionals get right? Where do they need more guidance? We all want to know.