Good information architecture is essential to providing a good UX and to accomplishing your business goals. But how can you tell if an IA is “good”? Start by looking at your analytics data.

Analytics systems keep a record of all the actions your users take, and reviewing behavior patterns reflected in this data can direct your attention to confusing, missing, or underperforming content categories that should be optimized or further investigated during user testing. Analytics data is especially useful when you have a large collection of content. Whether it’s blog posts, products, or support articles, presenting your content in categories that are both understandable and interesting makes users more likely to browse (and also improves your search rankings).

Homepage of Garden Design Magazine
Any design which relies on browsing to engage users can use analytics data to optimize topic menus. For example, this Gardendesign.com site has 3 sets of categories which could be analyzed by reviewing analytics data: the top global menu, the left-side topic navigation list, and the category pages featured in the body of the homepage.

Analytics Metrics That Suggest IA Problems

In order to use analytics data to improve information architecture, you must:

  • Identify which analytics metrics to consider
  • Interpret each metric according to the context of the overall design, user goals, and business goals.

Here are 5 analytics metrics that suggest a problem with your IA categories, and the follow-up questions you should ask to better understand your data. This interpretation step is critical, because some niche categories successfully serve important user goals, business objectives, or search engine optimization (SEO) strategies even though they exhibit these warning signs.

1.  Low Traffic to Categories

The volume of traffic to a category is the most obvious indicator of how useful or interesting the category is to your audience. Traffic can be measured as the number of views of the main category page (e.g., www.gardendesign.com/containers/), or by aggregating views of all the pages within a category (e.g., “/containers/shade.html,” “/containers/combos.html,” etc.). Note that different tools often use slightly different terminology to describe the same metric; the traffic to a particular page may be called ‘pageviews’, ‘visits’, or some other term depending on which analytics tool you are using.

Make sure to exclude repeat views by the same user. The overall traffic volume may be drastically inflated by users pogo-sticking, or repeatedly browsing back and forth from the category page to articles. Most analytics systems provide a separate metric which only considers unique views, and does not count repeated views by the same user in the same session.

How to interpret low traffic: Consider this metric in relation to the strategic importance of the category for the business and for users.

  • Is the traffic comparable with traffic to other categories? Calculate the total traffic to all categories, then the average traffic per category, and finally the ratio between each category’s traffic and this average. If the Container Gardening category only received 5% of the average traffic per category, it may be a topic that is irrelevant to most of your audience.
  • Are there other factors (such as page layout) that explain traffic disparities? Some categories may have higher traffic not because they are better categories, but simply because they are more visually noticeable, or appear multiple times. Don’t compare traffic between categories with significantly different prominence.
3 different duplicate categories on the Garden Design homepage
On Gardendesign.com, three of the categories from the left navigation are featured in the body of the page with attractive images. The analytics are likely to show much higher traffic to those three duplicate categories. These numbers would not necessarily reflect more interest in those topics; instead these categories’ popularity might be caused by the visual design. Proceed with caution before making changes to IA if the visual design might be heavily biasing the analytics data.
  • Would you expect higher traffic based on the importance or relevance of the topic? Sometimes mission-critical categories don’t get the traffic you would expect. In those cases, it’s worth trying alternative names for those categories to increase discoverability and findability.
  • Is the category strategically important, even though it has low usage? Sometimes the nature of a category is that it be used only once, or rarely. Or it can be used only by a small subset of valuable users. If that’s the case, then it’s probably worth keeping that topic around even though it only appeals to a small audience.

If the answer to all these questions is ‘no’, then consider eliminating or de-emphasizing the category. There’s an opportunity cost to having topics that nobody is interested in: instead, you could use the screen space to direct more attention to topics that would increase engagement or expand your audience.

2.  Low Conversions

Conversions represent desirable user actions (such as purchases or signups), and in most analytics systems must be manually configured as goals. You may find that some high-traffic categories have relatively few conversions. This suggests that those categories are less strategically important than you might assume based only on the raw number of visits and entrances.

How to interpret low conversions: Before making a decision based on low conversions, understand exactly what is being counted as a conversion, and look for other signals of value.

  • Is the category a significant source of traffic to other important pages? Even if a page itself does not convert, it may drive traffic to other pages that do convert (or serve other important functions in the user journey). For example, the How-to Info category page on gardendesign.com doesn’t generate many magazine subscriptions. But How-to could still be an important section if it is the main source of traffic to the Product section, which in turn generates significant affiliate marketing revenue.
  • Is the content part of a longer user journey, which requires several visits before completing a goal? If so, the category may influence visitors who later convert, but the connection may not be measured accurately by the analytics system if your metrics only capture conversions that happen in the same session, or if later visits originate from different devices.

If conversions are low and the category has no other strategic value, then it’s a good candidate for eliminating or at least renaming it to try to boost engagement.

3.  High Bounce Rates on Category Landing Pages

The whole point of a category page is to drive traffic to the content listed on the page. If users arrive at a category page and leave immediately, without clicking any of the options, something is wrong.

How to interpret high bounce rates: Consider where users were before reaching the category page, as well as what they see once they arrive.

  • Does the label accurately describe the category, and could it be misunderstood by users? High abandonment is often driven by unmet expectations. Link labels used on ads or in search results should accurately describe the page they lead to — a gallery of garden photos should not be described as “How to Design a Garden.”
  • Does the page layout prevent people from seeing the content? It’s worth visiting the page yourself or even running a quick usability test to make sure that the page content is visible and discoverable. If you use the same template for all category pages, it’s unlikely that just one category have a bounce-rate issue due to a layout problem. But sometimes content editors make one-off modifications that unintentionally impair the page usability. (Our course on Web Page UX Design discusses many such unpleasant issues.)

4.  Low Entrance Rates

The very first page a user sees in a particular visit is their entrance or landing page. Entrances are strategically valuable because they represent an opportunity to expand your audience (especially if you filter for entrances by new visitors, who have not visited the site in the past). If a category has few entrances compared to other categories, it does not effectively attract users and may need to be adjusted.

How to interpret low entrances: For the low-entrance categories, check other metrics:

  • Are entrances lower than you would expect based on other factors? If the topic is one you consider to be significant — for example, a category about “Award-Winning Gardens,” which showcases some of your best content — then i lack of effective promotion or poor labeling might be the problem.
  • Is the low entrance rate common across all channels? Segment arriving users by source, and analyze each channel separately. For example, if the category has decent social traffic but poor search traffic, it’s probably a topic that your users care about but you have an SEO problem, such as using different terms than the ones people search for.
  • Does the category have a high rate of conversions among the visitors it does attract? If so, it may be worth keeping since it’s effective for that niche audience.

Always check the entrance rate before you decide to eliminate a category, to make sure you don’t eliminate an important entry point into your content.

5.  High Volume of Search Queries

Search queries indicate what people want, and also suggest that they could not find it in your current IA. Generate a list of the most commonly searched terms, either directly from your search engine, or from within your analytics tool (if it is integrated with your search function.) The terms that users search for most frequently might be worth adding or prioritizing within your content categories. For example, if “xeriscape” is a commonly searched term, consider adding it as a separate category.

How to interpret high volume of search queries:

  • Is the search query already represented by a category? People often search for terms which exist in the current architecture. If a frequently-searched term corresponds to an existing category, consider promoting the category more prominently to make it easier to find.

Consider whether there is enough content to justify creating a category, and how well the content lines up with the business goals. Not every popular search term needs its own category. Only elevate those with a significant strategic advantage for the organization and its users.

Conclusion

Interpret analytics data within the broader context of content strategy, business and user needs, and SEO value. Look at all these dimensions before deciding whether a category needs to be removed, renamed, or kept as is.

In ambiguous cases, supplement your analytics metrics with other types of data. For example, conduct an A/B test to study the effect of renaming a category, or set up a short user survey on a particular category page to gather information about visitor motivations. By focusing your efforts on the warning signs above, you’ll be able to efficiently analyze and improve your IA based on analytics data.