What's Data Got to Do with It? Information Architecture and Analytics

By Laura A. Palmer

If you were asked to redesign the information architecture for a website, what's the first question you'd ask? You might want to know more about the product being sold or the purpose of the site. You'd ask about the site owner's perspective on problems with the current version. Several years ago, I'd have asked all those questions myself and probably a few more. But now, my first question would be, “How quickly can you get me access to your analytics?”

Some clients may be taken aback by your request. It's the modern day equivalent of asking to look under the petticoats before a formal introduction's been made. Yet, here's the reality of what analytics can offer information architecture: easily digestible data on user behaviors. What's compiled and displayed by a website's analytics provider should always be the first step towards reconsidering a site's architecture.

Data is everywhere and the ability to gather, store, analyze, and display it gets easier every month. Website data that once gave quaint measures on hits to a site and the number of pages viewed are now providing in-page analysis, compilations of keywords, and information on the number of smart-phone users who are hitting a site. There is a wealth of user behaviors here, in a scale and user-friendly format that many information architects (IAs) have only dreamed about.

What follows is an overview about information architecture and analytics—including some compelling reasons for why you should, if you haven't already, learn more.

Analytics: They're Not Just for Marketing Anymore

Here's a question for you: Who typically uses analytics data? Now, think about your answer—did marketing or IT pop up as your immediate response? Analytics can inform any number of different stakeholders in a website. Yet, it's very common to see analytics as belonging to just one department, like marketing. However, if you think about it, it's only marketing data if you're wearing a marketing hat. With your IA hat perched at a rakish angle, you're going to look at analytics in an entirely different way. You're going to see user behaviors en masse and, trust me, you're going to be delighted.

Analytics Basics

Google Analytics is a name you probably know—Google is ubiquitous in all things Web and it's free. Adobe's also entered the analytics business by powering their product with enterprise-level Omniture technology. No matter the analytics platform, all analytics packages work the same way: each uses a chunk of code placed behind the scenes in the HTML to collect information. Analytics data—as the information displayed on the product's dashboard—lets you see, as Google describes on their analytics page, how people found your site, how they explored it, and how you can enhance their visitor experience

Granted, you'll never be able to read visitors’ minds, but you'll be on hot on the trail of their journey. You'll be able to see how many folks follow an optimal path, complete a task, and have a presumably good experience. You'll also see if visitors are having a lost in the digital woods sort of experience. For example, are your visitors bouncing away from specific content like it's poison ivy? You've now got data that can start a process of inquiry and research. Before you know it, you'll be drilling down with custom segments and reports to unearth even more questions about your users’ actions and behaviors.

Why Do Information Architecture and Analytics Belong Together?

Let's first think about what we're asked to do as IAs. Morville and Rosenfeld are the authors of what's warmly referred to as “The Polar Bear Book.” It's the one with—you guessed it—the polar bear on the cover. While the bear strolls across the title outside, Information Architecture for the World Wide Web, inside Morville and Rosenfeld charge IAs with the responsibility to push the envelope and do our work faster and better.

The data delivered from analytics—that aggregate set of users’ interactions on a site—takes us a long way toward addressing that charge. While our tried and true methods (such as card sorts, paper prototypes, wireframes, and focus groups) still belong in our tool kit, analytics give us big-picture data all at once and in a highly accessible form. Granted, absorbing what it all means is like drinking from a fire hose at first, but we need to be very aware of how this robust set of information can support our processes and validate the decisions we make for site architecture.

The ideas of data-driven support and user validation are important for what IAs, as professionals, are all about. Much of what IAs do isn't well understood. Information architecture as practice and profession never meets with the same knowing glance as medicine or law. It just doesn't seem tangible or measurable to those looking in from the outside.

One way to give the profession additional weight is to intersect information architecture with optimal business decisions through data-driven strategy. If an IA can explain with both small-scale qualitative studies and large-scale data sets why they're doing something, the value in their work becomes crystal clear. Couple your information architecture decisions with terms like ROI (return on investment) and KPI (key performance indicator), and link what you're doing to goals and objectives, and the only girl (or guy) more golden than you will be Betty White.

Aren't Analytics Just a Gussied-up Server Log?

In short, yes. But, if you've been in the information architecture business for a while, you know the trouble with server logs. Quick n’ easy was never the name of the game when it came to getting logs and making sense of the data. And who in the world of mere mortals even understood what server log data was? The immediacy and resonance of today's analytics has been an all-around game changer.

Only a few years ago, requesting a server log and actually receiving it were events separated by days, not minutes. It's not that the data wasn't there—data warehouses were bulging at the digital seams. Yet, cumbersome and ad hoc were the status quo of the acquisition process, especially in Ashlee Vance's estimation. Supplicating the “data priests” to tease answers out of costly, fragile systems became the name of the game. By the time you had data, it was stale.

Overall, server logs was a meaningless phrase to most clients—it conjured images of the culinary arts intersecting with forestry. Small or mid-sized clients never knew what you meant or how to get the data. Instead, the IA was on the phone to the client's hosting service provider. After a call or three, reams of data arrived and that data still needed wrangling via various other tools. Right away, you'd lost the ability to provide what Sally Burford called “immediate and agile responses.”

Why are analytics so important today? If we want to push that envelope and work both faster and better, server logs and raw data aren't the logical starting place. Analytics data taken from tools that track user actions within a site is smarter, faster, and better looking (that is, easier for a human to consume and interpret). From the small client to the boardroom table ringed with C-level executives, telling people you're using a website's analytics data to support your plans will resonate. They'll know what data you're talking about and you'll have the buy-in to our work that's so critical for success.

There are No Answers, but the Questions Get Better

It's true—there are no answers. Data has never been about answers. And, it's dangerous to think that blanket solutions and snap decisions are what data can give us. Data should make us think and formulate questions that need answers. What we have with analytics data is a window into trends and patterns of user behavior. With a starting point and a framework derived from the data, the investigation can begin.

Before you start peering into the analytics crystal ball, you need to make the sorts of queries for which IAs are known. Information architects should always start by asking the site's owners about their expectations, goals, and objectives. Gather some basic information about the site and learn if it's perceived as fulfilling the needs of its users. Ask the folks who answer the phones or email if they get the same questions time and again. Try some of the steps and processes yourself—see how the experience works for you.

When you're trying steps and processes, benchmark times and paths for a general comparison point to what the analytics provide. Analytics data can show you how long a visitor spends on a page or the site. This data also shows where that visitor goes. Do the times and paths from your notes match up? Visitors might be taking sub-optimal routes to their goal; if you see that, you'll want to know why. Are there places where visitors just vanish? Other tools in your toolkit—like usability testing—may be needed to get that definitive answer.

From your preliminary fact gathering, you should already have questions coming to mind. What do you want to know about the website? Every site is different—audience, purpose, and context influence what a site's all about. Thus, there's no standard list of questions we should ask each and every time; you'll need to explore what the analytics offer and see what questions emerge.

That said, let's consider a few basic questions an IA could ask and answer with analytics data. And remember, some of these questions have implications down the road for the content strategist or the site's developer. As a good IA, you'll want to get other people involved; even your preliminary look into analytics can be the conversation starter a project and its personnel need.

How many visitors come to the site?

This is the most basic question. If the visitors are too few, there might be a problem with the SEO (search engine optimization) and findability strategies currently in place. As this is information architecture turf, you'll want to make notes to check out what's going on with the writing, tags, metadata, and other areas.

Of course, the other side of too few visitors is a whole lot of visitors. This isn't a bad thing unless they're leaving the site almost immediately or not doing what your client expects them to do on the site. You'll want to check into the time on site and depth of visit figures in your analytics and gather some more facts.

How long do they stay?

A better question might be: did your site's visitors purchase a product, make a donation, or do what they should have? Time isn't always an important metric if your site's visitors are productive.

However, if you've got lots of traffic staying for a long time and doing nothing, then there might be a problem. For example, can visitors find the page they need? What navigation path are they following? What's not supporting their actions on the site? The questions based on time can crop up pretty quickly here.

What content are they reading and how long are they staying on a page?

If you benchmarked that Page C took about 3.5 minutes to read and visitors are on the page for 6.5 minutes, what thoughts do you have? There might be something about the page—including navigation or labeling—that's perplexing your visitors. They may not know where to go or what to do.

What does the navigation summary say?

If a visitor lands on Page A, what's the next page they go to? Analytics will show you the percentage of visitors who go from one page to another. Do visitors go where you'd expect them to go? Is the jump from Page A to Page D logical? Has the page architecture hindered or helped them? If the visitors aren't following the predicted or optimal path, you should have a few more questions. This is a good time to dig into the in-page analytics and see what's happening. For example, you may see that 85% of the visitors click on the pretty graphic instead of the expected text link. That's a call to look at the page for placement of the graphic, readability of the text, and word choices.

What are they searching for?

If the site has an internal search function, the analytics will show you the search terms that make up a visitor's journey on the site. You'll also see what keywords brought them to the site. With this information, it's easy to check labels and navigation hierarchies to determine if what's currently in place is optimal for visitors.

Let's say your client sells sweater patterns for knitting enthusiasts. The client liked the label “cardigans for canines” and put it on the site. However, the data for visitors’ search terms say “dog sweaters” is what they're looking for. You can't deny a dog its rightful sweater and the case for change, no matter how cute that alliterative label was, is compelling.

No Site? No Problems!

Every business or cause has a website, right? At least, that's what we all think. However, there are always new business ventures, personal causes, or niche hobbies going online. It's not uncommon to be developing the architecture for a brand new site. As I said earlier, information architecture still has its tried and true methods like card sorts, paper prototyping with target users, and wireframe testing. Getting together with typical users and learning about the architecture they'd want to use is a natural.

If you're going to be building a site from scratch, basic investigative practices tell us—per Christina Wodtke and Austin Govella—what others are doing. It's easy to compare several sites in the same market and see what's happening. We can walk through their processes and try their menus, navigation, and so on—it's never been easier to gain a good sense of what's out there.

In addition, the array of online tools for information architecture is growing daily—there are many more ways to do what we've been doing. Business intelligence tools that provide analytics on competitors’ traffic, keywords, and other metrics are fast and easy to use. While these may not be free, the investment you make in what Compete.com describes as “website analytics for sites you don't own” may be worth it. This data can be a good way to start thinking about what a new site in a similar industry needs to do.

What Next?

There's more power in basic, free analytics than you might imagine. Information architects can leverage that power to gain insight into specific aspects of a site's user experience. Ever wondered if the international visitors use the navigation differently from their domestic equivalents? What about mobile visitors? Most any question you might have can be answered by building a simple analytics query. Just about anything you want to know is there for the asking.

Laura A. Palmer is an assistant professor in the information design and communication graduate program at Southern Polytechnic State University. She teaches courses in information architecture, content strategy, and Web development.