Mining for Data
Chances are, a few years ago you decided it was about time for your institution to create and maintain a professional, centralized website. You did away with the hodgepodge of independently designed pages and built the comprehensive, cohesive, well-branded site your students came to expect from your quality institution.
And, chances are, the commitment wasn't easy. Maybe you had to argue to justify the budget. Maybe you had to wrest some control from individual departments. Maybe you even had to alert some faculty to the presence of "the internet." But you've come a long way, the site is running smoothly, it looks good, and students and faculty are happy. You're done, right?
Yes! And no.
Like many college and university websites, yours was most likely designed primarily as a service tool for the campus community: prospective students, current students, faculty, and alumni. They visit, find the information they need, and leave. If the site is built and designed properly, their experience is a smooth one, one that entices prospective students and instills pride in the campus community.
And while the tendency is to leave well enough alone, doing so means you're probably not using your site to its fullest potential.
An institution's website is not only a wealth of information for its audience, it is a wealth of untapped resources for increasing alumni donations, capturing prospective students, and measuring the effectiveness of, say, ads placed in the local paper for summer sessions. Those still banking on increasing enrollment with that same CD-ROM authorized for distribution in 1996 might want to seek out books with titles like 100 Things People Do with Unwanted CDs and 1996-It Really Is About 10 Years Ago.
Web analytics isn't new, and most people are familiar with some numbers already: "100,000 hits this week, Bob" or "We're getting a lot of traffic from Google" could be common officespeak. And while these numbers are informative, they may not seem very useful. But they are.
Analytics isn't just "so you're going to count my hits," explains Casey Paquet, web manager at Eckerd College (Fla.). "There are all kinds of different ways to use this information."
As often is the case in higher ed, institutions will need to learn to do as business does. E-commerce sites have been using analytics to increase sales and to optimize their sites since before the dot-bust. But, it is a relatively untouched resource in the world of higher education. There's a whole other world out there and some IHEs are already taking advantage.
Ready to prospect for this new wealth? It will mean delving into the mysterious world of web analytics--the oft hard-to-navigate realm of numbers, stats, IPs, click-streams, referrals, conversions, and abandonments that a website generates. Using his or her webmaster as a guide, any administrator can learn how to bring these numbers into check.
Visit a newer local grocery store and you'll notice a few things: a bakery, a premium foods area, even an automotive section, background music, and an appealing design. They aren't there by accident. These companies studied how their "users" interacted with their stores and found useful trends in their shoppers: People are more likely to buy products at eye level; they buy more baked goods when they smell fresh-baked bread; out of convenience, they will buy non-grocery items. Music and store design can make shoppers feel less like they're performing a chore.
Stores were designed accordingly, around what their users wanted, in order to maximize sales. Shopper analytics, web analytics--not so different.
Simply, web analytics is a process of analyzing statistics generated by a site's users. Each time the site is visited, the server (or web analytics software) records the visitor's every click--from where and when they entered to where and when they exit, and everything in between.
Chances are, an institution's web server already has a statistics package built in by default. It can measure such things as:
How many people visited the site.
How many pages were viewed.
What sites visitors came from.
Where visitors are geographically.
What browser they are using.
These tools may also report how long each visitor stays and from what page they left the site. More complicated analytics packages can share how many times a specific visitor visited the site, when and where they went each time, and where they left, or "abandoned"--or even how long a person "hovered" over a link before clicking.
Sound like Big Brother? Well, that's another article, but it's the impressive and important array of statistics by which e-commerce sites live and breathe.
With this data, patterns emerge, forming the basis of web analytics: It is the "solution that people use to understand visitor behavior," explains Erik Bratt at WebSideStory, a San Diego-based digital marketing applications provider.
Analytics can turn server numbers into measurable benefits for the organization and the website's visitors. "With it, you can tell what content [visitors] are looking at, how they move through the site, what marketing campaigns are most successful, how far they went, how many people click through to admissions, how to optimize the number of admissions, how to streamline navigation, and even [how] to find out where people are dropping out of particular paths," adds Bratt.
It's all very valuable information--if you know what to do with it. And the trick is in the interpretation.
The biggest problem is that not all stats are what they seem. Even if the home page garners the most page views per month, it doesn't necessarily mean it's the most interesting page, or that it "deserves" this rank.
For example, if the only links to other areas of the site are on the home page, it will receive a lot of traffic. On the surface, higher page views suggest interest in the page--and you might walk away thinking that the home page generates lots of interest from its scintillating content and is therefore, interesting.
But digging deeper would reveal how users are truly interacting with the page. If the average time users spent on the page was low, and the average number of repeat visits was high, a different picture emerges about how users are truly interacting with the page. That is where the fun begins. Analytics usually provides the who, what, where, and when--and it leaves the why for the school's crack team of investigators.
Using the above example, it wouldn't take long to see that users were forced to go to the home page to get to where they needed to go, hence the high page views. Although this isn't necessarily good news, the information can be used to greatly improve your site's usability.
A case in point: When Director of Web Communications Brent Harker at Brigham Young University (Utah) noticed an increase in online traffic, he realized that it was largely due to the installation of WiFi spots around campus. Through web analytics, he determined that students on campus were visiting the BYU site more frequently--simply because they had more access.
From these questions emerge an equal amount of actionable items that might solve the mysteries uncovered. Indeed, according to Michael Stoner of Chicago-based marketing communications firm Mstoner, web analytics can measure many things, including "market reach, application fulfillment, [and] site-centric usability," and it can be an extremely useful tool in the continual development of a site, which experts say is necessary.
"It can even help make technical decisions," adds Harker, who measures which browsers and connection speeds are most popular with his users. "It gives us an idea of where to invest our programming resources."
Following are three ways that analytics can be used--for content, navigation optimization, and marketing. Decisions can be made reactively (uncovering an issue and taking steps to address it) or proactively (for measuring success and pushing content).
The most popular application of web analytics in both higher ed and business is content and navigation optimization. "Seems like a basic concept--that you should be looking at what your users are looking for and help drive them to it," Paquet says.
With such a diverse audience and a limited amount of "real estate" for navigation, it can be difficult for an institution's site visitor to find what's needed quickly and easily. Most ".edu"s follow the multi-demographic model, and the temptation to solve this challenge by assaulting the user with more and more link options and various unrelated content has been taken by many a failed site.
As any good webmaster knows, it's not a "build it and they will come" environment anymore; it's more like a "build it and they will come and tell us how to build it better" environment. A higher ed website is a perpetual publication--or at least should be. It is part news, part permanent content, part solicitation, and part branding/marketing. Things change all the time on campus and off.
Analytics can help put the right information in the right place. Thoughtful editing of copy, links, and organization can help make sites, pages, and forms more user-centric.
Pam Williams, senior program leader of the Division of Applied Technology at University of British Columbia, knows this well. She uses web analytics to track the programs prospective students are interested in, and then looks at where they abandoned the site. From that, she can make content decisions.
"It gives us the opportunity to offer prospective students the information they are looking for," Williams notes. "What can we offer to keep the visitor online?"
Williams is also thinking about web analytics as a classroom content area these days. She's helping to develop a program in the subject--one of the first in North America.
Blaire Garland, director of Institutional Marketing for Roanoke College (Va.), is taking Stoner's advice by using his web analytics package to "shape content" even after his firm redesigned the school's site. After a successful launch, Garland says he's in the "fine-tuning stage," using popular and top-level pages as a starting point for changes.
Additionally, Garland uses e-newsletters and blogs as "proving grounds" for other content by measuring interest, responses, and views on particular topics. He can then direct contributors to focus on subjects that generate the most traffic. To that end, he's developing a content team to help determine future content based on analytics.
At University at Buffalo, meanwhile, an automatic e-mail alerts web strategist Rebecca Bernstein's team of any sudden spike in frequency of a particular search term on their site. This allows for instant reaction to potential problems, as well as the adjustment of content to reflect the visitor requests.
"We were alerted that the name of a dorm was generating a lot of interest. It turns out that there was an activity at the dorm and we were able to place a link directly on the home page to help users find the information they needed," Bernstein cites as an example.
But it's not just keywords that can drive content. "In our internal portal, [administrators] can track stats by the content you use," she notes. Since users must sign in to access the internal portal, administrators can track individual statistics according to the person's status, such as faculty, international faculty, student on campus, or student off campus. "They know how a piece of content is valued in a different pattern by a freshman engineering student than a senior English student. We are able to deliver very fine-grade personalized content. You don't have to sift; we do the sifting for you," Bernstein adds.
Overkill? Not when IBM names you a Best Practices Partner for your work. Big Business learning from education--there's a thought.
While web analytics can help meet visitor content needs, Bratt encourages pushed content, as well. At an Amazon-esque level, users would learn, for example, that "people who took Biology 256 also took English 301." Or that those who took a cooking class also expressed interest in phrenology. Could that spark interest in less popular classes?
Let's say that phrenology department is petitioning for a link on the home page. Maybe the dean is adamant that the wonderful new site he just built with FrontPage, featuring orange text on a pink background, should be a prominent link.
Delving into an analytics package can confirm that, indeed, interest in that particular topic is low. Page views are only one to two hits per month--all from within campus. Searches for "phrenology" with the site search tool have been very low. Armed with this information, a site administrator might politely offer an alternative link solution, such as a bottom-of-page link.
Or, surprise! Say that, as the only university in the world offering a degree in phrenology, the school is getting an influx of traffic from internal search requests and lots of referrals from external searches as well. If users are going to great lengths to find this department, maybe the dean's got the right idea.
Paquet, too, is using traffic stats to suggest navigation edits to Eckerd's website, while BYU's Harker is taking a slightly different approach, using "a combination of web analytics and direct conversations with users." He conducts user surveys and focus groups to receive direct feedback. "We ask users what's most important to them, but we also look at the analytics to see what links they're using more than others."
These stats help reveal what users really want from a site. Tailoring the site's navigation to this end should result in higher user satisfaction. Just don't forget another factor in web analytics: common sense. If the alumni site is languishing, certainly don't pull it off the site.
The proactive aspect of web analytics involves measuring a specific set of data, in conjunction with a specific effort. If you've seen an ad with a "/something" in the URL, chances are it's measuring how many people were motivated to visit the site from that mention.
Banner ads have similar capabilities. In e-mail marketing, analytics can measure how many people opened the e-mail (the open rate), whether or not someone clicked on a link (the click-thru rate) and whether that person has html enabled.
Sound like a slick new marketing strategy? While it may seem kind of edgy in higher ed, it's second nature in business, says Paquet.
As director of e-marketing at St. Edwards University in Austin, Texas, Carmella Magnes has one of the most distinctive titles in higher ed today. With a background in e-commerce, she's familiar with the benefits of analytics. Indeed, it's the tool she uses to measure the success of her efforts.
Before building a campaign, Magnes works closely with her internal "clients" to determine their expectations: "Are they looking for lead generations, such as people filling out forms that require specific information? Or, perhaps they are looking to drive traffic to a specific site or department," she explains. Knowing their needs ahead of time enables her to establish key performance measurements before launch.
"Tracking response rates from a specific campaign involves measuring pre- and post-campaign stats not only on the landing pages, but overall," she adds. In other words, Magnes measures activity on the entire site, not just the promoted areas. This gives her a clearer measure of the campaign's success.
Garland at Roanoke tracks a number of statistics, including e-mail opens, clicks on links for both alumni and admissions, ad leads and clicks on landing pages, traffic from interactive ads, and completions of forms coming from outside advertising. If a particular effort is not performing well, he can take steps to correct the problem.
Bernstein employs similar techniques for the University at Buffalo. Discovering that the CTR (click-thru rate) on a link wasn't as high as expected, Bernstein decided to change the wording. The result? A significant increase in CTR on the link. The link? To the university's giving site. Are you sure you're not losing donations due to similar problems?
"With whose time and with what budget?" It's a question that commonly comes after an idea like using web analytics is voiced.
As Paquet explains, the perception is that this new undertaking will absorb additional time and resources that simply don't exist. He posits that it will be "nearly impossible" to convince people to devote time to analytics.
However, it doesn't have to be an expensive or time-intensive undertaking. Getting started with some simple initiatives can have a great impact with little effort.
Besides, asserts Bernstein, "no matter how small your resources are, they should be deployed in a strategic manner." And it's time well invested.
So you might scrap plans to overhaul the officially mandated "Student and Faculty Frog Toss of Togetherness" website and instead put efforts into optimizing the alumni page. If your current web server has a basic stats package already, start there. You'll quickly start learning which questions to ask to get the job done.
Kirk Snedeker is the web manager for Professional Media Group, publisher of University Business.