Data-Driven Design: Dare to Wield the Sword of Data – Part II

Dec 19, 2012

In my previous article, I talked about how web design can benefit from data, and now I’d like to recommend some ways in which data can be better fused with the creative process.

Although recent studies have debunked the myth of left- and right-brained dominance, the analogy still resonates with many people. Even if you feel as though you’re hardwired one way, in order to be successful you increasingly need to use both sides of your brain. However, for many quants and creatives it can be intimidating to venture outside of their normal comfort zone. I’ve found that having a process or guide can be helpful if you’re entering unfamiliar territory. Before I explore how data can play a more prominent role, let’s evaluate the standard web design process.

As I prepared to write this blog post, I researched a number of web design processes from different interactive agencies. Most of the web design processes could be summarized as having four main steps:

1. Plan (Strategy/Vision)
2. Design
3. Develop
4. Deploy (Launch/Deliver)

Having worked for a leading interactive agency more than ten years ago (Blast Radius), the overall process hasn’t changed much in that time. Conversely, web analytics (or digital analytics as we now like to call it) has evolved significantly from the basic, traffic-focused WebTrends and Urchin reports that I used as a marketer back in 1999-2001. Today advanced digital analytics tools from Adobe, Google, IBM, and others provide deeper, more actionable insights into digital performance.

Despite this change in the availability and business value of online data, data was rarely cited in the web design processes I found. Some of the processes mentioned upfront research consisting of user testing or competitive research in the Plan stage, but there was no talk of focusing on key online metrics or evaluating past performance with web analytics. In addition, the last step of the web design process (Deploy) never mentioned A/B and multivariate testing or even evaluating the results of the new site for possible enhancements.

If web design is one of the most measurable forms of design, why aren’t more web designers using data in their design approach? There are probably lots of excuses—not enough time, no access to web analytics or testing tools, clients aren’t asking for it—but I believe it’s long overdue for more data to be used in the web design process. In fact, Rob Randolph, Marketing Director at Cuker Interactive, felt that digital agencies where design is not increasingly informed by data are doomed to become dinosaurs, and he stated “by combining truly inspired creativity with business intelligence, you stand to win more often than you lose.”

Today, web designers have a wide array of analytics and optimization tools at their disposal. You have enterprise solutions from Adobe (SiteCatalyst, Discover, Insight, Test&Target), the ubiquitous Google Analytics, open-source options (Piwik, Open Web Analytics), and a ton of other unique point solutions (Crazy Egg, Clicktale, Mixpanel, Tealeaf, Optimizely). Let’s see how you can put them to better use in your web design efforts.

Rethinking the Web Design Process

If we deconstruct the standard web design process, we can see how data can be injected into the process to drive more success. I’ve identified six different ways in which data can and should be used within the web design process. The gray areas represent basic steps that need to be a part of every data-driven design. The blue areas are equally important steps that will scale in terms of the level of effort with the size, scope, and cost of the project. For example, for a simple display ad project might not need extensive insight gathering or A/B testing; whereas a major site redesign should include all of the data-related steps.

A. Insight Gathering

In most cases during the planning stage, you have access to all kinds of useful information collected in your web analytics tools. Rather than blindly forming a theory of what’s happening on the site based on different opinions, you can evaluate actual visitor behaviors and understand what’s really happening. Too often digital agencies tear down a website and re-design it without clearly understanding from the data what was and wasn’t working. Without data you could actually waste time trying to fix things that aren’t even broken (before you started).

Data can also help you better understand who you’re designing for. Rather than building a persona based on different assumptions, you can use the data to inform your personas in terms of what devices or browsers they use, where they are located, how they find your site, what content they consume, etc. Insights from the data can shed valuable light on your design project, and ultimately help refine your creative approach.

B. Clarify Goals & Metrics

When the desired outcome or goals of a design project are not clear, it is difficult to design for success. If you know what the business goals are, you can focus your design efforts accordingly. It’s better to demand clarity in the planning stage as it will help your design to hit the mark. Business objectives by themselves can be open to interpretation so tying them to specific metrics and targets can help to solidify what needs to be accomplished. The additional clarity and commitment that comes from agreeing upon the success metrics upfront will give your design a greater chance of success in the end.

Don’t fall into the trap of assuming what the key metrics will be without confirming what the business goals are. In many cases, if you don’t properly identify the right metrics upfront, it will be impossible to back into the numbers after the fact. What wasn’t tracked or measured is lost forever, which is challenging if you’re trying to understand how your design performed.

C. Deploy Analytics Tags

In order to measure the success of any web project, you need to have the right tracking in place. It sounds obvious but too often this step is overlooked, rushed, or trivialized, which results in missing or poor data. If you want to get meaningful insights out of your web analytics reports, it’s in your best interest to make sure someone has put some thought into the data collection and validated that it has been done correctly.

As an digital analyst, it can be painful to watch an expensive campaign or website launch when you know there’s inadequate or no data flowing in from it. If you’re not as geeky about data as me, it’s like not having a camera available for your kid’s first steps or when you bump into your favorite movie star or athlete. It represents an opportunity missed that can never be reclaimed.

In addition, if you’re just placing a basic tag with no customization, you’re passing up lots of valuable insights. Investing 20% more time in analytics tagging (we’re shooting for “ample” not “excessive” tagging) could easily translate into 80% more value out of your digital data. While some data is often better than no data (unless it’s bad data), I’ve found the data that has been tailored to your specific business needs will be far more relevant and useful than many of the default metrics and reports in your analytics tool.

Ideally, this step of setting up and validating the analytics tagging should be baked into your web development process so that there’s ample time and attention to do it right the first time. In addition, your web infrastructure should streamline, not complicate, this step. If you’re constantly battling an outdated content management system, your company will need to address this problem. Your web infrastructure can’t be an excuse for inadequate tagging and subpar insights.

D. A/B or Multivariate Testing

While web analytics can provide you with some useful insights for your design, nothing beats causal data where testing reveals what performs and what doesn’t. Designers who embrace optimization are often shocked at what creative works or doesn’t work when it is tested on actual users. Through testing you can better understand what influences the desired outcome, enabling you to hone your design before it is pushed to all of your online visitors. You also have the opportunity to test multiple creative variations rather than settling for just one design idea or approach. You might be disappointed when users don’t respond to a particular design, but you’ll learn more about their preferences as you continue to explore who they are and what they want.

At Adobe, the designers participate in monthly sessions where they review what performed and what didn’t in their testing campaigns. The web design team looks forward to those meetings, and the attendance has grown to more than 30 people. Senior Art Director, Ben Child, commented that it’s always interesting and surprising to see what performs for different audiences. As an example, they found that digital marketers respond to very different creative than what is used for design professionals.

E. Post-Launch Results

Rather than rolling onto the next design project without looking back, it’s in your best interest to understand if your design worked or not. In the fast-paced world of web design, not enough time is reserved for reflection and examination. If you had clear business objectives and captured the right metrics, it should be apparent how your design performed. The post-launch results form a baseline and provide valuable feedback on your design. If you check your ego and have an open mind, you’ll learn a lot from the insights hidden within your analytics data. They not only give you an opportunity to quickly respond to user experience issues, but the insights may inspire new directions in the future.

I’d recommend monitoring the key metrics after the launch and having a post-mortem review after the dust has settled and visitors are accustomed to the new experience (2-4 weeks). If you don’t have an analyst who can help with the analysis, don’t hesitate to learn how to use the web analytics tools so you can obtain the insights you need. In most cases, you don’t need an advanced degree in statistics to use these tools, just curiosity and a desire to learn.

F. Ongoing Optimization

They say design is never done. Even when a good design improves an online process or provides a better user experience, there is always room for iteration and improvement. In fact, a series of small improvements may eventually translate into a significant enhancement for your business over time.

One of the criticisms of testing or optimization is that you can fall into the trap of optimizing to a local maximum, focusing on only small enhancements and never making any daring design decisions. While it might be easy to blame data for standing in the way of bold design ideas, it is misplaced. The internal sacred cows, conventions, accepted practices, and heuristics at your company are actually what impede design innovation. These implied constraints put the data and testing in a box that drives them toward the only place they can go—a local maximum.

Whenever you challenge a perceived internal constraint or rule of thumb (e.g., “our home page always needs a rotating hero banner”, “our navigation menu has to be horizontal”, “only certain products can be featured on landing pages”), you remove the barriers to both your creativity and what can be improved. If the business goal is to increase a specific metric, then everything should be on the table (no untouchables) or you must accept a local maximum. If you believe in continual improvement of your design, then data is a valuable contributor to ensuring your design is both daring and successful.

If it doesn’t sell, it isn’t creative.
David Ogilvy, Advertising Executive

Beautiful design and elegant code are wonderful. Good job, you get a gold star—maybe even a Webby or CLIO award. However, a more effective user experience that drives significantly higher online sales (leads, subscriptions, engagement, etc.) is really what makes a design shine. I agree with Ogilvy that creative that drives positive results is what really matters. You’re only going to know if your design worked with data, and optimization can help your design to reach its full potential. Don’t fear the sword of data, master its insights to your advantage.

© 2024 copyright - analyticshero™

© 2024 copyright - analyticshero™