engaging experience

What's the Story?

As director of fundraising web strategy for a large institution, I create monthly web analytics reports for my internal clients. The process requires me to wade through extensive data on site usage, e-commerce transactions, marketing campaign results, and any other activity that produces measurable output.

Although my clients have full access to this information at all times, they can't digest raw data. They need a condensed version that lets them see how their properties are working and what improvements could be made. It's my job, as their internal consultant, to explore the data for peaks and troughs, trends and correlations, and to turn the data into meaningful, actionable recommendations.

I like to think of this process as "finding the story" in the data, discovering what really happened last month, and why. Stories are easy to tell, easy to remember, and easy to make use of in future planning. I give each client a summary memo with that month's key story. This is far more useful to them than a grid of numbers could ever be.

Here's an example. Last month a client sent an email driving visitors to an online holiday slideshow. The show was a non-commercial, nostalgic, feel-good piece offering warm holiday greetings and, by extension, inviting a sense of connection to the institution. The site enjoyed a huge traffic spike to the slideshow page for a day or two, and this drove up overall traffic numbers for the month. So the slideshow seemed like a terrific success.

But looking further at the numbers, I noticed the site also experienced a much higher than average bounce rate that month. That meant that most visitors arrived at the site, saw only a single page, and left. Looking back at the slideshow—the largest contributor of visits by far—I noticed it provided no call to action at the end. In fact, nothing on its page was clickable—there was no navigation, no footer, not even an email address. There was nothing for a visitor to do next. The slideshow ended in a dark alley.

So what's the client's "story of the month?" That while the email/slideshow combination was wildly successful in driving visitors to the site, the client missed an opportunity to leverage its success by asking the audience to explore related content.

And the recommendation? Keep the seasonal slideshow, because it clearly works to drive traffic. But next time, embed it on a page with site-wide navigation elements. Add a call to action at the end, and on the footer of the page, suggesting links to related content—seasonal events, other multimedia shows, or news articles that seem topical. This will keep some portion of the audience engaged, and continue to foster the sense of connection. Which is, after all, what the slideshow campaign is all about.

Google Flu Trends

From the Wisdom of Crowds department: Google Flu Trends shows that search queries for flu-related terms are predictive of flu activity in the United States. Although not everyone who searches for flu information is sick, patterns emerging from regional queries are indicative of infection rates and distribution.

These predictors can estimate how much flu is circulating up to two weeks faster than the models provided by the Centers for Disease Control.

Google's results have been published in Nature.

Magnetic Fields Visualization

This short film, recorded at the Space Sciences Laboratory UC Berkeley, offers a striking visualization of magnetic fields.

Hat tip: cboone

Site Bones Laid Bare

Mark Callahan of Mazamedia has produced a creative series, Internet Soul Portraits, in which he strips Google, MSN, and a handful of other popular sites of their content, revealing only their structure. Callahan calls it art, though they arguably read more like reverse-engineered site wireframes. But it's interesting to notice how easy it is to recognize some of these sites simply by their bone structure; the bones are part of the brand.

Junk Food For Thought

Kaiser Fung offers Junk Charts, a blog deconstructing confusing, obfuscating, or just plain ugly information graphics. It's amazing how many good looking charts are just plain bad information design. See, e.g., this post on a stock market performance graphic from the New York Times.

Airport As Organism

Interesting thread happening now at edwardtufte.com about whether improving airport taxiway maps might reduce runway incursions.

Maps are notoriously difficult to design well. Tufte writes:

Good maps layer and separate their information, partly in terms of a hierarchy of relevance but, more importantly, in a pluralism of differences. In good maps, active viewers can read several separate layers of information. These runway maps are extremely flat. These maps were probably put together piecemeal over the years, as each newly added piece of data competes with what was already there. For example, some type is added but it doesn't get enough emphasis compared with what is already on the map - and so the type goes into a bureaucratic box with a clunky arrow. Everything is saying "look at me now." This masks the relevant data sought by the user. It is like a bad cocktail party; someone is talking louder because someone else is talking loud . . . and the result is cacophony.

I'm less interested in the discussion of optimizing map design, and more interested in optimizing the design of the system of map-controller-pilot. The tower controller sees the whole landscape moving and shifting in real time - a three-dimensional, active matrix of planes, tarmac, and air. The controller doesn't need a map, because she works there every day, and she has memorized the map, memorized the airport, and so has a mental model of the system. But more important, I think, is that for her, the territory is the map.

Meanwhile, the pilot sees only what's on the other side of the windscreen, only what's in front of him. The pilot might be familiar with the airport, and so, like the controller, also might be working with a mental model. But the pilot's task on the ground is to move forward only at the command of the controller, to move to a point and stop, awaiting the next command. His operational information space is not three-dimensional - it is at most two-dimensional, and arguably even one-dimensional.

A map is, to use Andy Clark's word, scaffolding - an external prop that assists with cognitive processing and alleviates cognitive load. But, in this case - does it? The map is not the territory, and humans, when presented with both, have a doubled mental task. A two-dimensional map, even one that's well-designed and offers multiple layers of information, forces the pilot to switch between information spaces, adding both load and risk to the system. So, could we collapse the pilot's tasks to a single dimensional paradigm?

The key here is that no single element in this system has enough information to make a good decision. So the goal should be to help the system, not just the pilot, do a good job. We need to treat the system as the cognitive organism embodied in the landscape, and optimize based on the needs of the organism for communication, vision, and action.