Brownie Charts: Delicious Ways To Analyze Social Media Metrics

Digital marketing analysts seem obsessed with baked goods. When called upon to show something in percentages, they whip out a batch of pie charts — or doughnuts, such as the one shown below by AnyChart. Can you blame them? They have a high-stakes, high-pressure job. They deserve any comfort available.

These poor souls are expected to make complex behavioral relationships simple to grasp. Anything less is failure. Their bosses have little patience for long number tables or dozens of graphics.

As for me, I’ve shied away from circular charts (I’m not alone) — at least in web analytics. My recent preference is for grabbing more visual real estate. While I’m at it, I slip in a second dimension of information by using various colors. Charting pros call it a tree map, but in solidarity with my carb-obsessed colleagues, I call it The Brownie Chart.

Yes, a brownie chart. I may be missing something, but the chart below looks nothing like a tree. Sheesh! Actually, this so-called “tree map” (also by AnyChart), is a mash-up with another charting approach — heat mapping. In this case, the size of each segment is the share of land on the planet, and the color gradation, or “heat,” represents the share of world population. Look it over for a moment and you’ll come up with some interesting insights. For example, you could fit the far more populated India, China, plus much of Indonesia and Iran, into Russia. I didn’t realize Russia was so sprawling, and so sparsely populated.

Tree maps “re-shape” what would otherwise be slices of a pie or a doughnut into rectangles. These pieces fit  together, like in a puzzle, to comprise the whole. It truly is a “map,” allowing users to survey every scrap of real estate at a glance.

A Pan of Multi-colored Brownies

Visualization is all about making complex data comprehensible. Pie charts often do the opposite.

Imagine what a baffling pie chart you’d have if you forced it to contain all the data you see above. Although, to be fair, brownie charts do get some interactive help. Notice with this one what appears when you “mouse over” each oddly-cut brownie. Click on the graphic to play with the chart. What you see above is a screen capture of my own session with it.

Look in the low right corner. It’s what appeared as my mouse moved over the Papua New Guinea brownie. The box shows the numbers behind this brownie’s size and color.

An important aside: The person who made this world population chart got carried away. The info box should only show metrics that are supported by the visual. I would reserve other stats, such as infant mortality, for another graphic. That graphic would probably be a different type, like a bar chart — one better suited to show the correlation between population and infant mortality (assuming there is one). The mistake this population chart makes is all too common, and demonstrates how perilous our journey is when we set out to communicate clearly.

The danger of obscuring what’s important has frankly been the toughest nut to crack for me. Reports must get straight to the point. The facts should be laid out in a way that makes editorial and UI decisions obvious. (My visualization hero, Edward Tufte, has written convincingly about how unfocused reporting contributed to the decisions made behind the last, doomed flights of the Challenger and Columbia space shuttles. Sobering stuff.)

I’ll show you some mercy and just show one example here. It illustrates the utility of the hard-working brownie chart and also shows how to report on content interest index (CII). If you’re hungry for more, you can find another example of how I use web analytics brownie charting here, on my Digital Solid blog).

The example graphic below will demonstrate that the brownie chart, and the CII metric it reports on, successfully deliver the goods. And what goods would those be? Editorial feedback.

If you’ve been following my series on CII, you know that I’ve explained why editorial feedback is vital, yet is in extremely short supply.

Here’s an excerpt from that post. I led into the topic by talking about a documentary I’d seen on stand-up comedy:

This documentary reminded me of two immutable facts:

1. The origin of something that meets general public approval is not really an act of Creation – in the Biblical  sense. It’s more Darwinian — a gradual evolution.
2. This requires feedback, often harsh, and plenty of it!

In a comedy act, feedback is instantaneous. We web-based merchants of mirth (well, we’re merchants of something) aren’t as lucky. We rely on analytics.

That’s not good news, because until now simple web analytics hasn’t provided real feedback on user interest, at least on a page level. The CII metric coaches those responsible for specific content. It helps them see interest levels rise or fall with the editorial or UI changes they’ve made to pages.

Later in my post I compared the interest levels that people have in web content to the laughter produced by a stand-up comic.

When [the comedian Jerry] Seinfeld is listening for audience response, he automatically takes into account audience size. If there are a dozen people in the audience, getting 10 people to laugh is a big win. If it’s a theater of 2000, those same 10 voices laughing in a sea of silence would trigger major flop sweat.

Similarly, when generating a CII for a page, you need to factor in a second number. In addition to the number of [Facebook] “Likes” (and [Twitter] shares, email pass-alongs, etc.), you need to look at the number of total page views from that same time period.

Hmmm … If we need to show a bunch of pages, reporting on both the size of readership and changes CII … Why not chart them all on a single screen and apply a heat map? Wouldn’t that be helpful?

Yes it would.

Charting Interest Level Gains or Losses on EB.com

Everything Brownies, Inc. (“EB.com”) is a fictitious online business. The domain name I threw out there is real, but resolves to Encyclopedia Britannica, not to a brownie equipment and cookbook seller. (You can read why I did this on my companion blog post, at Digital Solid). The site might not be real, but the metrics I’ve charted for EB.com are similar to what I’ve measured for other web sites.

Notice that “Millie,” the imaginary owner of Everything Brownies and all-around web innovator, is doing very well for herself. Her site got more than 16 million page views — and that’s just the pages which her hypothetical web analyst chose to monitor with the CII metric. Why monitor these pages only?

Content interest index holds the most promise for improving conversions on a web site when the pages tracked are near conversion funnels.

Here’s an example:

I’ve moused over, and thus highlighted, the page Holiday Brownie Baking Kit. It did better than average in terms of changes to its CII: A score of 120 compared to 100 — giving us the 20 percent increase. You can click on the image to see the whole mouse-over message. This page has a product description — something that many people would want to print out, share via email, or post as a “Like” on Facebook. Perhaps there are even enough baking fans on Twitter to give it a Tweet or two. Whatever. All of these actions are factored into the CII scores for this and the pages we’re tracking.

Holiday Brownie Baking Kit is an important page because it’s sells a profitable item for Millie, and the call-to-action on the page leads right to the Order Now conversion funnel.

However, the score variation of 20 percent isn’t large enough to get very excited. This is the web, where metrics are never exact! So I’m guessing she hasn’t changed the content on this page since last quarter. Based on the CII change, it’s not worth further consideration this quarter.

Deluxe Baking Pan, on the other hand, shows far more CII “warmth.” It has an increase this quarter of 49 percent! That suggests she changed something in the content. This is a big deal for Millie. The page gets the most page views of them all — 4.8 million page views this quarter alone. (Perhaps it’s visited often by those pastry-fixated web analysts!)

Millie needs to compare the copy and graphics she recently changed (plus pricing, specifications, testimonials, etc.) with the content for that same page as it was presented last quarter. The reason: She’s done something right this past quarter! Perhaps she can even learn a thing or two that could improve the CII of other, similar pages.

And what do you think about the Toppings for Brownies page, a.k.a. Toppings? This product page had a 121 percent leap in CII. Was it seasonality that caused this sudden interest? Or some other externality, as analysts like to call outside influences? Or was it a content change? It sure would be a good idea for Millie to find out.

Likewise, Millie should find out what’s up with New York Style Brownies Recipe Book (it’s the greenest shape — it’s almost a square). That page doesn’t have nearly the page views of some of the other winners, but its CII score literally doubled! Its situation can be summarized this way: Opportunity = Low; Interest Level = Sky High!

While Millie is checking on content changes, should she wonder why that maroon-colored brownie shape, standing tall and narrow up against to the big winner, is suddenly such a dud? You bet she should. The page, Millie’s Brownie Recipe Book, lost big this quarter in terms of interest levels. She needs to find out what happened and fix it.

What’s cool about the CII metric is it causes eyebrows to be raised where, before it came along, eyebrows were so motionless you’d think they were botoxed. There was no excitement in the content manager’s heart … and no fear.

There was no actionable feedback.

“What If My Site Doesn’t ‘Do’ E-commerce?”

Think of it this way: We’re all selling things, whether it’s a concepts, services or a products. We also may be doing it directly, or just as often, through a supply chain. The fact remains that we’re selling. And if you’re selling, you should consider CII — with these provisos:

I’ve mentioned in prior posts that you do need large numbers of page views to get statistically valid results. You also need a content management system that lets you to see archives of past page versions. Otherwise, you’d be flying blind when it comes to figuring out what you changed in the content to change CII rates.

There are a few other places where CII wouldn’t be helpful. For instance, blog sites like this one have other ways to measure interest. The very fact that social network sharing is so common here negates the need for measuring CII.

But if your site has pages near just about any kind of conversion — whether it’s an email subscription, a request-for-quote page or a Donate Now like — you should consider using the CII, and charting it with this cool visualization technique.

Then let me know what you think!