Google Analytics – just a pretty face?

by Mike on June 6, 2013

Beer goggles can't help this

Beer goggles can’t help this

Smart design values an attractive presentation. But please allow me to advocate for the occasional dose of ugly.

This is partly a plea for honesty and self-confidence, but also a counterpoint against easy looks. Gertrude Stein‘s hatchet face forces us to pay her heed. Abraham Lincoln might not have been a looker, but he sure could turn a neat phrase or two.

If your core functionality is obscured by a pretty UI, there’s a problem. “Homely” can be a design feature too.

But you’re not some kind of hairy, Linux-loving, open source freak. You’re probably thinking, “what’s the problem with a polished interface?”.

Google Analytics and pancake makeup

Pretty chart in three clicks

Pretty chart in three clicks

Think about Google Analytics. Business-critical, technologically advanced, and absolutely free. It’s a gift to the world.

But it’s dressed up to go dancing instead of wearing something a little more appropriate for the office.

GA catches your eye and pulls you in. In three seconds, you’re making beautiful music together. But as the months and years wear on, you begin to wonder what you really accomplished.

GA has the ability to tap into the most sophisticated business data you can imagine. But the combination of simple interface and colorful, detailed results is deadly.

GA mastery before knowledge

GA encourages people to believe they have mastered the software when they’ve only skimmed the surface. Once you hook up the Javascript tag, you’ve built a handy-dandy log analyzer. Woohoo – I see pageviews!

And that’s where most of GA’s 15 million installed base stops (myself included, for far too long) to enjoy the feeling of accomplishment. They never understand the true purpose of GA is helping customers achieve your goals – like buying product.

Some top-tier shops have gone much further. Goals, API implementation, e-commerce, etc. They’ve used Google Analytics to understand and improve their businesses at a far deeper level. But even among elite users, the flaws might run deep. Unattributed value, pages that can’t be analyzed, and other problems can prevent GA from reaching its full potential.

Too much concealer

Google faces a serious problem: great complexity. The software works backwards from how most people think. It’s so intricate, even the help files are occasionally incorrect. Google can’t reduce the complications – they’re solving something big.

But I’d suggest the larger problem for both novices and experts is the beautiful user interface. It works like concealer makeup to cover any problems – and users never realize they need to step up a level.

To be fair, GA’s ecosystem has provided helpful books (written by PhDs!) and amazing product evangelists. But these steps are bolted on after the fact instead of being integrated into the product.

Most stats software could use a little wrinkle cream

A face only a motherboard could love

A face only a motherboard could love

Typical statistics software like SAS and SPSS sit on the other end of the beauty spectrum. They also handle huge complexity, but never cover up. These software packages are thick with big words and exposed concepts. The software simply won’t allow us to maintain any delusions.

That kind of brutal honesty is nice, but these expensive packages take clunky a bit too far. They feel like they were recently ported from a 1960’s era mainframe. And the open-source R? Even worse.

Just a hint of eye shadow

Just a hint of eye shadow

A little gloss goes far

So where’s the right balance?

A one-man development shop built an inexpensive stats package that hits the mark. Wizard relieves statistics complexity instead of hiding it.

It incorporates t-tests, Kolmogorov-Smirnov tests, and other sophisticated tooling without the UNIVAC feel.

Wizard keeps users close to their statistics through subtle color emphasis, interactive controls, and instantaneous calcs. It even exports Excel prediction models so users don’t goof up the results. This all comes together to help make the problem domain more approachable.

Applying makeup to your work

Of course, make your design attractive. But if you’re working in a complex problem space, ensure your users understand what they’re getting.

Ask them. And if they’re missing out on important details in their business domain, emphasize these parts more. No matter how ugly.

We users appreciate when your app’s killing floor is hidden behind closed doors. If I’m working on stats, I probably don’t need to understand the details of how to access the GPU for blindingly-fast calculations. If I’m analyzing my business, I probably don’t want the gory mess of stepwise regression models.

Just don’t overprettify.

Previous post:

Next post: