Groundhog Day (Or, Towards Reflective Equilibrium in Web Metrics)


“I am not a target market.”
– Douglas Coupland, Generation X: Tales for an Accelerated Culture

Canada celebrates Canada Day this Tuesday, and on Friday, Americans celebrate Independence Day. As holidays go, what’s not to like? You can go outside and fly a kite, watch fireworks, eat a hot dog, and maybe even ingest a little patriotic Kool-Aid. Certainly a far better day than, say, the Groundhog Day as perpetrated by the unhappy gods who pelted Bill Murray with hellish irony in the movie of the same name.

I’m thankful that my own life hasn’t been like the Murray Groundhog Day character’s. No life-insurance-hawking goons come up to me day after day with the same (“Phil? Phil Connors? Remember me? Ned! Needlenose Ned!”) greeting. Mornings don’t always bring the same recording of “I Got You Babe.” Arguably life has not only been different, it’s been better. Last July 1, I left a family barbecue midway through because I had promised to send out my just-launched subscriber newsletter on or before July 1. I went back to that proverbial “tiny one-bedroom apartment” which at the time doubled as my office, and worked for four hours in 40+ degree heat and high humidity (without a/c) cobbling together the subscriber list, formatting the HTML, and editing copy. (In $US that’s about a hundred and eighty degrees.) I melted, but the job got done. Life has definitely improved since then. Soon after that episode, I saw my shadow, smiled, asked the in-laws for money, and got the hell out of that tiny proverbial one-bedroom. (To you people from Vancouver rolling your eyes in the back of the room, please pipe down.)

But it wasn’t so for hapless weatherman Phil Connors. As Connors was repeatedly subjected to the same “day,” he became increasingly bored with its minutiae. Then, a brainstorm. Having so much knowledge about what was going to happen at what times could surely be turned to his advantage. The one great unattainable feat – winning the love of the endearing producer character (Rita) played by Andie Macdowell – could be attained if Phil played his cards right (and often enough). Or so he thought.

At first, the plan seemed to be working. Phil found that he could impress the object of his affections just a little bit if he took advantage of the fact that he knew what was going to happen next. He could be gallant or even clairvoyant on demand. He felt his power growing. The next phase would be to learn everything there was to know about her likes and dislikes, then cater to them.

It was working. The lady was charmed. Being Phil, his luck would eventually run out. But he found himself getting farther and farther along in the “day” before she inevitably said “Phil, I’m just not interested.” He was starting to enjoy this.

Then, something unexpected. He got his face slapped. He’d started to script the “day” so precisely that a little bell would suddenly go off in Rita’s head: this is too perfect… it’s too scripted… it’s CREEPY. She felt manipulated. People aren’t lab rats. You can’t hurry love. SLAP.

Fine, no problem, Phil said to himself. I’ll study harder. Script things more precisely. Do even better at catering to what she wants. Push her buttons in exactly the right way. She’ll be forced to succumb to my weird science.

SLAP. This time, it came a few minutes earlier in the day than the time before.


What was going on? These slaps were happening sooner every time. The harder Phil tried, the sooner his day ended. Confound this woman! Get me out of this day!

Is there a lesson in this for online marketers? If you’ve been paying attention to the game lately, you’ve noticed more and more attention being paid to measuring website results and engineering websites to produce desired actions from users. That’s no surprise, of course. You don’t want to bring potential customers to your e-commerce site just to have them browse around only to leave and never return. And if there has long been a science to retail (designing the layout of store shelves to increase revenues, etc.), shouldn’t this carry over to the extremely “engineerable,” testable, quantitative world of online sales?

Well, of course it should, but I’m going to argue that like Phil Connors, you can only go so far towards engineering the user experience before users sense your over-exuberance. Surely there must be a way to influence behavior in a generally favorable direction for your company without striving an unrealistic degree of control over the user’s every whim and synapse. I mean, is it so tough to understand that a lot of people flocked to the Internet, and certain sites in particular, because they were trying to get away from messages and processes that seemed too “corporate”?

I’ll say it again: we owe a lot to the “new web scientists” who are pushing metrics harder than ever. After years of wallowing in the miasma of unaccountable, complacent search engine marketing, for example, it’s refreshing to be reminded by Bryan Eisenberg that “conversion rates measure success, yet way too many Web marketers equate success with traffic.”

Equally important to the success of many web ventures is getting out of the idea that mere navigability is the key to designing a more usable website that will convert more visitors. That’s part of the point of what writers like Seth Godin (in The Big Red Fez: How to Make Any Website Better) are trying to get across, but navigation alone doesn’t create sales. Persuasion does. You can persuade with words; you can persuade with cleaner navigation. But you must persuade. Getting rid of some of your navigation options is probably going to be more important to this effort than making the existing options easier to find.

And don’t forget that persuasion begins at that point of first contact! Does your banner ad, email ad, pay per-click search engine ad, or organic listing do a good job of pre-qualifying your visitor, and preparing them to take a desired action on your site, such as filling in a free application form, buying your entry-level product, or requesting a white paper?

Like Eisenberg, Kevin Lee weighed in on search marketing last week, arguing that many marketers give up on it too soon because they aren’t executing, measuring, and testing properly. Lee even gives the “average” marketer a fighting chance to do better at what were previously seen as esoteric calculations of “profit optimization”: Lee suggests that knowing something like the average cost per lead or cost per order for different media is a rough way of ensuring that the right amount is being spent in these various media. For my own clients, it might apply something like this: a review of the numbers shows that our FindWhat campaign is running at $70 per lead while the Google and Overture campaigns came in at $30 per lead. This essentially means that too much of the budget is being spent on clicks from FindWhat. To make the proper adjustment, one needn’t shut off that hypothetical FindWhat campaign; one would simply bid lower (and accept a lower volume of traffic) until that cost per lead got closer to $30. Or, if volume were more important, and if the campaign would still be profitable at $50 per lead, the Overture and Google campaigns could be bid higher across the board, creating more clicks and leads at a significantly higher cost per lead. I can’t imagine running an online marketing campaign without attempting to follow at least these rough kinds of metrics.

Some of Lee’s other suggestions, though, may be unnecessary or simply beyond the capabilities of smaller companies. “Dayparting,” an obvious consideration for some retailers, relies on solid data about certain days and times of day when consumers are buying (and when they aren’t). As many know, though, sales can come in bunches. Is it worth risking not having one’s ads showing if there is a sudden, unexpected interest in one’s product? Is it worth an advertiser’s time (or money if they are outsourcing the work) to follow all of these metrics so closely? Ultimately, as with the financial markets, a hyper-empirical viewpoint (one that takes certain patterns in the data as “real” as opposed to being tendencies which could be overwhelmed by yet larger, unpredictable market forces) can be risky. Poring through a “stock trader’s almanac” that shows the stock markets performing in a certain way, on average, in the second week of July might seem like a good way to make money, until the trend reverses itself or proves spurious. Personally, the only thing I do about the “lunchtime effect” is to have lunch.

According to authors like Eisenberg, to truly succeed in redesigning a site to improve its conversion rates, you must test and measure. But how many factors can or should you be measuring? Doesn’t it depend on the size of your company? Even a good-sized company with 20+ employees may have trouble finding dedicated resources to turn the web development side of the company into a full-fledged “persuasive architecture laboratory.” We’re not all

But maybe more importantly, I have a sneaking suspicion that if you get it too perfect, that customer is going to slap your face.

If the love analogy doesn’t grab you, how about golf? If you don’t grip the club correctly and align your feet properly, as the Golden Bear always said, it doesn’t matter how much talent you have; you’ll have a great deal of difficulty making an accurate golf shot consistently. But that doesn’t mean aligning your feet correctly will turn you into Tiger Woods. Everyone needs to know the fundamentals. Most games are 80% fundamentals. But the rest is instinct… or magic. Sometimes it’s better to be lucky than good.

So: first things first. Move the big levers, and make those important changes that can increase ROI in big gobs (200%, 400%, etc.). Try something big like launching a separate site which is just a long, persuasive one-page sales letter written in consultation with a trained professional, replacing an old site that provided the user with a myriad of options. See a major improvement in conversion rates? Great. Rejoice. Buy yourself a new lob wedge. Kiss your spouse. Enjoy your Kool-Aid.

You may also like