Coffee, Sun & Technology

Pointed questions, or mini-surveys, can help you quickly resolve a problem you’ve discovered with analytics. A few weeks ago I was working with a customer on a registration process. The classic problem: very high drop-out rate on the registration page, and no obvious reason for that. Clean, simple page, 5 questions total, and just a “Register now” button.

Rather than working on alternate pages and A/B testing them, the customer decided to just pop-up a tiny window with just one question “Why aren’t you completing the registration process” – with a free text box and a “Submit my feedback button”. Very few responded, but for the ones who did, the pattern was clear: “Can’t read your security code”. This was changed and the increase in registrations was substantial.

I know of another customer who regularly calls random customers asking them direct feedback on a site feature which isn’t working well according to the numbers.

The moral of the story (I guess): Be pragmatic. Numbers crunching isn’t always the fastest path to higher conversions.

Clearly out of context but I had to mention Live 8 -

Every single day, 30,000 children die, needlessly, of extreme poverty.

On July 6th, we finally have the opportunity to stop that shameful statistic.

8 world leaders, gathered in Scotland for the G8 summit, will be presented with a workable plan to double aid, drop the debt and make the trade laws fair. If these 8 men agree, then we will become the generation that made poverty history.

You can visit the official site, where it takes seconds to sign the petition and hopefully get our leaders to do something for the kids.

Bob’s getting his PhD in 4-quadrant charts with nice shading effects - Hilarious post really, and as always, a nice punch line explaining the differences between stats and reports, analysis and optimization.

Jeff Nolan is starting a nice blog (Web Descant) with interesting posts on site design.

From his post “Writing Copy for Your Business Website”, two common sense (yet often forgotten) points:

Get to the Point!
Be brief. Try to write no more than half of the text you would have written for a more traditional publication such as a newspaper or magazine. If it’s any consolation, this should make your job a little easier.

Keep It Simple
The most important concept to understand when it comes to writing for the web is: don’t make your reader think too hard. They won’t waste time trying to figure out how your website works, they’ll just go somewhere else.

Thanks for sharing your thoughts with everyone, Jeff!

Using Yahoo’s “View Bids” tool:

Web analytics (6/27/2005)
Top bid: $7.36
2nd: $7.35
3rd: $5.97

(I haven’t checked this in a while but I *think* the bids more than doubled in the last 12-18 months)

This click fraud issue is a shame. John Battelle summarizes some of the key forces in play when fighting click-fraud, and references an article on Wired about BlowSearch, which has a novel approach to fighting click-fraud.

Google reminds me of Microsoft. Both of them in monopolistic positions, extremely profitable, with lots of smart people - both of them continuing to ignore their customers complains and piling up billions of dollars every quarter.

On Walmart and Target - “I’m a Target lover […] it comes down to design, pure and simple” says Darci Riesenhuber on Tom Peters’ blog. Meanwhile, Walmart’s about 4 times bigger than Target (WMT TGT). User experience or business efficiency, which is better? Food for thought.

Michael Nguyen on landing pages. Favorite quote:

They [the users] want to know:

- “Is this the right place?”
- “Is this how I imagined it would be?”
- “Should I click the back button?”
- “Does this look trustworthy?”
- “How much time is this going to take?”

Your landing page needs to address all these issues immediately.

All common sense – but worth reminding from time to time.

Take My Privacy, Please! says Ted Koppel from the New York Times - two quotes,

A nice one on Tivo (!!),

No one is suggesting that TiVo tracks what each subscriber records and replays. But could they, if they needed to? That’s unclear, although TiVo does have a privacy policy. “Your privacy,” it says in part, “is very important to us. Due to factors beyond our control, however, we cannot fully ensure that your user information will not be disclosed to third parties.”

And the journalist’s conclusion:

[…] We cannot even begin to control the growing army of businesses and industries that monitor what we buy, what we watch on television, where we drive, the debts we pay or fail to pay, our marriages and divorces, our litigations, our health and tax records and all else that may or may not yet exist on some computer tape, if we don’t fully understand everything we’re signing up for when we avail ourselves of one of these services.

Exactly.

Of course they’re not. Neither are Marketers regardless of what Seth Godin may think.

But in a growing era of math challenged masses it is not uncommon to be the perpetrator of lies albeit inadvertently. This affliction is especially true among online marketers using analytics with very limited time to ponder their conclusions, and/or have associated skills to evaluate the veracity of their conclusions. Three common untruths that seem pervasive are:

1. Normalizing for natural growth

Most metrics have a tendency to grow on their own accord, from unique page views to population growth, to people who have seen the daily show by Jon Stewart. This does not mean that Jon Stewart is getting more popular, it just says that those that have seen his show cannot undo the fact they have seen his show.

Marketers frequently ignore normal growth when they report new numbers. Maybe you heard that the number of Indians who keep pet elephants in their garage has nearly doubled since 1950. But the total U.S. population has nearly doubled in the same period, from roughly 150 million to 290 million. Thus we are far from being overrun with garage-dwelling pachyderms.

This trick is most used when marketers talk about an all-time box office record with unmatched collections at the ticket counter. What they fail to account for is that dollars are not worth as much as they used to and that the potential movie going audience is larger than before

2. Exceptions become the rules

All human beings like concrete facts rather than abstract statistics. Marketers and politicians know this, and cater to this inherent desire by using “best case” data points that prove their hypothesis. Think of ads for diet supplements, car mileage and other such examples – disclaimers tell the story.

Such measurements will lead one to provide data that fulfill the prophecy. Singling out specific campaigns that have a tremendous ROI are most always suspicious. Maybe those campaigns should be labeled with “results may vary” disclaimers???

3. Picking your measurement point carelessly (or, should I say carefully)

The genesis of this malaise is that of choosing to start or end a comparison at an abnormal point in an ongoing cycle. For example, the conversion this afternoon for electronic goods was 3.5%. Last week’s conversion was 2%. Hence, one can shake their fist in the air and claim a 75% improvement in conversion rates. Of course the conversion in the electronic category at this time of the year (people planning vacations) vary from 1.5% to 4%. So, this fluctuation is pretty normal. Measuring from the low on a bad day to a high on a great day is misleading.

This trick is most used when politicians talk about job gains and losses. One will talk about losses from a high point, whereas the other will talk about gains from a low point.

Of late I have seen many of these claims made by vendors as part of their marketing arsenal – I would urge all to keep the analytics community honest by demanding details.