Lies, Damned Lies

LIES, DAMNED LIES,
AND INTERNET STATISTICS

James R. Rosenfield

September, 2001

"There are lies, damned lies, and statistics."

Lord Acton or Benjamin Disraeli or Mark Twain first said this. Attributions vary. But whoever it was, he meant that you can do anything with statistics, innocently (but damagingly) if you don't understand numbers, and manipulatively if you have an agenda.

I've spent the last few weeks immersed in Internet and e-Commerce statistics, of which there are volumes. It's clear that Internet "statistics" are, as often as not, an amalgam of guesswork, wishful thinking, pie-in-the-sky optimism, end-of-the-world pessimism, and trivial commonplaces.

I will group the "information" I've gleaned into four categories:

1) Useful or at least interesting
2) Useless
3) Bad
4) Scary
5) Innumerate (innumeracy is to numbers as illiteracy is to words.)

I have confined myself to U.S. data not only to preserve my sanity, but also because of the law of large numbers. In a huge country, statistics (provided they're real) are more stable than those from other places.

1) Useful information

--Who trusts ads? You'd think no one would, but in a research environment, people don't want to portray themselves as absolute, 100% skeptics. So according to Forrester Research, 35% of adults trust ads in newspapers; 30% in magazines; 25% on radio; 24% on TV (you gotta be kidding!); 18% in direct mail; 14% on the Internet.

There is a precise correlation here between the longevity of the medium and its trustworthiness. Newspapers are the oldest medium, the Internet the newest. Exception: Direct mail, which has been around for at least 130 years. But direct mail's ease of entry - all you need is a stamp, a piece of paper, and someone's address -- makes it available to any thief or maniac who can read and write, a fact that has eternally compromised the medium's credibility.

--What's the demographic profile of the average Internet user in the U.S., in the year 2000? A bit younger, a bit more male, a bit whiter, a fair amount wealthier ($57,000 household income vs. $42,000), significantly more educated (41% college graduate vs. 22%), and more managerial/professional. No surprises there whatsoever, so this is "information" that supports what you already know, not necessarily a bad thing as long as you don't pay lots of money for it.

--Click through rates, year 2001? According to eMarketer, Inc., who aggregate research from many sources and promise disinterestedness, permission-based e-mail is at 3.2%; non permission-based at between 1.0% and 1.5%; and banner ads at .3%.

Click through rates have been declining, but that's in the nature of things: The novelty has worn off of the Internet, and most users now approach it with goals in mind. A person on an information gathering mission is likely to ignore banner ads, and to be greatly annoyed by pop-ups.

Permission e-mail, from my perspective, is doing pretty well, considering that most of it is created without much direct marketing skill. And, by the way, eMarketer says that 82% of Internet advertising in the year 2000 was direct response, in case there's any doubt about the nature of the Internet medium. Fifth four percent of total ad expenditures in 2000 were direct response, if you're to believe the Direct Marketing Association, which I don't. So reducing the DMA's self-interested exaggeration down to 50%, you have an even more dramatic difference between the physical world and the virtual world.

--How much is too much, when it comes to permission e-mails? According to FloNetwork, Inc./NFO Interactive, 12% of Americans would accept daily e-mails (lonely folks, one would think); 18% several times a week; 31% once a week; 10% every other week; 18% once a month; 6% less than once a month; and 5% don't know.

Generally, most companies and customers seem relatively comfortable with once a week, although this may well change as the total volume of permission e-mail increases. USAA, the vast financial services organization, tries to limit its permission email to once a month, but USAA is unusually sensitive about being intrusive.

Several studies indicate that the more experienced folks get with the Internet, the more negatively inclined they become towards commercial e-mail. That figures, since length of time on line would correlate with opportunities both to opt-in and to generate multiple cookies for the spam merchants.

E-mail message volume? 677.1 billion projected for 2001, a hair over 1 trillion for 2003. These are large numbers. Cut them in half and they're still immense.

--So, if you succeed in driving someone into your website via permission e-mails, what do customers want - and not want - when it comes to buying stuff from you? According to Indiana University/KPMG, in the year 2000 U.S. online shoppers wanted toll-free phone access to customer service (79% -- this is as a fail safe for when those servers go down); 73% want e-mail order confirmation; 67% e-mail shipping confirmation; 58% e-mail access to customer service; 56% online shipment tracking.

And once they're close to buying, what stops them? Well, according to the Boston Consulting Group, 52% of U.S. shoppers abandoned online transactions in 2000 because of invasive information requests (PricewaterhouseCoopers reports that 95% of Americans are in favor of opt-in; the other 5% probably don't understand what it is); 48% because the site takes too long to load. (Hint: If you're a consumer marketer, dumb down your technology. Lots of your customers and prospects are on dial-up.) Forty six percent are reluctant to enter credit card data; 45% complain of poor site design; and 42% mention website malfunction (those darned servers!).

2) Useless "information"

Robert Coen of McCann Erickson is the great guru of media expenditures in the U.S. His projections have been the ad industry's basic guidelines for several decades now. "Internet number continue to be a mystery to me," he says.

And how! So many of them are all over the place! Take abandoned shopping carts: eMarketer, which, again, is a disinterested aggregator of data, shows 2000 estimates ranging from 8% from the Software & Information Industry Association to 80% from AT Kearney. One would think they couldn't possibly be measuring the same thing, except they're separated by 10 other estimates ranging from 25% to 75%. This is completely useless. As are comparative estimates of "Projected US eAdvertising Expenditures Growth, 2001," which range from Merrill Lynch's -25% to ActivMedia 110%. Well, gee, Merrill's been bitten badly by the dot-com collapse, and ActivMedia might have some vested interests.

Comparative estimates of US eAdvertising expenditures for 2001 range from $5.4 billion (Robert Coen, at McCann Erickson) to those wild bulls at ActivMedia, at $23.5 billion. It's interesting to see the same crazed range for 1999, a year long past enough you'd think statistics would be signed, sealed, and delivered: from $1.9 Billion (CMR) to $16.7 billion (AdZone Interactive).

Click-through rates for permission e-mail were mentioned previously. I find eMarketer's 3.2% reasonably consistent with what I see at companies I work with, but estimates for 2001 start at 16.4% (IMT Strategies, who must be smoking something weird).

How many users can receive HTML e-mail? FloNetwork, Inc. has it at 70%, Netcentives 45 - 50%, a pretty big spread if you're trying to decide on your format.

And what about e-mail marketing spending? In this case, eMarketer is the bull, popping for over $2 billion in 2001, vs. the modest $1.3 billion range of Jupiter Research and Forrester Research.

3) Bad "information"

Useless "information" telegraphs its lack of utility through its absurdity. When the range is so wide, it's clear that no one knows what's happening.

Bad "information," on the other hand, can seem believable, which makes it dangerous.

Iconocast, for example, says that Amazon.com averages one promotional e-mail a month. I get several a week. (That's fine with me, because they're almost always well-targeted and appropriate.)

Knowledge Systems & Research, Inc. proffers some June 2000 information on "U.S. Online Consumers Who Say Advertising Influences Their Purchase Decisions." Strongly agree: 3%. Agree: 24%. Neither agree nor disagree: 45%. Disagree: 20%. Strongly disagree: 8%.

First of all, consumers simply don't know when they're being influenced or not. In fact, they're likely to look at a timely promotional e-mail as information, not advertising, even if they buy. Second, only a minority of consumers will ever admit to being influenced by advertising, since it behooves most people to claim that they're above such persuasions (which is exactly what the ad industry wants them to believe, by the way).

4) Scary information

Veronis, Suhler & Associates report that in 2000, Americans spent 261.5 minutes a day watching television. That's over 4 hours and 30 minutes, and it's actually a low estimate - you see higher ones. That means that Americans spend more time in front of the TV than doing anything else, except for working and sleeping.

No wonder we're the dumbest advanced civilization imaginable.

5) Innumeracy

Bad information combined with innumeracy breeds "statistics" that fall apart as soon as you probe them. Problem is, no one probes.

I commend to you a useful little book, published in 2001, called "Lies, Damned Lies, and Statistics," by Joel Best, Professor and Chair of Sociology and Criminal Justice at the University of Delaware. Best talks about "mutant statistics," statistics that sound reasonable until you think about them.

His favorite is one he calls "The Worst Social Statistic Ever." It's something you come across every now and then: "Every year since 1950, the number of American children gunned down has doubled." A student of his found this in an academic journal in 1995.

"…(L)et's assume," writes Best, "that the 'number of American children gunned down' in 1950 was one. If the number doubled each year, there must have been two children gunned down in 1951, four in 1952, eight in 1953, and so on. By 1960, the number would have been 1,024. By 1965 it would have been 32,768…In 1970, the number would have passed one million; in 1980, one billion…only three years later, in 1983…8.6 billion (about twice the Earth's population at that time). Another milestone would have been passed in 1987, when the number of gunned-down American children (137 billion) would have surpassed the best estimates for the total human population throughout history (110 billion). By 1995, when the article was published, the annual number of victims would have been over 35 trillion…"

A less dramatic, but hugely annoying, mutant statistic that appears often tells us thats that Americans are subjected to about 3,000 advertising messages per day. Jupiter Research, in fact, says that in 2001, Americans can expect a daily inundation of 3,838.

Well, we get a lot of marketing messages, but do the math: assuming a 17 hour waking day, 3,838 messages translate into 225.8 an hour, or 3.8 a minute. Even giving "marketing messages" the broadest possible construction (the Windows logo on your computer screen, the baseball cap promoting "Joe's Construction") this is clearly impossible.

Let's see what happens when you try to correlate this mutant statistic with TV viewership. If Americans spend 261.5 minutes in front of the TV, and if 20% of TV content is advertising, and if each ad averages 30 seconds, that would amount to 52.3 minutes of TV advertising per night, 104.6 commercials. Let's say that the average couch potato is idly leafing through a magazine, drinking some beers (the label equals a marketing message), and playing with the Internet at the same time: We'll double the dosage during TV time to 209.2 marketing messages, which will also more than accommodate 10 and 15 second TV spots.

If you subtract 261.5 minutes from 1020 minutes (the 17 hour waking day), you're left with 758.5 minutes. If you subtract the 209.2 TV-time marketing messages from the supposed 3,838 messages per day, you get 3,628.8 messages, which have to be forced into 758.5 minutes. That's 4.8 messages per minute that need to be accounted for outside of TV time, which is when people receive most of their advertising messages.

This means that people actually have to get more advertising messages when they're not
watching television: when they're working, chatting, seeing dentists, eating, gazing out the window, walking the dog, and going about their business. In fact, they have to get an advertising message every 20 seconds.

No way. To say the least, it doesn't add up. And lots of other "statistics" don't either.