Tested Advertising Methods

I’m not sure where I got the recommendation for Tested Advertising Methods (5th Edition) by John Caples and edited by Fred E. Hahn. It might have been a comment on Hacker News or possibly one of the fine recommendations that Patrick McKenzie is in the habit of dropping.

What’s it about?

John Caples was an advertising man — specifically, a copywriter. Copywriters write the actual advertising text — “copy” — which appears in any given advertisement.

What makes Caple’s book interesting for a lay business audience is that Caples worked in mail order advertising. Before the advent of the internet, mail order advertising was the only field in which an advertiser could get meaningful feedback about the performance of different kinds of ads.

The first part of the book is built on commonly successful properties that Caples noticed in his long career as a copywriting star. These range from general principles (“long copy sells better than short copy”) to very specific (words to start headlines with, eg. “How I blah blah blah” — just look at HN to see how well that one works!).

The second part is Caples discussing the central importance of testing to any advertising campaign. Caples is disinterested in well-accepted principles of advertising. All advertising must be tested, then tested, then tested again.

A/B testing

Of passing interest to me was how Caples mentions “A/B testing” and gives an explanation of its etymology.

A/B testing is best known today amongst web developers and marketers. In a website A/B test, visitors are shown, at random, different versions of a web page. Consequent clicks, purchases, signups etc are tracked. At the end of the test, the different versions of a web page (“A” and “B” … and “C” and “D” and so on) are compared for performance. The winner is adopted as the baseline for the normal webpage.

If you go to Wikipedia, that’s the basic description for A/B testing. But the term “A/B testing” actual refers to a kind of advertising test that was invented for newspapers, which Wikipedia doesn’t mention at all.

Newspapers are printed on giant print rollers; each roller has impressed on it multiple pages in a newspaper. Opposite sides of the print roller have the same page.

In an A/B test, the newspaper is paid to make two different impressions on each side of the roller. One side as advertisement “A”, the other side will print advertisement “B” in the same position. Because the two versions are alternated, this neatly avoids confounding variables from other comparative testing techniques. It’s the same newspaper, the ad is in the same position on the same page, the weather is the same, the geographic distribution of the different versions identical and so on and so forth.


Tested Advertising Methods may be suitable for a few audiences. Web marketers will find the simple common principles a useful starting point for their copywriting efforts; though most of the testing methods will be irrelevant given the possibilities of tracking users on the web. Small business owners may find it useful to help them zero in on simple, effective advertisements.

This entry was posted in Books, Business. Bookmark the permalink.