Tested Advertising Methods

Tested Advertising Methods was first published in 1932 by John Caples, a direct-mail copywriter at BBDO who had spent years running controlled headline tests in mail-order advertising. The book went through five editions across six decades. It's the closest thing copywriting has to a controlled-study dataset — Caples ran the tests, recorded the responses, and published what he found, including the failures.

His most famous ad opened with "They Laughed When I Sat Down at the Piano... But When I Started to Play!" A before-and-after narrative that ran for decades and is still cited as a case study in emotional specificity.

The one idea: a headline change alone can swing response 10-20x

The claim sounds like marketing hyperbole. Caples meant it literally and had the mail-order data to back it. He'd run the same offer to the same list with two different headlines and see a 2x, 5x, or 19x difference in response. The body copy, the offer, the price — unchanged. Just the headline.

The implication is that creative quality is primarily a headline problem. Not a strategy problem, not a media problem, not a budget problem. Most advertising underperforms because the headline doesn't stop the right person and give them a reason to read further.

Caples identified categories of headlines that consistently worked: headlines that appealed to self-interest ("How to..."), headlines that offered news ("Announcing..."), headlines driven by curiosity ("What makes a man likable?"), and headlines that hit a specific fear or desire with enough precision that the right reader felt personally addressed.

He also identified categories that consistently failed: clever headlines that didn't state a benefit, headlines that required context the reader didn't have, and headlines that tried to entertain rather than sell. "Humor is a dangerous device," he wrote. It's hard to test into consistently.

How it shows up in DTC ads today

Split-testing one variable at a time — the methodology Caples used — is the entire basis of modern creative A/B testing on Meta, TikTok, and email. The platforms did not invent this. Caples was doing it in 1928 with physical mailings.

In practice: a DTC brand running three hook variants against the same body and offer is running a Caples headline test. The Thumbstop Ratio is the modern metric for what Caples measured as response rate in the first 72 hours of a mailing.

A concrete example from Caples's data: two ads for a music instruction course. Headline A: "How to Play the Piano." Headline B: "They Laughed When I Sat Down at the Piano... But When I Started to Play!" Response rate on B was dramatically higher. The difference was narrative specificity — B created a scene, named an emotion (embarrassment turning to triumph), and let the reader project themselves into it. A was a functional description.

The pattern holds today. "How to sleep better" is Headline A. "I fixed my 3am wakeups in 11 days — here's what actually worked" is Headline B.

What operators get wrong

Testing concepts and variables simultaneously. Caples's methodology requires isolating the variable. If you change the headline, the visual, and the offer at the same time, you learn which package won — not why. Most DTC creative testing is package testing dressed up as variable testing. You can't improve systematically if you can't diagnose specifically.

Treating test losers as waste. Caples kept careful records of losing headlines because they revealed what buyers were not responding to. That negative data is as valuable as the winner. A losing hook that still scored 30% thumbstop tells you the concept has legs but the execution is off. A losing hook that scored 12% tells you the concept is wrong.

Related concepts

• Scientific Advertising — Hopkins established the testing discipline; Caples systematized it and published the results.
• Ogilvy on Advertising — Ogilvy cited Caples alongside Hopkins as foundational; they shared the headline-first conviction.
• Concept vs. Variation testing — the modern DTC framework for distinguishing what Caples would recognize as concept tests from execution tests.
• Thumbstop Ratio — the platform-native metric that serves the same diagnostic function as Caples's 72-hour response rate.

Caples's core finding has held for nearly a century: the ad that earns attention earns everything after it, and which words earn attention can be measured. The methodology predates digital by sixty years — which is exactly why it's credible.