Test-Measure-Refine: 80 Years Later It Still Works
-- Claude Hopkins, "Scientific Advertising," 1923. Republished 1998, NTC Business Books, page 294.
In uncertain times, it is a good idea to go back to the basics. The best introduction to e-mail marketing was written about 80 years ago by Claude Hopkins. In a slim, 95-page book called "Scientific Advertising," he covers all the ideas for successful e-mail campaigns today.
Of course, since he wrote it 80 years ago, you will not find discussions of what list server to use, but someone else in your organization should be worrying about that anyway. The book has been reprinted often and, unlike books today about new media advertising, which become obsolete in the three months it takes them to be printed, is still timely and relevant.
One of the fundamental lessons in "Scientific Advertising" is the importance of testing. Testing and measuring the results is one of the things that separates effective e-mail campaigns from simple e-mail blasts.
It is common today to hear campaign managers dismiss the importance of testing because of the low cost-per-thousand costs. The problem with this is that without testing, response rates do not improve. Campaigns became one-time events and not part of a process that improves response rates over time. Low CPM costs simply mean that tests can involve larger samples.
The basic test-measure-refine, or TMR, process is simple:
· Test two or more alternatives in a single mailing.
· Measure the results.
· Refine the results to improve future mailings.
It is important to measure several alternatives in a single mailing. The reason is simple: Each mailing is different, and the lesson learned is about which alternative is most effective. It is very difficult to extract this information if the alternatives are tested across mailings. This is one of the most basic mistakes made.
Many factors can be tested. It is usually wisest to focus on the most important:
· What is the offer?
· What is the positioning?
· Who is the target?
· What is the source of the names to be mailed?
All of these elements are critical to e-mail campaigns. The experience of the marketer and the campaign manager should guide the order of the factors to be tested, and the variations tested.
One of our clients used the TMR method to plan a series of e-mail campaigns. The result was a doubling of the response rate and, just as important, a better understanding of the prospect base. With this understandingof the prospect base. With this understanding, future campaigns improved.
Our client is a major retailer of women's health products. It is usually a good idea first to test the offer. This is generally the most important factor. Three offers were tested. The least expensive was a referral letter. In the middle was a sweepstakes offer. We also tested a high-cost offer consisting of a free product sample. The goal of the test was to learn which offer was most cost-effective.
The response rates were 3.2 percent for the referral letter, 3.5 percent for the sweepstakes offer and 7.8 percent for the product sample offer.
Though the product offer generated the highest response rate, it was the least cost-effective. There was not a substantial difference between the letter and sweepstakes response rates.
While the referral letter was more cost-effective, the marketer's experience was that sweepstakes do substantially better than referral letters and was reluctant to abandon the sweepstakes offer. The marketing and campaign managers decided to drop product sampling in further campaigns but continue to test the referral letter and sweepstakes offers.
Positioning. Next we tested the positioning of the e-mail piece. Three types of copy were prepared: copy focusing on the health benefits of the product; copy focusing on the nutrition benefits; and general copy emphasizing both. Each was tested using both the referral letter and sweepstakes offer.
The results of the campaign test are below. Notice that TMR helped refine the effectiveness of the offer and the positioning. The response rates increased to more than 5 percent from the 3.2 percent to 3.5 percent of the initial campaigns.
Health: Referral -- 2.8%; Sweepstakes -- 3.3%
Nutrition: Referral -- 3.6%; Sweepstakes -- 5.2%
Neutral: Referral -- 4.9%; Sweepstakes -- 4.4%
The health positioning was dropped as relatively ineffective, retaining the nutrition and neutral positionings. The nutrition/sweepstakes combination garnered the highest response rate, but only marginally higher than the neutral/referral combination. Since the cost of the sweepstakes offer was higher, future campaigns used the neutral/referral combination.
We considered targeting next and looked for a segment of the target population of women based upon age, geography, income, education and number of children in the household. For simplicity, we tested two cells: women younger than 35 and women older than 35.
The tests showed a response among women younger than 35 of 3.2 percent and a response among women older than 35 of 7.2 percent. This was an easy choice to implement. Future campaigns would target women older than 35.
Source of names. Now that the TMR approach narrowed the offer type, the positioning and the target, we tested the source of the names next. This was done to increase the size of the target population. The campaign was sent to the original name provider and two alternative suppliers.
Using the referral letter offer, the neutral positioning and females older than 35, the three list sources generated the following response rates:
Original list -- 7.3%; List 2 -- 5.9%; List 3 -- 5.2%
The original list had the highest response, and while the responses of the other lists were lower, they were all significantly higher than the original unrefined 3.2 percent response rate before the TMR process.
Today, there is a growing belief that targeting does not matter, given the low CPM price. Our experience is different. Lists burn out quickly. Good targeting decreases burnout and maintains the value and usefulness of a list. Good targeting can significantly increase response rate or maintain the same response rate, while decreasing the number mailed, saving other names for other days. The test-measure-refine loop is as valuable today as it was 80 years ago when Hopkins described it.