Testing mailings: how to create your own step-by-step plan in email marketing

The higher the rate for email marketing in your business, the more controversy about how letters should look. Trying to make the newsletter more efficient, we focus on the client's portrait and product knowledge, on marketing trends, on professional advice and general research.

The surest way is to constantly test all ideas in practice and rely on the reaction of your base. So you will create your own guide with the most effective solutions for your business. Disputes are not needed when there are numbers on hand.
')
Split tests and A / B testing

The main way to test letters. The bottom line: we make several versions of one letter and a helmet on test segments. We set the victory condition and the deadline (for example, 3 hours). At the end of the term, we determine the winning letter that goes to the main part of the database.

How to do:

• We formulate an exact hypothesis , which you either confirm or refute. Choose hypotheses that can then be turned into rules, for example, “in a regular mailing to our database, personalization in the subject line of the letter increases efficiency”.
• Based on the hypothesis, we are preparing a letter and its alternative (A / B test). In the split test alternative options can be made 2 or more. Letters should differ by only one element that you are testing.
• Customize the test . Usually mailing services offer ready-made templates for split tests. In SendPulse, it looks like this:

• We form a segment for the test . Typically, test letters are sent to 20% of the base, and the winning letter automatically goes to the remaining 80%. At the same time, each of the test options should receive at least 500 people.
• Choose a victory condition . If we test the topic, the letter will win with more openness. If we test elements in the body of the letter, we look at the number of transitions. But I recommend to take into account all indicators and choose the winner manually. For example, if the topic worked well, but did not fit the content of the letter, there would be fewer conversions from such a distribution, despite good openability.
• We carry out repeated tests on the hypothesis and summarize. Each hypothesis is better to test 3-4 times with an interval of 1-2 weeks. We remove the indicators both immediately after the test (3 hours after sending), and a day later, and better - 3-7 days. Results may vary significantly. We take into account everything - discoveries, transitions, target action, formal replies. The best solutions for all indicators we remember and continue to use in the future.

What is worth testing?

In general, you need to test everything.

Start with strategic moments: what type of content and communication style works best? Do you need branded design, which templates are more effective? What offers, promotions, gifts and STA more catch on the audience of subscribers? This data will be the basis of your strategy.

After that, test marketing subtleties that complement your strategy:

Theme



We test the length, the number of words in the subject, CAPS LOCK, word amplifiers, special characters, personalization by name and geolocation, question / call, digest / unique topics, Latin / Cyrillic, numbers, quotes, a specific sentence against the general wording.

Sender



We evaluate different versions of the sender's name - the name of the company, the name of the manager or manager, the name of the project, service, or fictional character on whose behalf the newsletter is sent.

Preheader



You can write a greeting, letter headers, a key sentence, CTA, or use the semantic continuation of the topic.

Letter writing



We test design colors, selection of pictures and banners, colors, size and position of buttons, text links, frames, branded elements, typeface, font color and size.

Content of the letter



It is interesting to analyze a personalized greeting, letter structure, letter length, reference to “you” or “you”, amplifiers, proposal wording, use of feedback, feedback request, use of purchase history, signature, basement.

We test the frequency

When you start to see the effect of your newsletters - interest in content, sales growth - you will be tempted to send newsletters more often. Most companies start with a weekly mailing, but then increase the number of letters to 3-4 per week, or even send them every day, including weekends.

At first, this solution has a short-term effect (more mailings = more sales), but you risk burning out the base with your napalm. I remember that last year Konstantin Benko from InfoBusiness2.Ru shared a terrible story about how in a year they reduced the number of subscribers from 350,000 to 10,000 precisely for this reason.

Comfortable frequency can be determined practically. The test is best done at least 2 months. In general, the longer you test, the more accurate the results. You can experiment for six months, if resources permit.

How to do:

We will test manually. Select 2 or more random segments in the database. The average performance of mailings to these segments should be the same. To check this, the first 2 weeks we send out to all segments equally according to the old scheme. If everything is ok, go to testing.

One segment is left as a control group. The rest of the helmet is the same series of letters with a different frequency. We remove indicators of each letter on each segment and we bring in the table. At a minimum, we look at openability, transitions, target actions and unsubscribes.

After 2 months, we look at the results by segment. First unsubscribing: 64% of people unsubscribe from mailings precisely because of the high frequency. If you see an alarming trend (the number of formal replies has increased by more than 3 times, and is constantly growing) - it is not worth it, even if you see sales growth. Calculate how this will reduce the base in a year and how much you will lose to those who unsubscribed prematurely.

Next, we look at the dynamics: how openability, transitions and target actions behave. The number of discoveries will fall by 10-20% - this is natural, but we compensated for this with additional letters.

We look at the total number of unique transitions in each segment, the average revenue per subscriber, and the total revenue per segment. If we see a significant result, we transfer the base to the winning scheme and continue to closely monitor the dynamics.

NB! On the dynamics of indicators affects not only the frequency. Perhaps, if your subscribers do not want to receive 2-3 emails per week, you should review the content of the newsletters. Also try setting the frequency based on the subscriber’s life cycle. New subscribers are ready to receive emails more often, for example, several times a week. Those who are in the database for six months or more - less often, 1-4 times a month.

We test time and day of sending

Any marketer will tell you that the best days for marketing emails are Tuesday, Wednesday, or Thursday. What are the best days for useful and entertaining content - Friday and even weekends. That mailings need to be sent at 9-12 am (according to some theories, even at 5-6 am).

For example, the recommendations of colleagues from the Pardot agency looked three years ago:



Have these trends changed and how they relate to your audience? Form a hypothesis based on the portrait of your client and test, even if they contradict the marketing canons.

Do split tests in the same way. At the end of the test, adhere to the chosen day, time and frequency of distribution. It will be more comfortable for subscribers, and it's easier for you to plan your work.

What is the bottom line:

1. Test required. This is the law.
2. Especially carefully test mailings when working with large databases.
3. We test only one item. Each hypothesis is checked 3-4 times.
4. We test only what can be used in the future, and further improve the distribution with the help of proven techniques.
5. We start testing the contents of letters with strategic elements, then we test the details.
6. If you decide to increase the frequency, we test and carefully look at the indicators. We take into account the total number of clicks and sales from each segment. We are attentive to the dynamics of formal replies.
7. We test the days of the week and the time of mailing based on the characteristics of the audience.

Successful tests and effective mailings!

Konstantin Makarov, CEO, SendPulse Successful Mailing Services