Land in the inbox on BFCM: test your audiences

1. Intro
2. 3 months out
3. 2 months out
4. 2 weeks out
5. Day of send
6. Post-send

Gradually test unengaged audiences

Imagine the impact of dropping a huge, heavy object in a swimming pool, like a piano, or a car. It would cause a huge ripple, and would probably do some serious damage to the integrity of the pool. Now picture a slightly smaller object, like a bowling ball. It’s smaller compared to the alternative, but would still make waves. Now imagine small droplets of water being added to the swimming pool, little by little. The impact is minimal and wouldn’t cause too much disruption to the rest of the pool.

You can think of audience testing in the same way: you want to test out your unengaged audience to see if you can get any of them back, but you don't want to risk making huge waves in your sender reputation by blasting a bunch of unengaged profiles. A few months before BFCM, start testing those unengaged segments by mixing them into campaign sends that also include your highly engaged customers (like little rain drops). This gradual approach limits the negative impact that those unengaged profiles may have on your deliverability since you are including them in small doses.

‍

‍

Plan your test campaigns

To test their audiences, Beantown plan to send one campaign daily. They will alternate every other day between sending to their engaged audience only, and then to their 90 day engaged segment plus a test group (increasing in size as per the guidance above).

Use the following sending strategy when scheduling your campaigns:

1. Split the test campaign 2. Return to engaged 3. Repeat the process

1. Split the test campaign

Build your campaign in Klaviyo. Once you're happy with the content, clone the campaign so there are 2 instances of the same campaign.

• ‍First send: send to your 30 day engaged segment
• Second send (begin sending 1 hour after first send, send over a 3 hour time span): send to your 31 to 90 day engaged segment + first test group of unengaged subscribers

You can configure this setting when you go to schedule your campaign by choosing Gradual send over several hours.

2. Return to engaged

The day after you send your first test campaign, send a campaign just to your 30 day engaged segment.

• Do not test audiences multiple days in a row.
• Always allow for 1 day of “sending as usual” in between tests.

Why do this? The purpose of audience testing is to gradually reintroduce unengaged profiles to your campaign sends without causing serious damage to your deliverability. Pulling back to engaged audience in between tests helps protect your sender reputation.

3. Repeat the process

As you continue testing, you should see your core engaged segment start to grow as you successfully re-engage some lapsed subscribers.

Be sure to follow our guidance on not increasing your audience size by more than 50% of the previous day’s audience to avoid significant damage to your sender reputation.

Note: The segment graphics in the tabs above are purely for illustrative purposes. To find out how to build an engaged segment, read this article.

Break your unengaged segments into test groups

Depending on the size and age of your account, you may have a few massive unengaged segments. When testing, you never want to increase your sending volume more than 50% of your previous audience size. This means if you have an unengaged segment with a lot of profiles in it, you will need to use Klaviyo’s audience sampling tool to break up the segment into smaller test groups that you can gradually add into campaigns.

The audience sampling tool allows you to break large unengaged segments into smaller lists that you can use for testing. This will give you more control over recipient count on campaigns, allowing you to minimize damage to your sender reputation by only including small batches of unengaged subscribers with each send.

‍

Example scenario

Beantown’s 90 day engaged segment is what they consider to be their primary engaged segment, and it consists of 10,000 profiles. They want to test their Unengaged L120 segment to see how many of them they can re-engage, but this segment contains 40,000 profiles, which is well over 1.5x their usual sending volume. Their first test send should not exceed 15,000 recipients, as this equates to their engaged segment of 10,000 x 1.5.

They use the audience sampling tool to pull a sample list of 5,000 profiles from their Unengaged L120 segment. They send their test campaign to the 90 day engaged segment of 10,000, plus the test group of 5,000 profiles from Unengaged L120.

On their next test, they will again increase their send volume by 1.5x since the last send, sending to their 90 day engaged segment plus a test group of 7,500 profiles from their Unengaged L120 segment.

Follow the steps below to achieve the outcome in the above example.