Rules of Thumb
How many contacts?
Motiva Message Testing AI thrives on data. While it doesn’t require millions of data points to be effective, it does need contact-level activity over some period of time. As a general rule of thumb, we like to see ~2,500 contacts flowing through a Motiva email step at a minimum, which would only support simpler optimization scenarios (e.g. small number of variations, short timeframe, etc.) Anything less runs the risk of weaker confidence results. For results to be shown, Message Testing requires a minimum of 150 contacts per email variation. For instance, if you have 3 variations, Motiva will require 450 contacts (3 x 150 = 450) before results are reported.
Generally speaking, the bigger, the better. And don't worry about scale; we work with some of the largest marketing campaigns in the world.
How much time?
Motiva learns fast by incrementally learning from audience response patterns but it still needs a bit of time. At a bare minimum we suggest two or three days, but if you can spare a longer campaign window, do it!
Try lowering the confidence threshold
The confidence threshold you configure directly affects how aggressively Motiva prunes lower-performing messages, which can increase overall email performance. By default, Motiva runs at 95% confidence, which means once Motiva reaches that threshold it will go all-in on a single variant. Try lowering this threshold. This can be really useful when you don't have much time for a campaign to run.
More than two variations
If you've done any traditional A/B testing, you probably only tested out two versions of a message at a time. With Motiva, try three or more. We regularly have customers running 25-way experiments, which just gives Motiva more to learn from.
Why? If you give Motiva two versions, it will run it similarly to a traditional A/B seeking to get to a single winner according to the Confidence Threshold that you set. This forces Motiva to send the same volume of each of the variations out until it finds that winner. The downside here is that Motiva isn't incrementally adjusting its send strategy along the way, since that would pollute the statistical validity of the results.
If you are running two variations, you'll likely want to lower the Confidence Threshold to 85% or 90%. This will force Motiva AI to try to find a winner sooner.
There's nothing wrong with only two variations or with A/B tests. It's just not taking fuller advantage of what Motiva can do for you, and your team isn't getting as much feedback or learning along the way.
The more creative and diverse your variations are, the more Motiva and your team will learn. Remember: 1) you can test out multiple variables at a time and 2) Motiva is always going to find the best mix of messaging approaches given your audience. Motiva will automatically boost higher-performing variations and shut down losing messages.
So don’t worry about guessing “right” when you’re considering creating variations. You’ll see better results by being bold and testing out lots of different ideas.
Have a plan
Motiva makes it easy to experiment with almost anything. But we think it’s important to go into it with a sense of what matters most to your business and how to find the quickest path to delivering and proving value. Consider the basic Motiva optimization loop like this:
- Define a good question / hypothesis. Something like, “Longer-form emails are more effective for audience X in this campaign.” Or “Secondary subject lines make no difference to open rates”.
- Set up the variations to test the hypothesis.
- Configure and activate Motiva, letting it run for a sufficient amount of time in order to generate an answer to the question.
- Use the results to improve future emails and generate new questions to test. Share those learnings with other team members.
The best Motiva AI use cases are ones where you have a clear sense of why you’re testing and the business results you’re hoping to generate.
Once you’re familiar with some of the basics of how to use Motiva AI, it’s tempting to apply it everywhere. But you still only have so much time in the day. Assess where the greatest business value could be gained by improving email marketing performance, and how that might be measured and demonstrated. That will probably lead you to a particular candidate use cases and help focus your efforts.
Things to try
When you log in to Motiva AI with your Eloqua ID, you’ll see a report on your homepage called “Weak Link”. Motiva automatically surfaces campaigns where performance is abnormally low compared to your other campaigns. These can be great places to employ Motiva’s optimization features!
Simple subject line testing
Subject lines (and secondary subject lines) can be a quick way to raise response rates for an email. But instead of testing just a couple of option, we encourage customers to really think out of the box.
Try 3, 5, or 10 subject lines! And don’t just limit yourself to minor variations. Experiment with different lengths, tone, language. The more widely diverse the options, the more you (and Motiva) will learn and the faster you’ll see results in your campaigns.
Multivariate subject line testing
Try generating three main subject lines and two secondary subject lines, which gives you a total of six variations to test. Simply select those assets in the Motiva configuration screen and watch Motiva optimize towards the best combination.
Send Time Optimizing
Let Motiva help you discover the optimal send times to use for particular audiences. Just configure Motiva to send over a certain time window, and it’s will automatically generate a dynamic report of response patterns. Once you have enough collected data, use those insights to restrict send times to the best performing windows. And try applying those send time restrictions to other steps in your campaign, if applicable.
Finding new audiences
One of the most powerful features of Motiva is its ability to surface audiences from behavioral response patterns. There’s nothing you have to do here, other than navigate to the “Who Responded” tab in Motiva and select meaningful contact attributes. This report gives you a quick sense of how audiences cluster around different message variations, and can help you diagnose and refine your audience segments.
You can even take your findings around audiences into other areas of your marketing strategy, from SEO to advertising to other, offline media channels.
Boosting Nurture or Onboarding
Inserting Motiva into an early email step in a multistep campaign can be a great way to generate both immediate response improvements and downstream effects. You might start simple with a click-through type optimization where you vary the body of the email, or possibly combine a few subject lines + body variations.