Predictive Content is Persado’s rapid form of language creation. This article defines Predictive Content, outlines when and how to use it in your campaigns, and answers some common FAQs.
What is Predictive Content?
Predictive Content is a Persado messaging creation method that employs machine learning to produce up to 4 Variants - variations on the control you provide us - designed to deliver performance. Our machine picks concepts and phrases that are predicted to drive impact based on best performing language, emotions, and benefits according to your brand, industry, and channel. By using machine learning, we generate up to 4 messages that are then Broadcast to your audience.
If you have performed Experiments in the past, we’ll draw on insights from your audience; otherwise, we’ll pull industry insights from our machine. To ensure language is not repeated across Persado customers’ content within the same industry, warnings display when overlapping or similar language is found in recent campaigns. Our Brand Content Strategy (BCS) team will then adjust the message to be more unique. As BCS becomes more familiar with your brand, they will also be able to better personalize the content to match your voice.
Using Predictive Content:
Enables you to take advantage of all that Persado has learned about different audiences and your industry - without having to perform Exploration for that specific campaign
Allows for a fast and easy way to use machine-generated content and still achieve a performance uplift
Reduces the time required to review and approve Variants as Predictive Content deploys at most 4 Variants per send, compared to Full Experiments with 16 Variants.
Predictive Content is a fast, easy way to create impactful messages for a high volume of campaigns.
Did You Know?
Machine learning is a subset of AI. It’s based on the idea that machines can be taught to learn and adapt through experience. Machine learning algorithms build a mathematical model based on sample data in order to make predictions or decisions without being explicitly programmed to perform the task. Machine learning technology helps us improve efficiency and accuracy thanks to the way it constantly learns over time.
How does Predictive Content work?
Predictive Content occurs in 4 key steps.
You provide Persado with the control.
Persado uses AI to generate up to 4 Variants using best performing language, emotions, and concepts according to your brand, industry, channel, and campaign.
You review and approve the Variants, and we incorporate updates as needed.
We help you deploy the Variants in a single send, maximizing time-to-impact and reducing overall timeline. Typically, we deploy no more than 4 Variants using Predictive Content and we can deploy as few as 1, not including the control.
Do I need to provide a control and metrics for Predictive Content?
Yes, because we still need to see what the control looks like. While metrics aren’t required, they are highly encouraged so we can scope the project, upon request.
Do I need to approve the Variants?
Yes. We always want your signoff on the Variants we create, regardless of creation method. Predictive Content results ultimately inform the AI of what language resonates with your brand, so we want to ensure the approved Variants match what is deployed.
Do we deploy a control alongside the Variants?
It’s recommended to deploy a control alongside Predictive Content, but not required. Including a control allows us to measure impact and learnings by seeing the difference in lift between the control and Variants. Without a control, you only receive the content without tracking or we estimate what the performance would have been.
How long does it take to generate Predictive Content?
Persado generates Variants based on your agreed-upon SLAs.
Does Predictive Content produce the same insights as Full Experiments?
Because Predictive Content is a head-to-head test of up to 4 Variants, it doesn’t produce the same depth of insights as a traditional Full Experiment, so you won’t be able to see element-level contribution to success or heatmaps. However, you will be able to see which message was most effective, track its primary KPIs, and see the way Persado tags the elements of each message. Aggregate insights are also available in Portal, so with enough tests, you’ll be able to see which emotions, descriptions, etc., consistently perform well.
Your Experiment Report for Predictive Content can be toggled by KPI. Your winning Variant displays at the top of the report with its metrics and tags.
Aggregate results are available for you to see total metrics across all Variants deployed and the control, as well as the best lift.
On the bottom of your Experiment Report, you’ll also find all Predictive Content Variants deployed in order of success rate by your selected KPI. Portal also lists the emotions featured in each message.
When should I use Predictive Content?
Predictive Content is an ideal option when:
Audience size and/or response rates do not allow for Exploration (i.e., the sample size isn’t enough to reach statistical significance, a key component of Experiments)
Speed-to-market is a priority
Technology limitations do not allow for testing of multiple creatives/Variations
If insights gained from the experimental design structure aren’t a priority
Whenever it makes sense for company needs.