When you’re reviewing Variants in Persado Portal, you may find yourself wondering: “What data was used to create these messages?” This article will give you more insight into how our language models take into account relevant data from the brand, account, industry, impression type, and channel to create content that makes sense for your particular campaign.
Exploration Variants (Full Experiments)
The Persado Brand Content Strategy (BCS) team expertly prompts our model based on your campaign goals and control message. BCS takes into consideration the type of language being used, the intended audience, the touchpoints, and any additional context you’ve provided via the Experiment Request Form.
Our platform then designs an orthogonally balanced test designed to learn what works best for your specific audience. It will include language (Emotions, Narratives, Descriptions, etc.) and formatting (positioning, bold/italic/underline, colors, etc.) that may be known to be high performing for:
Your industry
Your particular message type
Your audience.
Our BCS teams also have the ability to curate specific tests you’d like to incorporate into the design. One of the most powerful aspects of experimental design is its ability to cleanly test multiple factors within a single test!
Experimental design is meant to learn both what works well versus what doesn’t for your specific audience so that we can apply learnings to future campaigns. The goal is to find elasticity (i.e., the difference in performance between the best and worst performing message) in the first phase of testing, and then achieve lift in the Broadcast phase. This means that we sometimes include content that we anticipate won’t outperform the control message in the Exploration phase so we can validate what language we should avoid for your brand.
To learn more about the science of experimental design, check out this article. |
Predictive Content and Essential Motivation Variants
Our model first understands your control message: what it's about, what type of language is being used, who it's going to, and any additional context provided.
If you aren’t using a control (for instance, if you are generating content from scratch with Essential Motivation), our model draws from the same things it would pull from your control - but in this case, it uses the prompts you've provided around campaign info, product info, and any additional considerations.
Our model then generates content taking into account your unique brand profile, data from Experiments you’ve conducted with Persado so far, and relevant industry data. It will include language that may be known to be high performing for:
Your industry
Your particular message type
Your audience.
Need a refresher? Learn more about Predictive Content in this article, or Essential Motivation (self-service) here. |
Dynamic Motivation Variants
Inputs for Dynamic Motivation occur in a one-time setup phase to create a Dynamic Content Library. This means that you’ll provide the Persado team the email and/or web placements that will feature Dynamic Motivation content and identify the the topics/actions to be featured (e.g., Loyalty, Special Offers, Cart Abandonment) up front as opposed to having to provide a control for each campaign.
Similar to Predictive Content, our platform understands the most effective way to say each message as categorized by Narrative, Emotional, and Descriptive language. To do this, our model draws from brand profile data, relevant industry/channel/use case data, and data from any Experiments you’ve conducted with Persado so far.
Once served, these Variants are personalized in real time based on your first-party data. This means that you’ll ‘set and forget’ the campaign and our AI will automate the decisioning and language generation in market.
If you’re not familiar with Dynamic Motivation, check out this article for an understanding of the basics. |