Have you ever wanted to upload existing context for a campaign to prompt Persado’s AI? You can now provide files, free-form text, or file URLs that our language model will automatically ingest when improving or generating content. Persado language models will extract the relevant details and incorporate them directly into your generated Variants.
Here are the tactical steps and best practices to keep in mind when using this new Content Guidance feature.
How to Use Content Guidance
1. Input your Existing Content and/or Keyphrases as Usual
Follow your usual process for improving or generating content. Read more here.
2. Choose a Content Guidance Format
Select the type of input you want to provide in the new ‘Context’ module within the ‘Additional Refinement’ section of the content prompt. You can choose between a file upload, free-form text, or a file URL. Each option has its own system limits:
Upload Files: Add up to 5 files (maximum 20MB each) in JPG, PNG, or PDF format.
Add Text: Enter up to 10,000 characters of plain text.
File URLs: Add up to 5 file URLs. You may upload a file via URL provided it is a PNG, JPG, or PDF hosted online. Please note that the file must be accessible via a public or shared link and cannot be stored on your local drive.
3. Add a Description
For every file or text input you provide, you can optionally add a short description explaining how the content should be used. For example, you can specify that certain offer details must be included, that a product name and its attributes should appear in the output, or that the upload contains supporting context such as an email body when generating a subject line.
These descriptions help the Persado language models understand what is most important and ensure that the extracted details are incorporated accurately into the generated content. The model will also gather more subtle or nuanced inputs that can't realistically be entered into a free-form prompt, or a tagged input field.
After uploading all your inputs and adding descriptions, all of the additional context you included in your prompt will be shown at the bottom of the prompt input modal.
4. Generate Using Context
After adding all of your context, click ‘Generate’ as usual. All uploaded files, text inputs, file URLs, and descriptions are submitted along with your prompt.
In the example below, we’re improving a full email that starts with a generic “new credit card offer.” Because specific offer details were added through the Context module, the generated Variants now include those details (for example, “earn as high as 100,000 Membership Rewards points”).
Keep in mind: The context that you upload won’t surpass or replace your brand’s tone of voice. Instead, it will accompany it with information about offer positioning, campaign goals, etc.
This means you can upload internal documents that support the creative without any concerns about the style of writing impacting the copy itself. |
5. Edit Inputs to Regenerate
After generating content, you can edit the context included in your prompt. Click the ‘Add’ button to open the input module shown below, or hover over any existing upload and click the ‘Remove’ icon to delete it.
Use Case Examples and Best Practices
Need inspiration? Here are some ways you can leverage the Content Guidance feature.
✅ To reflect other impressions in the same funnel:
Upload an image or PDF of the email body for a subject line playground, or a mobile in-app modal for a push notification playground.
Type the URL of the landing page of the email.
Use a file URL with the landing page of the email.
Add the touchpoints of the email body or in-app modal as text.
✅ To include additional product information:
Upload a document with product information for vague control messages.
Use a file URL with the product page details.
Add free text with information about the product.
✅ When campaign briefs are available:
Upload a brief with info about the campaign.
Use a file URL with the brief.
Add free text from the brief of the campaign. This could be guidelines, campaign information, or content.
And here are some instances where we’d recommend you use other features instead.
When NOT to Use Content Guidance | What to Use Instead |
❌ To upload segment information | ✅ Use the 'Segments' functionality |
❌ To add important information about the campaign, to be used in the outputs | ✅ Use the ‘Other Content Considerations’ functionality |
❌ To suggest mandatory words or concepts | ✅ Use the ‘Always Use’ functionality |
❌ To generate content from scratch using only the uploaded document | ⚒️ We’re working on ‘Generate from Brief’ and Long-form functionality that will cover for these use cases. Stay tuned! |
FAQs
Does ‘Content Guidance’ work for all channels?
Yes! You’ll now see this additional field when generating OR improving content for any channel that is enabled for your account.
The most intuitive use for this feature is for channels that are part of a bigger funnel. For example: You’re optimizing an email subject line and want to use the text of the email body as additional context. The subject line may be vague, and because of being a space-constrained channel, it does not contain all the necessary info to proceed with content generation—but the additional material uploaded provides the right context and information to the model.
When should I use the 'Content Guidance' field?
The 'Content Guidance' feature is best leveraged when the initial message or keyphrases do not contain enough details and/or are very vague on what the campaign and product is.
If I use the ‘Content Guidance’ field, do I still need to enter keyphrases or other details?
Yes; even when using the ‘Content Guidance’ field, we recommend you enter keyphrases and campaign/product details as usual. Think of ‘Content Guidance’ as providing additional information about the context those details apply to.
For example: Let’s say you’re generating a headline and CTA for an email body. You could use the ‘Content Guidance’ feature to provide context that the rest of the email is about a credit card offer so your generated CTAs might read "Apply now" instead of more generic options like "Continue."
What is the difference between ‘Content Guidance’ versus ‘Always Use’ and ‘Other Content Considerations’?
The ‘Always Use’ and ‘Other Content Considerations’ fields are designed for keywords and phrases that you want to see (more or less precisely) in the final content outputs. The new ‘Content Guidance’ feature is intended as a way to provide more context to the model so that it can generate more accurate and relevant outputs. The actual content uploaded or typed in ‘Content Guidance’ does not necessarily appear in the outputs; rather, it’s used to ‘guide’ content generation.
Why am I not seeing anything in the outputs related to my ‘Content Guidance’ inputs?
The ‘Content Guidance’ fields are used to guide content generation, but this doesn’t mean that the input is repeated—whether verbatim or not—in the outputs. In some cases, the influence of the ‘Content Guidance’ input may be just a nuance in the content or a subtle hint. In other cases, the additional context may be used by not including certain words.
For example, say you upload email body context for a subject line, and the email body’s headline contains the word “Take advantage.” The model may infer that it’s better to not output a subject line with the same word. So, the absence of the email body wording is the way this additional context is being used.
In some cases, it’s also possible that the algorithm can’t find a way to properly incorporate information from all inputs due to character length limitations or other brand guideline specifications.