Note: Feature(s) Pending RolloutFeatures and functionality described on this page are part of our May 2025 release, which is rolling out to eligible customers May 15-June 19, 2025. See our Release Notes here to learn more. |
Objective
Provide direct instructions on how AI should tailor a given message by specifying tone, length, relevant data sources, and context.
Applies To
- Outreach Admins
Procedure
- Log in to Outreach as an Admin.
- Click Content > AI Content.
- Select Seller Content.
- Click Create.
- Complete the following fields:
- Personalization name
- Gen AI instruction
- Outline structure
- Click Save.
Additional Information
Gen AI Best Practices
Personalization name - be specific, this helps others across your team understand when and where they should use this personalization block
AI Personalization is managed by admins, content committees, or content creation teams.
Gen AI instruction (AI Prompt) - this can be as long or short as you would like. Additional best practices will be provided to help with prompting. Make sure to check the length in the read time area at the bottom of the prompt box.
Outline structure is important. This helps you with the tone of the messaging, if you would like the prompt to write things in specific ways such as keep it to one sentence, keep it casual, or don’t use the prospects first name. Adding these parameters gives the AI model more parameters to work within.
Buyer data - you can pick up as many as you would like. These work as a data waterfall method so if it can’t find the first signal you pick, it will default to the next, and the next, etc. You can mix signals, account and prospect information. We recommend you start with rarer signals like “IPO” on the top and more common ones towards the bottom like “New product Launch”, we also recommend using Account Description field towards the end, so that you always have some fall back. But if none of these signals or fields have value, the content will be generated using the AI prompt without any additional buyer data.
Seller content - adding seller content (industry, persona, product) can help the model with better messaging accuracy. If the Agent play you are running is about a specific product, use that product messaging here so it can use that to write emails on your reps’ behalf.
Variables - Use variables to tailor the prompt with data-rich context but stick to fields with high completion rates to ensure consistent results (e.g., {{first_name}} or {{account.name}}).
Preview - it’s always wise to generate a preview for any new personalization you make. This way you can make any adjustments to any of the above information to ensure the output you want is accurate.