Please note that this feature is currently in Beta. If you would like to learn more, reach out to your Growth Strategist.

Mutiny's AI allows you to scale personalization for an entire list of target accounts on key elements in your experiences. This guide will help you maximize the effectiveness of your AI-driven personalization efforts.

Getting Started

Follow the basic steps outlined in our How to personalize for accounts using Mutiny's AI article to set up your experience. This guide focuses on advanced techniques and best practices for crafting effective prompts.

Crafting Effective Prompts

The key to getting great results from Mutiny's AI is writing clear, specific prompts. Here's how to structure your prompts for optimal outcomes:

1. Specify the desired output: Start your prompt with a clear instruction, e.g., "Create a headline" or "Generate a value proposition".
2. Incorporate account variables: Use the + sign or @ symbol to include relevant account data.
3. Provide context: Give the AI information about your goals or constraints.
4. Request a specific tone or style: Describe the desired voice for your content.

Example Prompts

Here are some effective prompt examples for different elements:

1. Headline: 

"Create a compelling headline that addresses [COMPANY]'s specific challenges in the [INDUSTRY] industry, highlighting how our solution can improve their [PAIN_POINT]."

2. Value Proposition:

"Craft a value proposition that resonates with [COMPANY]'s size ([COMPANY_SIZE]) and emphasizes how our product can help them achieve [GOAL] while overcoming [CHALLENGE]."

3. Call-to-Action:

"Create a persuasive call-to-action that speaks directly to [COMPANY]'s role as a [COMPANY_TYPE] in the [INDUSTRY] market, encouraging them to take the next step in addressing their [PAIN_POINT]."

Tips for Refining Prompts

• Start broad and then narrow down based on results
• Experiment with different combinations of account variables
• Use specific instructions for tone and style (e.g., "professional but friendly", "data-driven and authoritative")

Common Pitfalls to Avoid

1. Being too vague or general
2. Overloading the prompt with too many variables
3. Neglecting to specify the desired outcome or format

Advanced Techniques

Try these advanced prompting techniques for even better results:

1. Conditional statements: 

"If [INDUSTRY] is finance, emphasize security features; otherwise, focus on ease of use."

2. Multiple options:

"Generate three different approaches for addressing [COMPANY]'s [PAIN_POINT]."

Evaluating and Iterating

After generating content:

1. Review variations for relevance and impact
2. Use the "what else would you change?" section to refine the personalization
3. Apply learnings to future prompts

Leveraging Account Data Effectively

• Prioritize variables that yield the best results (e.g., industry, company size, pain points)
• Combine multiple variables for more nuanced targeting
• Ensure your account data is complete and up-to-date for best results

FAQs

• How can I improve my prompts if I'm not getting the desired results?
  • Try being more specific in your instructions, experimenting with different account variables, or breaking down complex requests into smaller, more focused prompts. You can also reach out to your Growth Strategist for personalized advice.
• Where is the AI getting the Account information?
  • Mutiny's model is looking at any data aligned to an account via the Account list. This can be either through your CSV uploads or your first-party data integrations.

  • You can verify what account info is available to the model by checking the list of variables listed in the “insert a variable” dropdown or by navigating to the list view for this experience’s list to see what values are populated.

• How is the AI determining how to personalize more deeply?
  • The AI is using your first party data, as well as crawling your own website for information about how you may market yourselves to accounts based on variables. For example, if you have a page focused on servicing a certain vertical, that may factor into the AI model
• How do I add more information about an account for the model to personalize by?
  • You can sync more data via your first-party data integrations. See how to map fields for Salesforce or Hubspot. You can also upload a custom CSV with as many account fields as you'd like via Lists.
• Will this work on any element on the site?
  • Right now, this feature will work with text-based elements with at least 3 words. 
• Can we use this feature with components like sidepops?
  • This feature is only available for page content experiences.
• How do I discard and write a new prompt after applying variations to my experience?
  • If you'd like to discard, select "View variations" from the element and go back into the AI modal. There will be a discard option, which will take you back to the beginning of the generation process.

Remember, effective AI personalization is an iterative process. Don't be afraid to experiment and refine your approach based on the results you see.