Advanced interaction settings

    All CollectionsAgentsAdvanced interaction settings

    Advanced interaction settings

    This guide covers all the detailed settings for controlling how your agent interacts with users on social media.

    "How do I control which posts my agent responds to?"

    Use the Respond on setting (Facebook Only):

    • Every post: Your agent will respond to comments on all posts (ads and organic)
    • Only ads: Respond exclusively to comments on sponsored/promoted posts
    • Only organic: Respond only to comments on non-sponsored posts

    This is helpful when you want to prioritize responses on specific content types.

    "Can my agent be selective about which comments to answer?"

    Yes! Configure the Response Frequency:

    • Always: Respond to every comment that meets your criteria
    • When relevant: Only respond when AI determines a reply adds value (e.g., skip emoji-only comments or simple acknowledgments)

    "Should my agent respond to comments that were hidden?"

    The Respond to Moderated Comments toggle controls this:

    • When ON: Your agent will still generate replies to comments that were hidden by moderation
    • When OFF: Hidden comments won't receive responses

    This is useful if you want to keep engagement with commenters whose content was automatically moderated.

    "How do I control my agent's communication style?"

    The Tone of Voice setting has five options:

    • Direct: Straightforward, no-nonsense, factual, and clear
    • Formal: Business-like tone with professional expertise
    • Balanced: Neutral approach that maintains engagement
    • Warm: Friendly atmosphere that builds genuine connections
    • Lighthearted: Witty and cheerful with a dash of entertainment

    Choose based on your brand personality and audience expectations.

    "Can I make my agent sound more human?"

    Configure these settings for a more natural feel:

    • Use Emojis: Toggle ON to allow emojis in responses
    • Response Time:
      • "Natural" adds varying delays to mimic human behavior
      • "As soon as possible" responds immediately

    "How does my agent handle multiple languages?"

    Use the Language setting:

    • Detect automatically: The AI determines the appropriate language based on the comment (recommended)
    • Specific language: Force the agent to use a particular language for all responses

    "Can I control how detailed my agent's responses are?"

    The Answer Length setting gives you three options:

    • Concise: Short, to-the-point responses
    • Standard: Balanced length (recommended for most cases)
    • Detailed: Comprehensive, in-depth responses

    Publishing modes and confidence-based automation

    "What's the difference between 'All replies' and 'High confidence' publishing?"

    ReplyZen offers two publishing modes when you have both "Enable Replies" and "Post Automatically" turned on:

    • All replies: Every AI-generated response is published immediately (full automation)
    • High confidence: Only publishes replies when the AI is confident about the response quality

    "How does confidence-based publishing work?"

    When using "High confidence" mode, your agent analyzes each reply and assigns it a confidence score. Replies above your threshold are published automatically, while those below are saved as drafts for your review.

    What affects confidence scores:

    • Comment clarity and context
    • Whether the agent has relevant knowledge to provide a good answer
    • Complexity of the question or topic
    • Potential for misunderstanding or sensitive content

    "What confidence levels should I choose?"

    You can select from four preset confidence levels:

    • Permissive (50%): Publishes most replies automatically, only saves very uncertain ones for review. Good for high-volume pages where speed matters more than perfection.

    • Balanced (65%): Recommended for most users. Provides a good balance between automation and quality control, catching moderately uncertain replies while still automating the majority.

    • Cautious (80%): Only publishes confident replies. Best for brands that prioritize response quality or handle sensitive topics where accuracy is critical.

    • Strict (95%): Only publishes very confident replies. Ideal for highly regulated industries, premium brands, or complex B2B scenarios where every response must be perfect.

    "Will I still get responses for low-confidence comments?"

    Yes! Your agent still generates responses for all eligible comments. Low-confidence replies simply appear in your Comments dashboard under the "Draft" section instead of being published immediately. You can review, edit, and approve them manually.

    This means you never miss an opportunity to engage—you just get to ensure quality for uncertain cases.

    Advanced link handling

    "How can I make my agent include links in responses?"

    The Append Link setting (Facebook Only) controls this:

    • No: Never include links
    • When relevant: Let AI decide if a link makes sense in context
    • Always: Include a link in every response

    "Where do these links come from?"

    Use the Link Source setting:

    • Automatic: AI chooses the most relevant link based on context
    • Fixed link: Always use a specific URL you define

    "Can I track which links come from my agent's replies?"

    Yes! Use URL Parameters to add tracking to all links in your agent's responses:

    • Toggle ON: Automatically adds tracking parameters to every link your agent includes in replies
    • Configure: Click to set up specific parameters like UTM tracking for Google Analytics

    This works independently of link shortening - you can track links whether they're shortened or not.

    Common use cases:

    • Track conversions from social media comment responses
    • Measure engagement from different campaigns or post types
    • Identify which agent responses drive the most traffic

    Example: A link like https://yoursite.com/product becomes https://yoursite.com/product?utm_source=facebook&utm_campaign=agent_replies&utm_content=commenter_123

    "What about link analysis and shortening?"

    • Scrape Post Link Content: When ON, analyzes URLs in the original post to provide better context
    • Link Shortening: When ON, automatically shortens links using cml.sh

    FAQ: Common interaction questions

    "Will my agent like comments automatically?"

    If you enable Like Comments Automatically (Facebook Only), your agent will like all incoming comments.

    "My agent is responding too much/too little. How do I adjust this?"

    Try these steps:

    1. Adjust your Response Frequency setting
    2. If using "When relevant," try switching to "Always" for more responses
    3. Check if you have enabled responses on the right post types

    "Why isn't my agent including links in responses?"

    Check these settings:

    1. Make sure Append Link is set to "When relevant" or "Always"
    2. Verify your Link Source setting is configured correctly
    3. For Facebook-specific features, remember they don't work on Instagram due to API limitations

    "How do I make my agent sound more like my brand?"

    For best results:

    1. Choose the appropriate Tone of Voice
    2. Add brand-specific information in the Additional Context & Instructions
    3. Build a comprehensive Knowledge Base with your brand's details and preferred responses

    "My agent is publishing too many/too few replies automatically. How do I adjust this?"

    If using "High confidence" mode:

    1. Too many low-quality replies being published: Choose a stricter confidence level (Cautious or Strict)
    2. Too many good replies ending up in drafts: Choose a more permissive confidence level (Permissive or Balanced)
    3. Want more control: Switch to "All replies" OFF mode to review every response manually
    4. Want full automation: Switch to "All replies" mode to publish everything immediately

    "Will confidence-based publishing work with my existing moderation settings?"

    Yes! Confidence-based publishing works alongside all your moderation settings:

    1. Comments are first checked against your moderation rules
    2. If approved for response, the AI generates a reply and assigns a confidence score
    3. High-confidence replies are published automatically
    4. Low-confidence replies are saved as drafts for your review

    Moderation and publishing confidence are independent systems that work together.

    "How do URL Parameters work with different link sources?"

    URL parameters are added to all links your agent includes in replies, regardless of the link source:

    • Automatic links: Parameters are added to AI-selected links from your post content
    • Fixed links: Parameters are added to your manually defined link
    • Shortened links: Parameters are applied first, then the link is shortened (preserving the tracking)

    "What tracking parameters should I use?"

    The default configuration includes standard UTM parameters:

    • utm_source: Platform (facebook/instagram)
    • utm_medium: Always "social"
    • utm_campaign: "agent_replies"
    • utm_content: Unique commenter ID
    • utm_term: Post ID

    You can customize these or add your own parameters like campaign_id, ad_group, or any other tracking values your analytics tools need.

    NEXT ARTICLE

    Advanced moderation settings