Use this article to design effective Custom Scenarios for your AI Email Agent. These best practices help you create clear criteria, accurate responses, and conversation flows that move leads forward without unnecessary rework.
Before you start
Before creating Custom Scenarios:
Turn on AI Replies
Review your existing default and custom scenarios to avoid duplication
Start with a small number of scenarios and expand over time
Write effective criteria
Criteria determine when a scenario runs. Criteria reduce false positives and prevent overlapping scenarios.
.png)
Best practices
Write criteria that:
Use third‑person perspective
Start with:
Lead mentions…
Lead asks about…
Lead requests…
Describe a specific intent or phrase
Account for common variations in how leads express the same need
Examples
Good criteria
Lead mentions they need pricing information for enterprise plans
Lead asks about integration capabilities with Salesforce
Criteria to avoid
When you see pricing questions (incorrect POV)
They want to know more information (too vague)
Write clear Agent Response (Response Instructions)
Agent response directs the AI Email Agent how to respond when a criteria matches. Well‑written instructions produce consistent, on‑brand replies.
.png)
Best practices
When writing Agent Response:
Write as direct instructions to the AI Email Agent
Include all necessary context in the instruction
Specify:
Links
Resources
Attachments
Define the tone and approach, for example, helpful, consultative, urgent, and so on.
State exactly what information must appear in the response
Important
Custom Scenarios are not linked to campaign documents. Include all required information directly in the Agent Response.
Recommended structure
Use this structure to keep instructions clear and complete:
Context – What background the agent needs to know
Message content – What specific information to communicate
Resources – Links, documents, or materials to include
Call to action – What to ask the lead to do next
Example
Agent Response
Respond with a helpful tone acknowledging their interest in pricing. Explain that our enterprise plans start at $5,000/month and include dedicated support, custom integrations, and unlimited users. Include this link to our pricing page: [www.company.com/pricing]. Mention that our sales team can create a customized quote based on their specific needs and offer to schedule a 15-minute consultation call.
Decide whether to continue or end the conversation
Each Custom Scenario must either continue the conversation or close it.
.png)
Continue Conversation
Choose Continue Conversation when you want the lead to reply again.
Use this option when:
The lead is early in their buying journey
The question requires follow‑up
You want to gather more information before sales engagement
How it works
When the lead replies, their message is evaluated against all reply scenarios again. This allows the AI Email Agent to adapt as intent changes.
This allows the AI to:
Detect new topics or questions the lead raises
Adapt to changing lead intent
Guide leads through a natural conversation journey
Trigger different scenarios based on how the conversation evolves
Examples
Lead asks general product questions → Continue (nurture their interest, allow them to ask follow-ups)
Lead asks about use cases → Continue (gather more context, let conversation develop)
Sample scenario:
Criteria: Lead mentions they're evaluating multiple solutions
Agent Response: Share a comparison guide and case studies relevant to their industry. Ask about their key decision criteria and timeline to help them evaluate effectively.
Status: Continue Conversation
When lead replies, their message are evaluated against all scenarios again - they might ask about pricing, request a demo, or share specific concerns, triggering the appropriate response.
End Conversation
Choose End conversation and update its status in AI Email when no further AI interaction is needed.
Use this option when:
The lead requests direct human contact
The issue requires sales or legal involvement
The question is fully resolved
The lead unsubscribes or declines further contact
How it works
When you select End conversation and update its status in AI Email, the conversation ends with the AI's response. The lead's reply won’t go through the scenario evaluation flow again. You must assign an AIE status to categorize the conversation outcome.
Examples
Lead asks for custom contract terms → Close (requires human negotiation)
Lead's question fully answered with resource → Close (if no follow-up needed)
Test and refine your scenarios
Testing helps you catch issues before scenarios go live.
Best practices
Use Preview to test multiple lead messages
Confirm:
The correct criteria fires
The response includes all required details
Links and resources work as expected
Refine Agent Response if the response does not match your intent
Common issues to avoid
Criteria that are too broad
Missing links or context in Agent Response
Closing conversations too early
Creating scenarios that compete with each other
Use Criteria to refine default scenarios
Criteria make default scenarios more specific.
.png)
How Criteria works
A scenario runs only when the lead meets both conditions:
Default criteria + additional criteria = Scenario match
Example
Default scenario: Qualified
Default criteria: Lead shows positive intent
Additional criteria: Lead mentions needing more than 50 licenses
In this case, the scenario runs only if the lead meets both criteria.
Pre‑launch checklist
Before enabling Custom Scenarios, confirm that:
Create scenarios you can anticipate leads will reply to. Design scenarios that naturally invite responses and keep conversations flowing.
Avoid duplicate scenarios. Do not create multiple custom scenarios with the same meaning or that would trigger on identical lead messages.
Test each scenario thoroughly. Use the Preview feature to verify each scenario triggers correctly and generates appropriate responses.
Start small: 1-3 scenarios first. We recommend creating and testing 1-3 custom scenarios before building more. This allows you to refine your approach and understand what works best.
Use Criteria to make default scenarios stricter.