Language support

Learn about supporting synthetic customers who speak languages other than English.

Traits

Think in archetypes and conversational styles

The most useful personas are defined by communication behavior, not by demographics (for example, age or gender). Focus on traits that actually change how the conversation plays out.

Effective trait focus:

• Patience level, vocabulary complexity, emotional volatility
• How they escalate (or don't)
• Whether they provide information proactively or need to be asked

Less useful trait focus:

• Age, gender, occupation (unless relevant to the scenario)
• Physical appearance
• Hobbies unrelated to the conversation

Use short, descriptive phrases

• Good example: Impatient, direct, uses short sentences, escalates quickly if not helped within 2–3 exchanges, refers to previous negative experiences.
• Suboptimal example: This person is kind of impatient and doesn't really like waiting around for things.

Include behavioral cues, not just adjectives

Describe what the persona does, not just what they are. Good examples include:

• Asks the same question repeatedly if the answer isn't clear
• Responds with single words or very short phrases
• Threatens to switch providers when frustrated

Define escalation and de-escalation triggers

Tell what makes this persona more upset and what calms them down, for example:

• Becomes cooperative when the agent uses their name and acknowledges the inconvenience
• Gets increasingly frustrated if put on hold or asked to repeat information

Typical responses

Include three (3) to five (5) dialogue snippets to anchor the persona’s voice. Ensure they:

• Vary by mood: Include a calm opening, a frustrated middle, and a satisfied close.
• Sound human: Use contractions, slang, and natural imperfections rather than "textbook" English.

Examples for an "Impatient Executive" persona:

• I don't have time for this.
• Can you just fix it?
• I've already explained this twice.
• Look, I'm in a meeting in 10 minutes.
• Fine, that works. Thanks.

Examples for a "Confused Customer" persona:

• I'm not sure what you mean by that.
• Can you explain that more slowly?
• I think I pressed the wrong button.
• Oh, thank you so much, dear.

Persona maintenance

• Audit for distinction. Review your simulation results. If every conversation sounds the same regardless of the character, your personas are too generic. Sharpen their unique traits to ensure they sound like distinct individuals.
• Separate style from situation. To keep your data clean and effective, ensure each category sticks to its specific role: The persona is about the communication style, tone, and behavioral traits, The scenario is about the specific situation, the problem, and the customer's goal. Avoid adding the scenario’s emotional context (for example, "is angry about a bill") to the persona’s permanent traits.
• If you're creating different personas for different channels, assign tags to classify them. This makes retrieval easier when needed.