Email Templates: consistent replies in seconds
Citai Team
April 10, 2026 · 10 min read
The problem: every agent replies differently
Without templates, each team member writes emails in their own style. The result: brand inconsistency, incomplete information, and wasted time reinventing responses.
How it works
Create templates
Each template has: name, subject (with variables), HTML body with variables in double braces, language (ES/EN/PT), and category (welcome, support, follow-up, confirmation, farewell).
Dynamic variables
Use double-brace variables: {{client_name}}, {{client_email}}, {{company_name}}, {{subject}}, {{date}}. They auto-replace with thread data.
Seed templates
Citai generates predefined templates per language at account creation. They’re editable and don’t count against plan limits.
Usage from threads
In a thread, click “Insert template” to select and auto-fill a template with thread data.
Variable system: dynamic fields in detail
Variables are the heart of templates. Here is the complete system:
Contact variables
| Variable | Description | Example |
|---|---|---|
{{client_name}} |
Thread contact name | “Maria Garcia” |
{{client_email}} |
Contact email | “[email protected]” |
{{subject}} |
Active thread subject | “Premium package inquiry” |
Company variables
| Variable | Description | Example |
|---|---|---|
{{company_name}} |
Your tenant/company name | “Acme Corp” |
{{date}} |
Formatted current date | “April 14, 2026” |
Custom variables
You can create your own variables in the template body. When inserted into a thread, if a variable has no automatic data, an editable field appears for the agent to fill manually before sending.
Example: {{order_number}}, {{product}}, {{refund_amount}} — these don’t auto-fill but standardize the structure.
Measuring template effectiveness
Key metrics to monitor
- Response time: Compare average response time before and after implementing templates. Typical reduction: 60-80%
- Consistency: Check if responses maintain the same tone, information, and structure. Without templates, agent variation can be 40%+
- Reuse rate: What percentage of responses use templates? Usage below 30% indicates templates don’t cover frequent cases
- Customer satisfaction: Standardized, complete responses improve CSAT. Monitor negative feedback post-implementation
How to optimize
- Review most-used templates: If a template is used 50 times a month, invest time perfecting it
- Identify gaps: If agents write manual responses for a recurring topic, create a new template
- Informal A/B testing: Try two versions of the same template and compare positive response rates
Before and after examples
Example 1: Support response (before)
“Hi, if your order hasn’t arrived I’d suggest waiting a few more days. If it still doesn’t come let us know. Regards.”
Problems: No personalized greeting, no reference number, no concrete timeline, no signature.
Example 1: With template (after)
"Dear {{client_name}},
Thank you for contacting us about your order. We’ve checked the status and the estimated delivery time is 5-7 business days from the purchase date.
If your order doesn’t arrive within that timeframe, please reply to this email with your order number and we’ll arrange a priority review.
We’re at your disposal.
{{signature}}"
Improvements: Personalized, informative, clear action, professional.
Example 2: Sales follow-up (before)
“Hey! Just writing to see if you saw our proposal. Hope to hear from you!”
Example 2: With template (after)
"{{client_name}}, good morning.
I’m writing to follow up on our proposal sent on {{proposal_date}} regarding {{subject}}.
Have you had a chance to review it? I’m available for any questions or to schedule a clarification call.
{{signature}}"
Template maintenance and versioning
Best practices
- Quarterly review: Schedule a review every 3 months to update information (prices, policies, products)
- Ownership: Assign a responsible person per category. Support team maintains support templates, sales maintains commercial ones
- Internal changelog: Document changes to each template. When updating a seed template, notify the team
- Don’t duplicate: If two templates are very similar (>80% identical content), merge them into one with conditional variables
When to create vs update
| Situation | Action |
|---|---|
| Price/policy change | Update existing template |
| New product or service | Create new template |
| General tone change | Update all templates in the category |
| Recurring negative feedback | Rewrite the template with improvements |
| New language required | Create translated version of key templates |
Common mistakes when creating templates
- Too generic: “Thanks for writing, we’ll respond soon” adds no value. Each template should solve something specific
- Too rigid: If the template leaves no room for personalization, agents will avoid it. Include editable sections marked with
[customize here] - Variables without fallback: If you use
{{order_number}}but the agent doesn’t have it, the response gets sent with the placeholder visible. Always verify before copying - Ignoring cultural context: A formal English template (“Dear Sir/Madam”) may sound odd in casual markets. Create regional variants if needed
- Not updating after changes: The template says “shipping in 3-5 days” but the policy changed to 5-7 days. Periodic review is critical
Sidebar > Email Assistant > Templates
Try Citai for free
Create your intelligent knowledge base in minutes. No credit card required.
Create free accountRelated articles
Email Assistant: reply to customer emails with AI in seconds
New Citai module that generates professional email replies using your knowledge base. Pre-analysis, 4 tones, embedded images and full history.
Read article → ProductResponse Style: customize your AI agent's voice
Configure how your email agent writes: signature, preferred tone, custom instructions, format, and length of generated responses.
Read article → ProductEmail Threads: manage complete conversations with AI context
Organize emails into threads with automatic analysis, smart alerts, in-thread search, and response generation with full conversation context.
Read article →