Suggestion Chips
Suggestion chips are quick-reply buttons that appear in chatbots to help users respond faster by tapping preset options instead of typing.
What Are Suggestion Chips?
Suggestion chips are selectable, visually distinct, pill-shaped elements that present users with fast, predefined reply options in chatbots and conversational UIs. These chips appear below or near a bot’s prompt and represent the most likely or contextually appropriate user responses. When tapped or clicked, the chip’s value is submitted as the user’s answer, and the chips disappear to keep the interface uncluttered.
Purpose:
- Accelerate input by providing one-tap answers
- Guide users toward valid inputs, reducing ambiguity and errors
- Increase conversation completion rates by preventing dead-ends
Example:
Bot: “What size pizza would you like?”
Suggestion Chips: [Small] [Medium] [Large]
Key Characteristics of Suggestion Chips
Ephemeral: Chips are usually visible only for a single prompt and disappear after selection or user input, helping keep the conversation clean and focused.
Pre-Defined: The set of options is generated by the chatbot logic, tailored to the current conversation context.
Non-Persistent: Chips are not permanent navigation or action buttons—they exist only for the relevant dialog turn.
Visual Distinction: Pills or bubbles, often rendered according to Material Design guidelines, with clear padding and separation from the main chat stream.
One-Tap Selection: Users can respond instantly, reducing typing effort and mistakes.
Adaptive: Chips can be contextually generated, showing only relevant or valid options based on conversation state or user profile.
Suggestion Chips vs Other UI Elements
Suggestion Chips vs Buttons
| Feature | Suggestion Chips | Buttons |
|---|---|---|
| Appearance | Pill/bubble, ephemeral | Rectangular, may be persistent |
| Disappear After Tap | Yes | Not always, often persistent |
| Use Case | Lightweight, context-sensitive choices | Navigation, actions, persistent |
| Placement | Below prompt/input area | Inline with cards, messages, footers |
| Platform Example | Dialogflow, Messenger Quick Replies | Messenger, Telegram, Web widgets |
Suggestion chips are for quick, contextual, one-off decisions; buttons are for ongoing actions or navigation.
Suggestion Chips vs Quick Replies
Terminology: “Quick replies” is the term for this pattern on Facebook Messenger and Telegram. Google names them “suggestion chips” in Dialogflow and Assistant.
Behavior: Both chips and quick replies disappear after use, are limited in number, and are intended for rapid input.
Technical Note: Some platforms allow quick replies to return structured payloads or collect user data.
Suggestion Chips vs Persistent Menus
Menus: Always visible, providing global navigation/actions (e.g., “Help,” “Settings,” “Restart”).
Suggestion Chips: Ephemeral, tied to the latest prompt.
Suggestion Chips vs Smart Chips
Smart Chips: In Google Docs, these are document-embedded elements for links, people, or files—not related to conversational UI.
Technical Implementation
General Workflow
- Bot Prompt: The chatbot sends a message requiring input
- Display Chips: The client displays suggestion chips below the prompt
- User Action: User taps a chip; the chip’s value is sent to the backend
- Bot Response: The bot processes the input and responds; chips disappear
Platform-Specific Guidance
Dialogflow
Native Support: Suggestion chips are supported in Dialogflow CX and ES integrations for Google Assistant, web chat, and compatible channels.
Technical Limits:
- Up to 8 suggestion chips per prompt
- Only available on devices with
actions.capability.SCREEN_OUTPUT - At least one simple response is required before chips
- Chips not allowed in a
FinalResponse
Dynamic Chips: Use webhooks to generate context-sensitive chips in response payloads.
Design: Pair chips with cards or responses; do not repeat the same info in both.
Facebook Messenger
Quick Replies:
- Up to 13 quick replies per message
- Each reply can have text, an optional image, and a payload
- Disappear after selection or free-form input
Persistent Buttons: For navigation or static actions, not ephemeral input.
Telegram
ReplyKeyboardMarkup:
- Provides custom keyboards for one-tap replies
- Keyboards can be set as “one-time”, causing them to disappear after a reply
- Keyboard buttons can be arranged in rows, and the keyboard can be hidden after use
Inline Keyboard: Used for persistent, action-triggering buttons.
Sample Code: Dialogflow Webhook (Node.js)
// Sample fulfillment snippet for suggestion chips in Dialogflow webhook
const {WebhookClient, Suggestion} = require('dialogflow-fulfillment');
function sendSuggestionChips(agent) {
agent.add("What size pizza would you like?");
agent.add(new Suggestion('Small'));
agent.add(new Suggestion('Medium'));
agent.add(new Suggestion('Large'));
}
The Suggestion class is available in the dialogflow-fulfillment library. Chips can be generated dynamically based on user input or context.
Best Practices for Suggestion Chips
Limit the Number: Present three to five options at most. Too many choices overwhelm users and degrade UX.
Be Contextual: Show only relevant options based on user state or conversation context.
Keep Labels Short: Use concise chip text (under 20 characters) to prevent wrapping or layout issues.
Enable Disambiguation: Use chips to clarify ambiguous user input or to confirm detected entities.
Accessibility: Ensure chips have sufficient color contrast, are screen-reader friendly, and have clear labels.
Handle Free-Text Inputs: Validate and process user input even if they ignore chips and type a response.
Platform Compliance: Respect platform limits for chip count and label length.
Common Pitfalls and How to Avoid Them
Too Many Chips: Avoid presenting long lists. Stick to 3–5 options.
Disconnected Chips: Every chip must trigger a valid handler or intent; never leave a chip unlinked.
Persistent Chips: Ensure chips disappear after selection to prevent confusion.
Unvalidated Input: Always handle cases when users type their own response instead of selecting a chip.
Vague Labels: Use clear, context-specific text (“Yes”, “No”, “Medium”) rather than ambiguous phrases.
Ignoring Platform Limits: Each platform has chip/quick reply count and character limits (e.g., 8 for Dialogflow, 13 for Messenger).
Use Cases and Practical Scenarios
Slot Filling in Ordering Flows
Prompt: “Choose a drink size.”
Chips: [Small] [Medium] [Large]
Outcome: User taps a chip to fill the slot without typing.
Disambiguation After Invalid Input
Prompt: “Sorry, please select vehicle type:”
Chips: [Car] [Truck]
Quick Surveys or Feedback
Prompt: “Was this helpful?”
Chips: [Yes] [No]
Branching Conversation Paths
Prompt: “What would you like to do?”
Chips: [Browse] [Order Status] [Support]
Collecting Sensitive Data
Prompt: “Share your contact info.”
Chips: [Use my email] [Use my phone number]
Note: Chips can be pre-filled with user data where supported.
Implementation Tips
Design Considerations
Visual Hierarchy: Ensure chips are visually distinct but not overwhelming
Touch Targets: Make chips large enough for easy tapping on mobile devices
Spacing: Provide adequate spacing between chips to prevent accidental taps
Responsive Design: Ensure chips work well across different screen sizes
Performance Optimization
Lazy Loading: Load chip options only when needed
Caching: Cache frequently used chip configurations
Async Updates: Update chips asynchronously to avoid blocking UI
Analytics and Monitoring
Track Selection Rates: Monitor which chips are most/least selected
Identify Drop-offs: Detect when users ignore chips and type instead
A/B Testing: Test different chip configurations for optimal engagement
Advanced Features
Dynamic Chip Generation
Generate chips based on user context, conversation history, or external data sources.
Localization
Support multiple languages by dynamically loading localized chip labels.
Personalization
Customize chip options based on user preferences or past behavior.
Progressive Disclosure
Show basic options initially, with an option to reveal more advanced choices.
FAQ: Suggestion Chips
Q: What are the main benefits of using suggestion chips?
A: Faster, error-resistant, and structured input for users, leading to higher conversation success rates.
Q: How do suggestion chips differ from buttons?
A: Chips are ephemeral and prompt-bound; buttons are often persistent and used for navigation or static actions.
Q: Can users ignore chips and type their own response?
A: Yes. Always validate input for both chip selection and free-form text.
Q: What if my platform doesn’t support chips natively?
A: Use quick replies or similar elements, or design custom pill-shaped buttons that mimic chip behavior.
Q: What about accessibility?
A: Ensure good color contrast, readable fonts, and ARIA labels as per Material Design accessibility guidelines.
Summary Table
| Feature | Suggestion Chips | Buttons | Quick Replies | Persistent Menu |
|---|---|---|---|---|
| Appearance | Pill, ephemeral | Rectangular, persistent | Pill/bubble, ephemeral | List, always visible |
| Disappear on selection | Yes | Often no | Yes | No |
| Use case | Fast, guided input | Navigation, actions | Fast input | Global actions |
| Max options (FB) | 13 (recommended 3-5) | 3 per card | 13 (3-5 optimal) | - |
Key Takeaway
Use suggestion chips (quick replies) for rapid, context-driven user input. Limit the number for clarity, keep labels concise, and always handle free-form input.
References
- Dialogflow Rich Responses Documentation
- Material Design 3: Chips
- Facebook Messenger Quick Replies
- Telegram Bot API: ReplyKeyboardMarkup
- Dialogflow Fulfillment Library (Node.js)
- Activechat: Chatbot Buttons vs Quick Replies
- Stack Overflow: Slot Filling Chatbot, Suggestion Chips in Prompts
- YouTube: Dialogflow Chatbot Bot Prompt Suggestion Chips
- Google Docs: Insert Smart Chips
- Wikipedia: Natural Language Understanding
- Dialogflow: Slot Filling
- Activechat: Privacy Policy
- Activechat: Terms of Service
Related Terms
Conversational AI
AI technology that understands and responds to human conversation through text or voice, learning fr...
False Positive
A false positive is when an AI system incorrectly flags something as matching a criterion when it ac...
Quick Replies
Quick Replies are temporary buttons in chat apps that offer preset answer options. They disappear af...
Task-Oriented Bot
A chatbot designed to help users complete specific tasks like booking flights, tracking orders, or s...
Voiceflow
A no-code platform that lets teams build AI chatbots and voice assistants by dragging and dropping c...
Contact Deflection
A customer service strategy that helps people solve problems on their own through self-service tools...