Dialogflow CX vs ES Complete Comparison: 2026 Version Selection Guide

"What's the difference between CX and ES? Which one should I choose?"

This is the most common question from people new to Dialogflow. Choosing the wrong version is costly: choose ES and find it lacks needed features, and you'll have to redo the entire project; choose CX but only use basic features, and you'll spend several times more money each month.

This article tells you the differences between the two versions in the most direct way and provides a decision framework to help you choose.

To first understand the basics of Dialogflow, refer to the Dialogflow Complete Guide.

---

Version Evolution History

From API.AI to Dialogflow

Dialogflow's predecessor was API.AI, a startup focused on natural language understanding. After being acquired by Google in 2016, it was renamed Dialogflow and integrated into the Google Cloud ecosystem.

Post-acquisition Dialogflow (now called ES) grew rapidly, becoming one of the most widely used conversational bot platforms in the market. But as enterprise customers increased, ES's architectural limitations gradually emerged.

Why Google Released the CX Version

ES's architecture was designed for simple conversations. When enterprises wanted to do:

• Complex multi-turn conversations (10+ steps)
• Multi-person collaborative development
• Rigorous version control and testing processes
• Cross-department conversation flow integration

ES became very difficult to maintain. Intents kept increasing, Context relationships became increasingly messy, eventually becoming "only the original author can understand" situations.

In 2020, Google released Dialogflow CX, specifically to solve these problems. CX is not an upgrade of ES, but a completely redesigned architecture.

Illustration: Dialogflow Version Evolution Timeline

Scene Description: A horizontal timeline chart showing from left to right: API.AI founded in 2014, acquired by Google and renamed Dialogflow in 2016, Dialogflow CX released in 2020, current situation in 2025 (both versions running in parallel). Each node has brief explanatory text.

Visual Focus:

• Main content clearly presented

Required Elements:

• Based on key elements in description

Chinese Text to Display: None

Color Tone: Professional, clear

Elements to Avoid: Abstract graphics, gears, glowing effects

Slug: dialogflow-version-timeline

---

Dialogflow ES (Essentials) Deep Dive

ES is the "classic" Dialogflow, suitable for most small to medium conversational bot projects.

Core Features Introduction

Intent-based Architecture

The core of ES is Intent. Each Intent contains:

• Training Phrases: What users might say
• Response: Bot's reply
• Context: Conversation context
• Parameters: Extracted information

One-Click Integration

ES has built-in multi-platform integration, completable in the Console:

• LINE
• Facebook Messenger
• Telegram
• Slack
• Web Widget

Small Talk

ES provides preset chitchat functionality. The bot can respond to common greetings like "hello," "thank you," "who are you" without setting them up one by one.

Knowledge Connector

You can import FAQ documents, and Dialogflow automatically generates corresponding Intents. Suitable for quickly building knowledge base Q&A.

Use Cases

Scenario	Description
Simple FAQ Bot	Auto-replies for 20-30 common questions
Single Function Bot	Single process like reservations, queries, registrations
MVP Validation	Quick validation of chatbot business viability
Personal Projects	Learning, experimentation, side projects
Limited Budget	Free tier of 1,000 requests per month

Limitations and Drawbacks

1. Conversation Flow Difficult to Visualize

ES has no flow chart editor. Conversation logic is scattered across various Intents and Contexts, making it hard to grasp the full picture as projects grow.

2. Context Management Difficult

ES uses Context to control conversation flow, but Context relationships can only be remembered mentally or documented externally. After 50+ Intents, Context relationships become very complex.

3. Limited Version Control

ES can only export/import JSON, without built-in version management. Multi-person collaboration easily leads to overwriting each other's work.

4. Basic Testing Features

Only basic conversation testing panels, without batch testing, regression testing, and other advanced features.

5. No Native WhatsApp Integration

If you want to integrate WhatsApp Business, ES requires building your own middleware service.

---

Dialogflow CX Deep Dive

CX is designed for enterprise-level complex conversations, with a completely different architecture from ES.

Visual Flow Editor

CX's biggest feature is the visual editor. Conversation flows are presented as flowcharts, designable through drag-and-drop.

Flow: Split conversations into multiple independent flows, such as "Reservation Flow," "Query Flow," "Complaint Flow."

Page: Nodes within each Flow, representing conversation states.

Route: Connections between Pages, defining transition conditions.

This architecture makes complex conversations intuitive. Even with 100 conversation nodes, you can understand the overall logic by looking at the flowchart.

Multi-Flow Parallel Processing

ES conversations are linear, only in one flow at a time.

CX supports multi-Flow parallelism. Users can ask about "business hours" while in the "reservation flow," and the system will pause the reservation, answer the question, then return to the reservation flow. This is closer to real conversation situations.

Advanced Version Control

CX has built-in version control system:

Feature	Description
Version	Create snapshots for Flows, can rollback anytime
Environment	Create development, testing, production environments
Traffic Split	A/B test different conversation flow versions

Teams can test new features in development environment, then deploy to production after confirmation. No worry about breaking live services.

Conversation History and Analytics

CX provides detailed conversation analytics dashboard:

• Conversation success rate
• User drop-off point analysis
• Intent trigger statistics
• Conversation path visualization

This data helps you continuously optimize conversation design.

RAG Knowledge Base Integration

CX supports Data Store integration, connecting to:

• Website content
• PDF documents
• Structured data

The bot searches for relevant information from the knowledge base and combines answers using generative AI. No need to create Intents for every question.

Illustration: Dialogflow CX Flow Editor Interface

Scene Description: A screenshot of the Dialogflow CX Console showing the visual flow editor. The screen shows a complete reservation flow: from "Start" → "Ask Date" → "Ask Party Size" → "Confirm Reservation" → "Complete," each node represented by boxes, arrows connecting to show flow direction.

Visual Focus:

• Main content clearly presented

Required Elements:

• Based on key elements in description

Chinese Text to Display: None

Color Tone: Professional, clear

Elements to Avoid: Abstract graphics, gears, glowing effects

Slug: dialogflow-cx-flow-editor-interface

To learn complete CX operations, refer to Dialogflow CX Tutorial: From Beginner to Advanced.

---

Feature Comparison Table

Feature	Dialogflow ES	Dialogflow CX
Architecture Design	Intent-based	Flow-based
Visual Editor	No	Yes
Multi-Flow Support	No	Yes
Version Control	Export/Import	Built-in Version Management
Environment Management	No	Yes (Dev/Test/Prod)
A/B Testing	No	Yes
Conversation Analytics	Basic	Advanced Dashboard
Batch Testing	No	Yes
RAG Knowledge Base	Knowledge Connector	Data Store + Generative AI
LINE Integration	Built-in	Requires Custom Middleware
Messenger Integration	Built-in	Requires Custom Middleware
WhatsApp Integration	Requires Custom	Native Support
Voice Integration	Supported	Supported (More Features)
Multi-language Support	Supported	Supported
Maximum Intent Count	2,000	Unlimited (Uses Flow/Page)
Team Collaboration	Limited	Fully Supported
API Version	v2	v3
Learning Curve	Gentler	Steeper

---

Pricing Comparison

Pricing is an important consideration in version selection. The two versions have different billing models.

ES Pricing Model

ES charges by request count:

Type	Free Tier	Overage Cost
Text Requests	1,000/month	$0.002/request
Voice Input	15,000 seconds/month	$0.0065/15 seconds
Voice Synthesis	-	$0.004/second

Features:

• Free tier resets monthly
• Suitable for scenarios with many conversation turns but few total conversations
• Cost grows linearly with request count

CX Pricing Model

CX charges by Session (conversation) count:

Type	Cost
Text Session	$20/1,000 Sessions
Voice Session	$45/1,000 Sessions
Generative AI	Additional charges

Features:

• New accounts get $600 USD free credits (within 12 months)
• One conversation counts as one Session regardless of turns
• Suitable for complex scenarios with many conversation turns

Cost Calculation Examples

Scenario: E-commerce Customer Service Bot

• 5,000 conversations per month
• Average 8 turns per conversation
• All text conversations

ES Cost Calculation:

Total Requests = 5,000 × 8 = 40,000 requests
Billed Requests = 40,000 - 1,000 = 39,000 requests
Monthly Cost = 39,000 × $0.002 = $78 USD

CX Cost Calculation:

Total Sessions = 5,000
Monthly Cost = 5,000 / 1,000 × $20 = $100 USD

In this case, ES is about 22% cheaper than CX.

Scenario: Complex Consultation Bot

• 2,000 conversations per month
• Average 20 turns per conversation
• All text conversations

ES Cost Calculation:

Total Requests = 2,000 × 20 = 40,000 requests
Billed Requests = 40,000 - 1,000 = 39,000 requests
Monthly Cost = 39,000 × $0.002 = $78 USD

CX Cost Calculation:

Total Sessions = 2,000
Monthly Cost = 2,000 / 1,000 × $20 = $40 USD

When conversation turns are many, CX is actually more cost-effective.

Illustration: ES vs CX Cost Comparison Chart

Scene Description: A dual-axis chart, X-axis is "Monthly Conversations," Y-axis is "Monthly Cost (USD)." Two lines in the chart: blue line represents ES (steeper slope), orange line represents CX (gentler slope). The two lines cross at a point labeled "Break-even Point." Below the chart is text explaining recommendations for different scenarios.

Visual Focus:

• Main content clearly presented

Required Elements:

• Based on key elements in description

Chinese Text to Display: None

Color Tone: Professional, clear

Elements to Avoid: Abstract graphics, gears, glowing effects

Slug: dialogflow-es-cx-cost-comparison-chart

For more detailed cost calculations and money-saving tips, refer to Dialogflow Pricing Complete Analysis.

---

Can't figure out the costs? Dialogflow billing has many details, and actual costs may differ significantly from estimates. Book a free consultation to have us help you calculate costs precisely and choose the right version.

---

Version Selection Recommendations

Scenarios for Choosing ES

1. Simple Conversation Flow

If your bot only needs to:

• Answer 20-30 common questions
• Handle single process (reservation, query, registration)
• Conversations don't exceed 5 turns

ES's features are more than sufficient.

2. Limited Budget

ES has 1,000 free requests per month, very cost-effective for small projects or MVP validation.

3. Need to Launch Quickly

ES has a gentler learning curve, and built-in platform integrations let you quickly connect LINE, Messenger.

4. Team Has No Dedicated Developers

ES can complete basic features with almost no coding, suitable for non-technical backgrounds.

Scenarios for Choosing CX

1. Complex Conversation Flow

If your bot needs:

• 10+ conversation branches
• Complex conditional logic
• Multi-turn deep conversations

CX's visual editor makes development and maintenance much easier.

2. Multi-Person Collaborative Development

Teams of more than 2 people developing chatbots simultaneously need CX's version control and environment management.

3. Need WhatsApp Business Integration

Native WhatsApp integration is only supported in CX. If your target market is Southeast Asia, Europe, or South America, WhatsApp is a necessary channel.

4. Enterprise-Level Requirements

Rigorous testing processes, A/B testing, detailed conversation analytics—only CX has these features.

5. Expected Long-Term Maintenance

If this bot will be iterated for years, CX's architecture is easier to maintain and extend.

Decision Flowchart

Start
  │
  ▼
Does conversation flow exceed 10 branches?
  │
  ├─ Yes → Choose CX
  │
  └─ No → Continue evaluation
          │
          ▼
        Need WhatsApp integration?
          │
          ├─ Yes → Choose CX
          │
          └─ No → Continue evaluation
                  │
                  ▼
                Team has more than 2 people developing simultaneously?
                  │
                  ├─ Yes → Choose CX
                  │
                  └─ No → Continue evaluation
                          │
                          ▼
                        Limited budget?
                          │
                          ├─ Yes → Choose ES
                          │
                          └─ No → Choose CX (leave room for future expansion)

Illustration: Version Selection Decision Tree

Scene Description: A decision tree flowchart displaying version selection logic from top to bottom. Each diamond node is a question, rectangular nodes are conclusions (Choose ES or Choose CX). Nodes are connected by arrows labeled "Yes" or "No." Overall color scheme is clear and easy to read.

Visual Focus:

• Main content clearly presented

Required Elements:

• Based on key elements in description

Chinese Text to Display: None

Color Tone: Professional, clear

Elements to Avoid: Abstract graphics, gears, glowing effects

Slug: dialogflow-version-decision-tree

---

FAQ

Can ES Projects Migrate to CX?

No direct migration possible. ES and CX have completely different architectures (Intent-based vs Flow-based), and there's no automatic conversion tool. If switching from ES to CX, you need to redesign the conversation flow.

It's recommended to evaluate clearly at the project's beginning to avoid having to redo later.

Can You Use Both ES and CX?

Technically yes, but not recommended. The two versions have different APIs, SDKs, and Consoles—maintaining two systems simultaneously increases complexity.

How Steep Is CX's Learning Curve?

If you're familiar with ES, transitioning to CX takes about 1-2 weeks to adapt to new concepts (Flow, Page, Route). If learning from scratch, CX's official tutorials take about 20-30 hours.

What If You Choose the Wrong Version?

• ES not enough, want to switch to CX: Need to redesign, but can reference ES Intent logic
• CX too expensive, want to switch to ES: Need to simplify flow, break complex logic into multiple simple Intents

Both situations require significant work, so choosing correctly from the start is important.

Will Google Stop Supporting ES?

No indication of this currently. Google continues updating ES; the two versions are designed for different needs. But looking long-term, CX receives more investment in new features.

---

Next Steps

After selecting a version, you can start learning in depth:

Chose ES:

• Dialogflow Complete Guide - Learn from the beginning
• Dialogflow Intent and Context Tutorial - Master core concepts
• Dialogflow LINE Bot Integration Tutorial - Quick launch

Chose CX:

• Dialogflow CX Tutorial: From Beginner to Advanced - Complete CX learning
• Dialogflow Pricing Complete Analysis - Calculate CX costs precisely

---

Still Not Sure Which Version to Choose?

Version selection affects the entire project's development approach and cost structure. If you:

• Aren't sure about your conversation flow complexity
• Uncertain about future expansion needs
• Want someone to help evaluate

Book a free consultation to have experienced people analyze your needs and choose the right version.

We'll respond within 24 hours, and consultation is completely free.

---