Delaverse AI Knowledge Base

Back to Home

Knowledge Base

Security and Updates
Data Privacy and Securitysoon
Platform Version Historysoon
Roadmap for Future Updatessoon
Learning Resources
Video Tutorials (Beginner to Advanced)soon
Webinars and Podcastssoon
Free Articles and AI Resourcessoon
Case Studiessoon
Delaverse AI Platform v2.0

Customer Behavior Analysis and Sales Optimization

Build intelligent customer behavior analysis workflows that collect behavioral data from multiple sources, analyze patterns and churn risks, identify cross-selling opportunities, and deliver strategic insights for marketing, sales, product, and support teams with AI-powered optimization recommendations.

Customer Behavior Analysis and Sales Optimization workflow

Target Audiences

Marketing Teams

Personalize campaigns and increase conversion rates

Sales Teams

Identify cross-selling opportunities and retain customers

Product Teams

Align products with customer preferences effectively

Support Teams

Improve interactions and reduce churn rates

CEO

Make strategic decisions based on customer behavior insights

Prerequisites

Access to a Delaverse account
Customer behavioral data (e.g., purchases, interactions, support tickets) ready for upload
Google account access for Google Sheets and Google Docs Nodes
API access to customer databases (e.g., CRM) with authentication details (if used)
Access to Telegram with a bot token for the Telegram Node

Step 1: Create a New Project

Log in to Delaverse: Visit playground.delaverse.ai and sign into your Delaverse account
Select New Project: On the projects dashboard, click the New Project box
Name the Project: Enter a title, e.g., Customer Behavior Analysis and Sales Optimization, in the Project Name field and click Create Project
Enter the Playground: You'll be directed to the project's Playground, where you can add nodes

Step 2: Add Customer Behavioral Data Input Nodes

To analyze customer behavior, data is collected from multiple sources: Google Sheets, File Database, Google Docs, or API (e.g., CRM).

2.1. Google Sheets Node

Uploads tabular customer data

1
Add Node: From the right-hand menu, click Google Sheets Node
2
Configure Node: Click the node and enter a title, e.g., داده‌های مشتری (Customer Data)
3
Sign in with your Google account and create a new sheet
4
Enter data (e.g., purchases, interactions)
5
Click Update
6
Save Changes: Click Save Changes and close the editor with the top-left X button

Suggested Content:

Purchase Table: Columns for customer (e.g., ID123), purchase date (e.g., 1404/03/01), product (e.g., phone), amount (e.g., 5M IRR), quantity (e.g., 1)
Interaction Table: Columns for customer (e.g., ID123), date (e.g., 1404/03/01), type (e.g., call, email), subject (e.g., support)
Support Table: Columns for customer (e.g., ID123), ticket (e.g., TKT456), date (e.g., 1404/03/01), status (e.g., resolved)

2.2. File Database Node

Uploads data files (e.g., CSV or PDF of customer behavior)

1
Add Node: From the right-hand menu, click File Database Node
2
Configure Node: Click the node and enter a title, e.g., فایل‌های مشتری (Customer Files)
3
Upload files (e.g., CSV of purchases, PDF of interaction reports)
4
Review the list of uploaded files
5
Save Changes: Click Save Changes and close the editor with the top-left X button

Suggested Content:

CSV: Purchase reports with columns for customer, date, product, amount
PDF: Support or customer behavior analysis reports

2.3. Google Docs Node

Uploads text-based customer reports or notes

1
Add Node: From the right-hand menu, click Google Docs Node
2
Configure Node: Click the node and enter a title, e.g., گزارش‌های متنی مشتری (Customer Text Reports)
3
Sign in with your Google account and create a new document
4
Click Update
5
Save Changes: Click Save Changes and close the editor with the top-left X button

Suggested Content:

Support notes: Details of customer interactions (e.g., complaints about delays)
Text reports: Preliminary analysis of customer purchase behavior

2.4. API and Request API Nodes

Retrieve customer data from databases (e.g., CRM)

Add API Node:
Add API Node:
From the right-hand menu, click API Node Click the node and enter a title, e.g., API مشتریان (Customer API) Enter the base URL (e.g., https://crm.example.com/api) Configure Authentication: No Authentication, Basic Auth, Bearer Token, or API Key If headers or query parameters are needed, click Add Header, enter name and value Click Save Changes and close the editor with the top-left X button
Add Request API Node:
Add Request API Node:
From the right-hand menu, click Request API Node Click the node and enter a title, e.g., دریافت داده‌های مشتری (Fetch Customer Data) In the Used API section, select the previous API Node Choose the request type (e.g., GET) and enter the endpoint (e.g., /customers or /interactions) Configure tabs: Request, Headers, Parameters, Body, Response Click Execute Request to view results Connect the API Node's output to the Request API Node's input

Suggested Content:

Purchase data: Customer, date, product, amount
Interaction data: Customer, date, type (e.g., call, chat), subject
Support data: Customer, ticket, date, status
Note: Users can use Google Sheets, File Database, Google Docs, or API. Combining sources is possible but requires precise instruction tuning.
Important: Completing these steps is straightforward with the necessary access. However, if you need to connect an API through your system and lack the expertise, consult your company’s technical team. If you don’t have a technical team or they’re unavailable, submit a ticket to us. Request workflow creation by our team, and we’ll prepare a cost invoice based on your subscription level. After payment, we’ll work with you to build the workflow to meet your needs.

Step 3: Connect Data to Analysis Nodes

This step depends on whether you need real-time data updates. Choose one of the following approaches.

3.1. Connect Google Sheets to Vector Database Node (Static Data)

Add Node: From the right-hand menu, click Vector Database Node.
Configure Node: Click the node and enter a title, e.g., Customer Analysis Database. Click Update.
Connect Node: Connect the Google Sheets Node’s output to the Vector Database Node’s input.
Verify Data: Click the Vector Database Node to ensure data is correctly aggregated.
Save Changes: Click Save Changes and close the editor with the top-left X button.
Important Note: When connecting Google Sheets to the Vector Database, updating the data in Google Sheets does not automatically update the Vector Database, as the data is stored statically. To update the data:
Delete the file in the Vector Database associated with the Google Sheets.
Disconnect the Google Sheets and Vector Database nodes.
Update the Google Sheets data.
Reconnect the Google Sheets and Vector Database nodes to load the updated data.

3.2. Connect Google Sheets Directly to AI Assistant (Dynamic Data)

For real-time data updates without using the Vector Database:

Add Function Call Node:
From the right-hand menu, click Function Call Node.
Click the node, enter a title, and press the Parameters button.
For non-technical users: Select the “Read from Sheets” template to auto-configure parameters.
In the generated JSON schema, ensure the “name” parameter matches your preferred function name.
Copy your Google Sheet’s ID from the Google Sheets Node and paste it into the section marked your_sheet_id in the JSON schema.
For technical users: Click Build with AI, write your custom instructions in Persian, and click Build Smart Parameters.
Connect Nodes:
Connect the Google Sheets Node’s output directly to the AI Assistant Node’s input, and connect the AI Assistant Node’s output to the Function Call Node’s input.
Note: The Request API Node does not connect to the Vector Database Node; connect it directly to the AI Assistant Node.

Step 4: Add the AI Assistant Node

This node analyzes data and generates analytical reports for customer behavior, churn, and cross-selling:

Add Node: From the right-hand menu, click Analyzer Node
Configure Node: Click the node and enter a title, e.g., تحلیلگر رفتار مشتری(Customer Behavior Analyzer)
Select AI Model: Select an AI model (e.g., Open AI)
Instructions: Choose the Customer Behavior Analyzer template or edit the instruction (see below)
Connect Nodes: If using Google Sheets with static data, connect Vector Database to AI Assistant. If using Google Sheets with dynamic data, connect Google Sheets directly to AI Assistant via the Function Call Node. If using API, connect Request API directly to AI Assistant
Save Changes: Click Save Changes and close the editor with the top-left X button

Customer Behavior Analyzer Instruction Template

You are an AI assistant for analyzing customer behavior, churn rate, and identifying cross-selling opportunities. Your goal is to analyze customer behavioral data to identify patterns, at-risk customers, and suggest complementary products.
- Retrieve customer data from the Google Sheet with the following ID: {your google sheet ID}
- Retrieve customer ticket summary reports from the Google Doc with the following ID: {your google sheet ID}
- Retrieve customer interaction data from the file database with the following ID: {your google sheet ID}
- Store the data in a vector database.
- Perform the following analyses:
  - Customer Behavior: Identify patterns and preferences (e.g., frequent purchases of product X), categorize customers (loyal, occasional).
  - Churn Rate: Identify customers at risk of churn (e.g., reduced purchases or unresolved tickets), calculate churn probability.
  - Cross-Selling: Suggest complementary products/services (e.g., headphones for phone buyers).
  - Issues: Identify bottlenecks (e.g., slow support response times).
  - Suggestions: Recommend strategies for customer retention and sales growth (e.g., personalized discounts).
- Write responses in Persian with a professional and friendly tone.

Critical Function Call Requirements: After providing your analysis, you must immediately call the function {your function’s name} with the full analysis text as the content parameter. Do not call the function with empty parameters like {}.

Mandatory Steps:
1. First: Provide your complete analysis response in Persian.
2. Then: Immediately call the function {your function’s name} with the same full analysis text as the content parameter. Call the function with the following values:
  
   {
     "doc_id": "your doc id",
     "content": content
   }
The content parameter must be the complete analysis in Markdown format, not empty or summarized, but the full text.

Reminder: The function call is mandatory and must include your complete analysis as the content.
Important: Ensure the analysis result is sent to Telegram in its entirety.

Step 5: Deliver Reports

Users can receive reports via Telegram, Google Sheets, or Google Docs.

5.1. Telegram Node (Optional)

To deliver reports via Telegram:

Create a Telegram Bot:

• In Telegram, go to @BotFather, send /start, then /newbot• Choose a name, e.g., @CustomerBot, and copy the bot token

Configure Telegram Node:

• Add Node: From the right-hand menu, click Telegram Node• Enter the bot token and click Test Connection• Configure Advanced Settings (Welcome Message, Group Access, etc.)
Advanced Settings: Welcome Message:سلام! آماده گزارش رفتار مشتریان هستید؟ 😊 (Hello! Ready for customer behavior reports? 😊). Set Group Access, Access Restriction to All, Message Limit to 10 messages per minute.

5.2. Google Sheets Node (Optional)

To store reports in a table:

Add Google Sheets Node:

• From the right-hand menu, click Google Sheets Node• Enter title, e.g., گزارش رفتار مشتری (Customer Behavior Report)• Sign in and create new sheet with columns: customer, behavior, churn probability, cross-sell recommendation

Add Function Call Node:

• From the right-hand menu, click Function Call Node.• Then click the node, enter a title, and press the Parameters button.• For non-technical users: Select the "Write to Google Sheets" template to auto-configure parameters.• In the generated JSON schema, make sure the "name" parameter matches your preferred function name.• Copy your Google Sheet’s ID from the *Google Sheets Node* and paste it into the section marked your_sheet_id in the JSON schema.• For technical users: Click "Build with AI", write your custom instructions in Persian, and click "Build Smart Parameters".

5.3. Google Docs Node (Optional)

To store reports as text:

Add Google Docs Node:

• From the right-hand menu, click Google Docs Node• Enter title, e.g., گزارش متنی مشتری (Customer Text Report)• Sign in and create new document for report content

Add Function Call Node:

• From the right-hand menu, click Function Call Node.• Then click the node, enter a title, and click the Parameters button.• For non-technical users: Select the "Write to Google Docs" template to auto-configure parameters.• In the generated JSON schema, set the "name" parameter to match your preferred function name.• Copy the Google Doc’s ID from the *Google Docs Node* and paste it into the section marked google_doc_id in the JSON schema.• For technical users: Click "Build with AI", write your custom instructions in Persian, and click "Build Smart Parameters".

Step 6: Add Container and Analyzer Nodes (If Using Telegram)

For comprehensive data and conversation analysis:

Add Container Node

• From the right-hand menu, click Container Node• Click the node and enter a title, e.g., کانتینر رفتار مشتری (Customer Behavior Container)• Place all previous nodes (Google Sheets, File Database, Google Docs, API, Request API, Vector Database, AI Assistant, Telegram) inside the Container

Add Analyzer Node

• From the right-hand menu, click Analyzer Node• Click the node and enter a title, e.g., تحلیلگر جامع (Comprehensive Analyzer)• Choose the Customer Behavior Analyzer template or edit the instruction (see below)• Connect the Container Node's output to the Analyzer AI Assistant Node's input

Comprehensive Customer Analyzer Instruction Template

For Assistant Messages:
You are an AI assistant tasked with comprehensive analysis of customer behavior based on assistant messages.
Your goal is to generate key metrics, identify long-term trends, and provide recommendations for customer retention and sales growth.
Required Actions and Analyses:
    • Key Metrics: Churn rate, average purchase value, engagement rate.
    • Periodic Trends: Changes in behavior (e.g., reduced purchases) over months.
    • Issues: Bottlenecks (e.g., unresolved support tickets).
    • Recommendations: Strategic solutions (e.g., personalized campaigns).
    • Deliver outputs in text format for Telegram with a professional and friendly tone.

Step 7: Connect to the Frame Chat Node

This step enables delivering AI Assistant responses via an interactive chat frame on your website or app, with testing capabilities:

Basic Configuration

• Add Frame Chat Node from right-hand menuEnter title, e.g., Feedback FrameSet company name (e.g., Delaverse)

Domain & Style Configuration

• Domain & Style Configuration

Implementation & Testing

• Connect AI Assistant to Frame Chat• Go to Code tab and copy HTML code• Use Test Frame to preview responses

Testing and Verification

• After connecting to the AI Assistant, a Test Frame option appears• Enter sample queries to review responses• Ensure responses are accurate, professional, and aligned with input data• Click Save Changes to persist all settings

Step 8: Add the Trigger Node

To automate the initiation of your financial analysis and forecasting workflow with precise timing:

Add Node:

• From the right-hand menu, click Trigger Node and drag it onto the canvas in the Delaverse Playground.

Configure Node:

Click the node and enter a title, e.g., “تریگر گزارشگیری مدیریت” (Management Reporting and Analysis Trigger).

Timing Settings: In the “Timing Settings” section, choose the scheduling type from the dropdown: 

  • Interval: Select a value and unit (e.g., 1 day) for regular execution, and set the timezone (e.g., Asia/Tehran).
  • Cron Expression: Enter a Cron pattern, e.g., “0 8 * * *” for daily at 8 AM, following the provided example.
Message Text: Enter “مطابق دستورالعمل، عمل کن” (Act per instructions) to guide the AI Assistant Node in executing the workflow’s analysis and reporting tasks.

Maximum Executions: Leave blank for unlimited runs or enter a number (e.g., 10) to cap executions.

Trigger Status: Toggle the status button to “On” to activate the trigger.

Status Info: After setup, review execution history (e.g., runs completed) and click “Update Schedule” if timing changes are needed.

Connect Nodes:

• Connect the Trigger Node’s output to the AI Assistant Node’s input to initiate the workflow.

Connect Nodes:

•Configure timing and message text, then click “Test Trigger” to simulate a workflow run. Verify outputs appear in Telegram, Google Sheets, or Google Docs, depending on your setup.

Save Changes:

•Click “Save Changes” to apply all configurations, then close the editor with the top-left “X” button to return to the Playground.

Why It’s Needed:

•The Trigger Node automates the start of non-API workflows,working with the AI Assistant Node to initiate the workflow.

Key Tips for Success

Data Source Selection: Use Google Sheets, File Database, Google Docs, or API. Combining sources may require precise instruction tuning
Training Content: Use real data (e.g., purchases, interactions, tickets)
Persian Language: Keep all settings and outputs in Persian
Testing: Verify reports in Telegram, Google Sheets, or Google Docs
Continuous Saving: Click Save Changes after every modification

Final Output

Customer behavioral data is collected from Google Sheets, File Database, Google Docs, or API
Analytical reports (patterns, churn risks, cross-selling opportunities) are generated
Reports are delivered via Telegram (text responses), Google Sheets (tables), or Google Docs (text reports)
Comprehensive insights and strategic recommendations are provided through the Analyzer Node
General Note on Workflow Customization
This workflow, like all workflows in Delaverse’s knowledge base, is designed as a sample for educational purposes. Users can extensively customize nodes, settings, and input files to meet personal or organizational needs. For assistance, consult our 24/7 support chatbot or submit a ticket for guidance. If you’re unable to build or modify the workflow yourself, request professional workflow creation via a ticket; we’ll provide a cost invoice based on your subscription level, and after payment, we’ll collaborate to build it. For new node development (e.g., a custom node not yet available), submit a ticket with your requirements, noting that such requests may incur higher costs due to development efforts. Our team is here to ensure your automation success! 😊

Ready to Transform Your Customer Analytics?

Start building your intelligent customer behavior analysis workflow today. Analyze behavioral patterns, predict churn risks, identify cross-selling opportunities, and optimize sales strategies with AI-powered insights that drive customer retention and revenue growth.