Skip to main content

How to Set Up Data Collection

Nikita Podelenko avatar
Written by Nikita Podelenko
Updated over 2 weeks ago

How to Set Up Data Collection

Data Collection allows you to automatically extract structured information from your AI-powered calls. Instead of manually reviewing transcripts to find specific details, SkipCalls AI can identify and save key data points like budgets, dates, preferences, and contact information directly after a call finishes.

Why use Data Collection?

  • No manual note-taking: Focus on the conversation while the AI handles the data entry.

  • Consistent data format: Ensure all your call data follows the same structure for better reporting.

  • Easy CRM integration: Automatically push extracted data to your CRM via webhooks.

  • Better follow-up capabilities: Quickly see the most important details of a call at a glance.

How to access

  1. Navigate to the Agents section in your dashboard.

  2. Select the agent you want to configure.

  3. Click on the Collections tab.

Adding a data field

To start collecting data, you need to define individual "fields" that the AI should look for during the call.

  1. Click the Add Field button.

  2. Field Name: Enter a unique identifier (e.g., customer_email, budget_amount). We recommend using lowercase with underscores.

  3. Data Type: Select the format of the data:

    • Text: For names, emails, addresses, or general comments.

    • Number: For budgets, price ranges, or decimal values.

    • Integer: For counts, ages, or whole numbers.

    • Yes/No: For boolean questions (e.g., "Is the customer interested?").

    • Date: For appointment dates, deadlines, or availability.

  4. Extraction Instructions: This is the most important part. Tell the AI exactly what to look for and how to interpret the conversation (see tips below).

  5. Click Add Field to save.

Writing good extraction instructions

The better your instructions, the more accurate the data extraction will be. Be specific about the context and edge cases.

Example for a budget field:
​ Extract customer's budget. Look for: explicit amounts, price ranges, or phrases like "can spend up to". Return the maximum amount mentioned as a number. Return null if not discussed.

Best practices:

  • Be specific about what keywords or phrases the AI should watch for.

  • Give examples of how the information might be mentioned in a conversation.

  • Specify what to return if the information is missing (usually "null").

Built-in Examples

You can use these common examples to get started quickly:

Field Name

Type

Instructions

customer_budget

Number

Extract budget amount. Look for price ranges or "up to X".

preferred_date

Date

Find appointment or installation date mentioned by the customer.

has_existing_system

Yes/No

Does the customer mention currently having a similar system?

Where collected data goes

Once a call is completed, the AI analyzes the transcript and populates your defined fields:

  • Call Details: View the extracted data directly in the call history section.

  • Webhooks: If you have webhooks configured, the structured data is sent in the payload.

  • CRM Sync: Data can be automatically mapped to your integrated CRM fields.

Tips for success

  • Start small: Begin with 3-5 most important fields and refine them before adding more.

  • Test and iterate: Review the extraction results after your first few calls and adjust the "Extraction Instructions" if needed.

  • Use snake_case: Using names like appointment_date instead of Appointment Date makes it easier to work with integrations.

Did this answer your question?