Train AI with CRM data
Leverage Merge’s CRM API endpoints to efficiently train AI models around prospect recommendations, win/loss patterns, and more
1. Authentication and connection
Your users authorize your application to access their CRM platform by going through an authorization flow using Merge Link.
2. Data extraction from CRM
Use the GET /Accounts endpoint to retrieve account-related data.
Fetch contact details using the GET /Contacts endpoint.
For transactional data or interaction histories, utilize the GET /Opportunities and GET /Tasks endpoints.
3. Data pre-processing
Clean the data by removing duplicates in your customer data, handling missing values, and filtering out irrelevant fields. Transform data into a format suitable for AI training, possibly involving normalization, encoding, or feature engineering.
4. Model training
Split the CRM data into training and testing sets. Use the training set to train an AI model tailored for tasks like customer churn prediction, sales forecasting, or lead scoring.
5. Model evaluation
Validate the model’s performance using the testing set. Compute relevant metrics to gauge the model’s effectiveness.
6. Feedback loop
Closely monitor ongoing deals and opportunities for continuous AI training and insights. For example, identify win/lost patterns based on number of touchpoints, number of involved stakeholders and their seniority, revisions around close data and contract value. Continuously train your AI model using Object Tracking from Merge to power insights.
Use Remote Fields to send back any AI insights back to the customer’s CRM platform. For example, AI predictive close percentage.