Infer's Model Build Process
2 likes654 views
Infer's model building process leverages historical data from automation systems and external customer signals to train predictive models that score leads and contacts. The approach incorporates over 1,000 relevant signals, including company revenue and employee count, to enhance model accuracy and drive customer insights. Continuous model review and testing ensure the effectiveness of the predictions and their applications in acquiring data.
Infer's Model Build Process
- 1. Infer’s Model Build Process
- 2. Proven to Create Lift Customer Data Pre-Built Connectors API Open • Historical data from your automation systems is used to train the predictive models • Connectors sync the data so that Infer is able to score leads and contacts and push updates
- 3. Automation data alone is not enough
- 4. You’ve got to be vicious about acquiring data
- 5. External Signals Customer Data • Employee Count • Company Revenue • Sells Globally • First Trademark < 7 Years Ago • Is Hiring Support Agents • Credible Domain • Actively Advertising • Recently Expanded to New Office • Raised Funding of $20-$100M • Featured in Inc. 5000 • Uses Amazon Web Services • Not Spammy 1,000+ Signals External Signals
- 6. Machine Learning Data ScienceExternal Signals Customer Data Machine Learning External Signals Customer Data Machine learning looks at all of the data and determines what signals are statistically relevant
- 7. Customer Data External Signals Machine Learning Historical Backtesting Train Set Test Set Used to train the predictive model Used to test the accuracy of the predictive model Testing Framework Machine Learning External Signals Customer Data
- 8. Customer Data External Signals Machine Learning Historical Backtesting Model Review • What signals are used in the model • How accurate is the model at predicting winners • Recommendations for applying the scores Review
- 9. Proven to Create Lift