01
CRM systems collect data. People still have to interpret it.
Sales teams receive leads from forms, imports, events, partners and many other sources. A record may contain:
- Company name
- Job title
- Notes
- Product interest
- Free-form description
- Industry description
Someone often has to decide what that record actually means. Is this a healthcare company? An enterprise opportunity? A partner lead? A renewal? A high-value inbound request?
02
Train from decisions your team has already made
Suppose your CRM contains thousands of accounts that sales operations has already assigned to customer segments. That can become training data.
| Company Description | Segment |
|---|---|
| Regional hospital network | Healthcare |
| Industrial pump manufacturer | Manufacturing |
| Online apparel marketplace | Retail |
StayCharted AMT can learn the relationship between the descriptive data and the category your business uses.
03
Common CRM predictions
A StayCharted AMT model could help predict:
- Industry
- Lead type
- Customer segment
- Territory group
- Opportunity type
- Product interest
- Routing queue
- Account tier
Use StayCharted AMT to fill recurring classification fields in your existing CRM.
04
Fit it into the existing workflow
Teams can start with CRM exports in CSV or XLSX. Once the model proves useful, an integration can make the process automatic.
- Website / Lead Source
- CRM
- StayCharted AMT Prediction API
- Segment / Category / Route
- CRM Record Updated
This makes StayCharted AMT useful even when Salesforce, HubSpot or another system remains the center of the sales process.
There is no built-in Salesforce or HubSpot connector. Start with file exports, or have a developer call the API and apply its category suggestions in your CRM.
05
Start with one field
You don't need to clean up your entire CRM at once. Pick one field that people repeatedly fill in manually. Train a model for that decision. Measure the results. Then decide whether another field deserves its own model.