Custom Data Definitions

Custom data definitions teach Pingd's AI about your specific data — what fields mean, how your team uses terminology, and business rules that affect interpretation. Better definitions lead to more accurate, relevant answers.

Why Data Definitions Matter

Your CRM data is unique to your business. Field names like BANT_Score__c or Tier_Level mean nothing to an AI without context. Data definitions bridge that gap.

Without definitions:

"What's the BANT score for Acme Corp?"
"I found a field called BANT_Score__c with value 78, but I'm not sure what this represents."

With definitions:

"What's the BANT score for Acme Corp?"
"Acme Corp has a BANT score of 78/100, which is strong. This measures Budget (confirmed), Authority (VP-level sponsor), Need (active pain point), and Timeline (Q2 decision). Scores above 70 typically indicate a well-qualified deal."

Column Mappings

With multiple data sources (Salesforce, Snowflake, S3, etc.), the same business concept often has different field names across systems. Column mappings solve this by creating logical names that work across all your data sources.

Cross-Connector Translation

For example, "account owner" might be called:

• Account.OwnerId in Salesforce
• OWNER_EMAIL in your Snowflake accounts table
• rep_email in your S3 exports

Create a mapping called account_owner that points to all three, and your agent will understand they're the same thing.

Setting Up Mappings

1. Go to Settings → Data → Column Mappings
2. Click Create Mapping
3. Enter a logical name (e.g., account_owner)
4. Map it to fields from each connected data source:

- Salesforce: Account.Owner.Email - Snowflake: accounts.OWNER_EMAIL - S3: account_data.rep_email

1. Add a description: "Email address of the account's assigned sales rep"

Benefits of Mappings

• Unified queries — "Show me John's accounts" works regardless of which system the data comes from
• Cross-system insights — Compare Salesforce opportunities with Snowflake pipeline data seamlessly
• Simpler maintenance — Change field names once in mappings, not in every custom instruction

Setting Up Definitions

Navigate to Settings → Data → Custom Definitions.

Field Definitions

For each custom field, you can provide:

• Display Name — Human-friendly name (e.g., "BANT Qualification Score" instead of BANT_Score__c)
• Description — What this field represents and how it's used
• Value Interpretation — What different values mean (e.g., "Above 70 = strong, 40-70 = moderate, below 40 = weak")
• Business Context — How your team uses this in practice

Terminology Glossary

Define company-specific terms your team uses:

Term	Definition
"Whale account"	Enterprise account with ARR potential > $500K
"Red zone"	Deal at high risk of churning or lost
"Champion"	Internal advocate at the prospect company
"Land and expand"	Start with small deal, grow within account

Add terms at Settings → Data → Glossary.

Business Rules

Define rules that help the AI interpret data correctly:

• "Deals in 'Verbal Commit' stage are 90%+ likely to close"
• "Accounts marked 'Strategic' should always be flagged in pipeline reviews"
• "If an account has no activity in 30+ days, it's considered 'dormant'"
• "Our fiscal year starts in February"

Bulk Import

For teams with many custom fields:

1. Go to Settings → Data → Custom Definitions
2. Click Export Template to get a CSV template
3. Fill in definitions for your fields
4. Upload the completed CSV

Best Practices

1. Start with the top 20 — Define the fields your team asks about most. You don't need to define every field on day one.

1. Include examples — "Revenue Band: A = $1M+, B = $500K-$1M, C = under $500K" is better than "Revenue classification."

1. Describe relationships — "Territory is linked to the rep's region, which determines their manager and quota."

1. Update regularly — When you add custom fields to your CRM, add definitions here too.

1. Get input from reps — Ask your team what fields confuse them. If it confuses a human, it definitely confuses the AI.

Pro Tip: Great data definitions make your agent dramatically more useful. Spending 30 minutes on definitions is the highest-ROI configuration you can do.

How Definitions Are Used

When your agent processes a query:

1. It identifies which fields are relevant to the question
2. It loads the definitions for those fields
3. It interprets the raw data using your definitions
4. It responds using your terminology and business context

This means the agent speaks your team's language, not database column names.

See also: Connecting Your CRM · Configuring Your Agent · Writing Custom Instructions