Crm Integration

CRM Integration

Overview

Most CRM problems are process problems disguised as software problems. Teams add fields, automations, and sync rules before deciding what the CRM is supposed to be the source of truth for.

This skill designs a CRM integration that supports GTM execution without turning into rep admin theater: clear lifecycle stages, field ownership, enrichment sync, dedupe, and stage-gate rules.

When to Use

Use this skill when the user asks to "set up our CRM for sales", "configure HubSpot", "configure Salesforce", "set up Attio", "design deal stages", "integrate enrichment with CRM", "clean up CRM data", "define lifecycle stages", or "build CRM reporting".

Use crm-toolkit for platform selection, contacts-vs-leads architecture, and blueprints — then platform setup skills (hubspot-setup, salesforce-setup, attio-setup) for rollout.

Authoritative Foundations

Winning by Design — Bowtie Model

CRM design should cover the full bowtie: acquisition, conversion, retention, expansion, and advocacy. Most teams only model pipeline and ignore post-sale stages.

HubSpot — Lifecycle Stage Model

HubSpot's lifecycle model is useful for SMB and mid-market teams because it connects marketing, sales, and CS handoffs with clear stage definitions.

Salesforce — Opportunity and Object Architecture

Salesforce works best when object relationships, required fields, and automation rules are designed before admins start customizing pages.

Attio — Programmable CRM Model

Attio's flexible object model works well for teams that need relationship intelligence, custom objects, and lightweight GTM workflows without Salesforce overhead.

Prerequisites

• Defined GTM motion: PLG, sales-led, founder-led, or hybrid
• ICP tiers and segments
• Sales stages and exit criteria
• CRM owner/admin
• Enrichment sources
• Sequencer and marketing automation tools
• Reporting requirements

Step-by-Step Process

Phase 1: Define Source-of-Truth Rules

No integration should write fields it does not own.

Phase 2: Build Lifecycle Stages

Recommended stages: subscriber/raw lead, lead, MQL, SQL, opportunity, customer, expansion candidate, evangelist/referral source. Each stage needs entry criteria and exit criteria.

Phase 3: Design Deal Stages

Enterprise motions can add MEDDICC fields. SMB motions should keep required fields lighter.

Phase 4: Integrate Enrichment

Enrichment writes only enrichment-owned fields, never overwrites manually corrected CRM fields without confirmation, uses source + timestamp columns, and exposes failed enrichment as a visible status.

Phase 5: Reporting and Governance

Minimum dashboards: funnel conversion by lifecycle stage, pipeline by stage and owner, stage aging, source-to-revenue attribution, data completeness, enrichment coverage and freshness.

Output Format

markdownCopy code
# CRM Integration Blueprint

## Source-of-Truth Rules
| Field Group | Source | Allowed Writers | Overwrite Rule |
|---|---|---|---|

## Lifecycle Stages
| Stage | Entry Criteria | Exit Criteria | Owner |
|---|---|---|---|

## Deal Pipeline
| Stage | Required Fields | Exit Criteria | Automation |
|---|---|---|---|

## Integration Map
| Tool | Reads From CRM | Writes To CRM | Risk |
|---|---|---|---|

Quality Check

Before delivering, verify:

• Every stage has entry and exit criteria
• Field ownership is defined
• Enrichment cannot overwrite human-owned fields silently
• Required fields are minimal enough reps will use them
• CRM supports marketing, sales, and CS handoffs
• Dashboards cover funnel, pipeline, stage aging, and data quality

Common Pitfalls

1. Two-way sync everywhere. Creates conflicts and loops. Fix: define read/write ownership per tool.
2. Too many required fields. Reps stop updating CRM. Fix: require only fields needed for the next stage.
3. No stage exit criteria. Pipeline becomes subjective. Fix: write objective exit rules.
4. CRM as the process. Tools do not create process. Fix: define process first, configure second.
5. No data freshness. Old enriched fields linger forever. Fix: source and timestamp every enrichment field.
6. No lost reason taxonomy. Closed-lost learning disappears. Fix: standardized lost reasons with notes.

Execution Artifacts

• references/framework-notes.md — named frameworks, citation anchors, and operating assumptions
• templates/output-template.md — copy-paste deliverable structure for the user
• scripts/check-output.py — local checklist validator for required sections This skill includes lightweight artifacts the agent can load on demand: Use the artifacts when the user asks for an implementation-ready deliverable, a repeatable workflow, or a quality check rather than generic advice.

Related Skills

• hubspot-setup — HubSpot-specific implementation
• salesforce-setup — Salesforce-specific implementation
• attio-setup — Attio-specific implementation
• gtm-operations — governance and operating cadence