How to Audit Your Own HubSpot Portal Using AI Tools

Modern AI assistants and HubSpot's Model Context Protocol make it possible to perform comprehensive portal audits without hiring expensive consultants. Here's your practical guide to getting started.

As HubSpot consultants, we've seen countless portals over the years, from pristine setups to digital dungeons where workflows go to die. The common thread? Most businesses have no clear picture of what's actually happening under the hood of their CRM. Until recently, comprehensive portal auditing required either expensive external consultants or countless hours of manual detective work.

That's changing fast.

With the emergence of AI powered tools and HubSpot's Model Context Protocol (MCP), businesses can now perform sophisticated portal audits themselves. While this doesn't replace the need for expert strategic guidance, it can provide valuable insights and help you identify issues before they become problems.

What to expect in this blog post?

• Why most HubSpot portals need regular health checks
• The AI powered audit approach
• Your HubSpot audit checklist
• Setting up your DIY audit environment
• What you can discover
• Common issues AI audits uncover
• What AI auditing can't replace
• When to go beyond DIY
• Getting started today
• The future of HubSpot auditing

Why most HubSpot portals need regular health checks

HubSpot grows with your business, but that growth often creates complexity that goes unnoticed:

• Marketing emails with declining performance
• Workflows that were disabled during updates but never re-enabled
• Forms collecting data that nobody monitors
• Social accounts disconnected after password changes
• Integrations that stopped syncing properly
• Properties created for one-time campaigns but never cleaned up

The traditional approach involves manually checking dozens of different screens across multiple hubs, a time-consuming process that most teams postpone until problems arise.

The AI powered audit approach

The Model Context Protocol (MCP) creates a direct connection between AI assistants like Claude and your HubSpot data. Instead of clicking through countless screens, you can ask questions (known as prompting) like:

• "Which workflows haven't run in the last 30 days?"
• "What forms have the highest abandonment rates?"
• "Are all my social accounts still connected?"
• "Which email campaigns are underperforming compared to industry benchmarks?"

This systematic approach mirrors HubSpot's own audit methodology but automates the data collection and analysis.

Screenshot from Claude Desktop executing the prompt "Let's do a HubSpot audit on my portal (using MCP)"

Your HubSpot audit checklist

Based on HubSpot's own audit framework, here are the key areas you should regularly review:

1. Account & license efficiency

• Subscription levels: Are you on the right tier for your needs?
• Contact limits: How close are you to your tier limits?
• Feature utilization: Are you paying for features you're not using?
• User seat optimization: Do all users need their current access level?

AI Assistant Query: "Analyze our subscription usage and identify underutilized paid features."

Screenshot from Claude Desktop showing an excerpt of above's prompt with ROI Optimization opportunities

2. Technical foundation & security

• Domain connections: Are your domains properly connected?
• Tracking code: Is it installed on all relevant pages?
• Email authentication: Is your sending domain configured correctly?
• User permissions: Are access levels appropriate for current roles?
• API access: Are there unused private apps or tokens that should be deactivated?

AI Assistant Query: "Check our domain connections, user permissions, and identify any security configuration gaps."

3. Data integrity analysis

• Duplicate records: How many duplicate contacts or companies exist?
• Data completeness: Which important properties have low completion rates?
• Data consistency: Are there formatting inconsistencies across similar fields?
• Record associations: Are contacts properly linked to companies and deals?

AI Assistant Query: "Perform a data quality assessment and highlight integrity issues across our database."

4. Marketing email performance

• Deliverability rates: How do your bounce rates compare to industry standards?
• Engagement metrics: Are open and click rates declining over time?
• List health: Do you have segments that haven't been engaged recently?
• Suppression management: Are unsubscribe processes working correctly?

AI Assistant Query: "Analyze our email performance over the last 6 months and flag any concerning trends or deliverability issues."

5. Forms and conversion points

• Form performance: Which forms have the highest abandonment rates?
• Lead flow: Are leads being properly assigned and followed up on?
• Conversion tracking: Are all important conversions being captured?
• Data capture: Are forms collecting the right information for your sales process?

AI Assistant Query: "Show me form performance data and identify optimization opportunities."

6. Workflow and automation health

• Active vs inactive: Which workflows were disabled and why?
• Error rates: Are any workflows failing silently?
• Performance: Which automations are taking too long to execute?
• Logic validation: Are workflow conditions still relevant to current processes?

AI Assistant Query: "Audit all workflows and highlight any that need attention or have performance issues."

7. Integration status & data processing

• Connected apps: Which integrations are active and syncing properly?
• Data flow mapping: Where is data coming from and going to?
• Sync frequency: Are integrations running at optimal intervals?
• API limits: Are any integrations approaching usage limits?
• Data processing compliance: Are data transfers between systems properly documented?

AI Assistant Query: "Review all integrations, map data flows, and flag any sync issues or compliance concerns."

8. Social media and advertising

• Account connections: Are all your social accounts still connected?
• Publishing performance: Which platforms are generating the best engagement?
• Ad account status: Are your advertising integrations working properly?

AI Assistant Query: "Check social media connections and advertising account performance."

9. Data retention & compliance

• Data retention policies: Are you keeping data longer than necessary or required?
• Contact lifecycle: Do you have processes for managing inactive contacts?
• Deletion workflows: Are there automated processes for data cleanup?
• Audit trails: Can you track data changes and access patterns?

AI Assistant Query: "Review our data retention practices and identify any compliance risks or cleanup opportunities."

Setting up your DIY audit environment

Prerequisites:

• HubSpot admin access
• Basic comfort with technical setup (no coding required)
• Claude Pro subscription (for MCP support)

Step 1: Create a HubSpot Private App

1. Navigate to your HubSpot settings → Integrations → Private Apps
2. Create a new app with read-only scopes for:
   • CRM objects and schemas
   • Workflows and automation
   • Users and teams
   • Custom objects

Step 2: Set Up MCP Connection

1. Download Claude for Desktop (or any other client that supports MCP)
2. Edit the JSON configuration file (in Claude Desktop: Settings → Developer → Edit Config)
3. Add the below JSON code in the configuration file and replace "your-token-here" with... your token provided in Step 1

{
  "mcpServers": {
    "hubspot": {
      "command": "npx",
      "args": ["-y", "@hubspot/mcp-server"],
      "env": {
        "PRIVATE_APP_ACCESS_TOKEN": "your-token-here"
      }
    }
  }
}