Build org charts with Deepline

The problem with manual account mapping

Most B2B sellers do account mapping the same way: open LinkedIn Sales Navigator, search the company, export to a spreadsheet, manually research each person, guess who reports to whom.

This takes 1-2 hours per account. The data decays at 22.5% per year. By the time you finish 20 accounts, the first five are already stale.

Deepline's org chart builder does this in one command. It searches multiple providers, enriches each contact, infers hierarchy using title-based seniority classification, and outputs structured data you can push directly to your CRM.

How to run it

In Claude Code, use the /orgchart skill:

/orgchart Map the engineering leadership at stripe.com

Or start from a specific person:

/orgchart Build the org around linkedin.com/in/manpreet-singh at Stripe

The skill reads your target, searches for employees at the company, classifies seniority, and infers who reports to whom.

How it works

People discovery. Searches Apollo across multiple seniority tiers (executives, directors, managers) to find employees at the target company. Runs multiple passes to get broad coverage.

Profile enrichment. For each person found, enriches the profile with LeadMagic or similar providers to get full job titles, locations, and LinkedIn URLs.

Team clustering. Fetches job listings from Crustdata to understand team structure. Job postings reveal departments ("Engineering - Identity Platform") that titles alone miss.

Hierarchy inference. Classifies each person's seniority from their title, then predicts reporting lines using a scoring model.

The seniority classification model

Deepline classifies titles into 11 seniority levels:

The classification checks patterns in order. "Senior Vice President" matches before "Vice President." "Head of Engineering" maps to Director level.

The manager prediction model

Once everyone has a seniority level, Deepline predicts who reports to whom using a scoring system:

score = seniority_gap + team_match + geo_proximity

Seniority gap: +10 for exactly one level above. +5 for two levels. +2 for three. Your manager is usually one level above you, not three.

Team match: +8 for same team. +3 if teams overlap (e.g., "Identity" is a substring of "Identity Platform"). You probably report to someone in your own department.

Geo proximity: +2 for same city. +1 for same country. Remote teams exist, but co-located managers are more common.

The highest-scoring candidate above threshold (5 points) becomes the predicted manager. If no one scores high enough, the person is placed at the top of their branch.

Example output

For a target like "Manpreet Singh, Head of Engineering - Identity at Stripe," the builder returns:

{
  "people": {
    "manpreet-singh": {
      "name": "Manpreet Singh",
      "title": "Head of Engineering, Identity",
      "seniority": "director",
      "team": "Identity",
      "location": "San Francisco, CA",
      "linkedin": "linkedin.com/in/manpreet-singh"
    },
    "charles-huang": {
      "name": "Charles Huang",
      "title": "VP Engineering",
      "seniority": "vp",
      "team": "Engineering",
      "location": "San Francisco, CA"
    }
  },
  "hierarchy": {
    "root": "charles-huang",
    "target": "manpreet-singh",
    "edges": {
      "charles-huang": ["manpreet-singh", "other-director"]
    }
  },
  "summary": {
    "target": "Manpreet Singh",
    "manager": "Charles Huang",
    "peers": ["Other Director"],
    "direct_reports": ["Amy Seaman", "Dan Manager"],
    "total_people": 12
  }
}

The edges object maps parent to children. Claude Code can render this as an interactive HTML org chart or push the contacts to your CRM with hierarchy metadata.

Generating the visual

If you want an HTML visualization:

/orgchart Build org chart for Stripe engineering and generate HTML output

Claude Code produces a self-contained HTML file with interactive nodes. Click a person to see their details. The chart uses the same color coding by team that job listings revealed.

Accuracy and limitations

The org chart builder is inference-based. It does not have access to internal HR data. Accuracy depends on:

• Public data availability. If someone has no LinkedIn or no title in Apollo, they may be missing from the chart.
• Title clarity. "Director of Special Projects" is harder to classify than "VP Engineering."
• Company size. The model works best for companies with 50-500 employees. Very large companies have too many layers; very small ones have ambiguous titles.

For most B2B account mapping use cases, the inference is good enough. You get the buying committee structure in minutes instead of hours, and you can manually correct any errors before sending outreach.

Pushing to CRM

The org chart builder outputs structured JSON that Claude Code can push to HubSpot or Salesforce:

/deepline-gtm Push the enriched contacts to HubSpot with hierarchy tags

Each contact gets custom properties for:

• Seniority level
• Team/department
• Manager name (if inferred)
• Direct reports count

This makes it easy to filter your CRM by "all directors at target accounts" or "everyone who reports to the CTO."

Related plays

• Account mapping play - Full documentation for the account mapping workflow
• Company to contact waterfall - How the underlying contact discovery works
• Claude Code skills - All available skills including /orgchart
• Quick start - Install Deepline and run your first command

$ curl -s "https://code.deepline.com/api/v2/cli/install" | bash