Configuring Salary Benchmarking for the /apply Workflow

The salary benchmarking feature is an optional layer on top of the /apply fit evaluation. When configured, it shows how a target company’s compensation compares to a baseline — helping you calibrate whether a role is financially competitive before you invest time in a full application. The feature is entirely data-source agnostic: it works with union salary statistics, Glassdoor exports, manually researched figures, or any other benchmark you can express in the expected JSON format.

How It Works

salary_lookup.py reads a salary_data.json file in the repo root and matches company names using fuzzy logic. The matcher handles:

Danish and Nordic characters — ø, æ, å, ö, ä, ü are matched against their anglicized equivalents (o, ae, aa, etc.) so “Ørsted” matches “Oersted”
Legal suffixes — common Danish corporate suffixes (A/S, ApS, I/S, P/S, IVS, AMBA) are stripped before matching so “Novo Nordisk A/S” matches “Novo Nordisk”
Noise patterns — geographic suffixes (Danmark, Denmark, Nordic, Scandinavia), group/holding suffixes, and parenthetical sub-entity labels are stripped before comparison
Partial word overlap — multi-word company names are matched by shared core words, with a minimum score threshold to filter out weak matches

During /apply Step 1 (fit evaluation), if salary_data.json exists, Claude runs salary_lookup.py for the target company and includes the result in the fit evaluation output. If the file is missing, the salary step is silently skipped and the rest of the workflow continues normally.

salary_data.json Format

The tool expects a JSON file with a metadata block describing the data source and baseline, and a companies array with one entry per company:

{
  "metadata": {
    "source": "My Union Statistics 2025",
    "index_baseline": 100,
    "index_label": "Index",
    "baseline_description": "Index 100 = median salary for private sector"
  },
  "companies": [
    {
      "company": "Novo Nordisk A/S",
      "city": "Bagsværd",
      "categories": {
        "all_employees": { "count": 500, "index": 108.5 },
        "engineering": { "count": 120, "index": 112.3 }
      }
    },
    {
      "company": "Ørsted A/S",
      "city": "Fredericia",
      "categories": {
        "all_employees": { "count": 200, "index": 105.2 }
      }
    }
  ]
}

Field reference:

Field	Required	Description
`metadata.source`	Yes	Where the data comes from (for reference only)
`metadata.index_baseline`	Yes	The baseline value (e.g. `100` for index data, `0` for absolute salaries)
`metadata.index_label`	Yes	Label for the index column in output (e.g. `"Index"` or `"Monthly salary (DKK)"`)
`metadata.baseline_description`	No	Human-readable explanation of what the baseline represents
`companies[].company`	Yes	Company name as it appears in your source data
`companies[].city`	No	City/location — used for the `--city` filter when multiple entries share a name
`companies[].categories`	Yes	Named salary categories, each with `count` and/or `index` fields

The category keys (all_employees, engineering, etc.) are free-form — use whatever labels match your data source. The display formatter renders them as title-cased strings (All Employees, Engineering).

Setup Options

Manual
Excel
Research template

Create salary_data.json by hand in the repo root with data from any source — union statistics, Glassdoor, salary surveys, networking conversations, or personal research.

# Create the file in the repo root
touch salary_data.json

Then paste in the JSON structure shown above and populate the companies array with the employers you want to track. You can start with just a handful of companies and add more incrementally.

If your salary data comes from an Excel file (common for union statistics exports), use the included converter:

pip install openpyxl
python tools/convert_salary_excel.py path/to/salary-data.xlsx \
  --source "My Salary Data 2025" \
  --baseline 100 \
  --baseline-desc "Index 100 = median salary"

The converter auto-detects the Excel layout — it looks for a “Company” or “Firma” column and an optional “City” or “By” column, then treats the remaining columns as salary data, automatically pairing count and index columns where it finds them.The output is written to salary_data.json in the repo root, ready to use immediately.

To build your dataset incrementally from personal research, start with this empty template and add companies as you encounter them during your job search:

{
  "metadata": {
    "source": "Personal research",
    "index_baseline": 0,
    "index_label": "Monthly salary (DKK)",
    "baseline_description": "Approximate monthly salary before tax"
  },
  "companies": [
    {
      "company": "Example Corp",
      "city": "Copenhagen",
      "categories": {
        "entry_level": { "index": 42000 },
        "senior": { "index": 55000 }
      }
    }
  ]
}

This format uses absolute salary values rather than an index. Set index_baseline to 0 (or any non-comparative value) and use index_label to describe the unit. Add one company object per employer you research — the fuzzy matcher will find them during /apply even if the exact name in the posting differs slightly from what you typed.

Using the Tool Directly

You can run salary_lookup.py outside of /apply to look up a company manually:

# Basic lookup by company name
python salary_lookup.py "Novo Nordisk"

# Filter results by city when multiple entries share a name
python salary_lookup.py "Ørsted" --city "Fredericia"

# Output as JSON (for scripting or piping)
python salary_lookup.py "COWI" --json

# List all companies in the dataset
python salary_lookup.py --list-all

The tool prints a formatted table showing each matched company, its location, the salary categories available, and the index value compared to the baseline. The vs Baseline column shows the percentage difference from index_baseline — a positive value means above the baseline, negative means below. If no match is found, the tool suggests trying a shorter or different name and notes that entries in the dataset may include legal suffixes.

salary_data.json is listed in .gitignore by default and will not be committed to your fork. Salary data sourced from union statistics or employer agreements is often proprietary or confidential, and keeping it out of version control avoids accidental exposure if you ever make your fork public.If your data is not sensitive and you want to track it in git, remove the salary_data.json entry from .gitignore.

If salary_data.json does not exist when /apply runs, the salary benchmark step is silently skipped — no error, no interruption to the workflow. You will see a note in the fit evaluation output indicating that salary data was not available for the target company. This means you can set up the rest of the framework and come back to salary benchmarking later without any friction.

Get Started

Core Workflow

Extended Commands

Customization

Reference

Configuring Salary Benchmarking for the /apply Workflow

How It Works

salary_data.json Format

Setup Options

Using the Tool Directly

Build docs developers (and LLMs) love

Get Started

Core Workflow

Extended Commands

Customization

Reference

Documentation Index

​How It Works

​salary_data.json Format

​Setup Options

​Using the Tool Directly

Build docs developers (and LLMs) love

How It Works

salary_data.json Format

Setup Options

Using the Tool Directly