FranchiseDepth/franchise-database-sample

FranchiseDepth Franchise Database: free sample

Fees, Item 19 earnings, closure rates, growth, outlets by state and agreement terms for 25 US franchise brands, parsed from each brand's Franchise Disclosure Document (FDD) as filed with a US state. It is in exactly the same format as the full database, so you can test your queries, notebooks or AI workflows before scaling up.

Full database: 2,485 brands, with 13,225 Item 19 figures, closure rates for 1,865 brands and outlet growth for 2,265. Available at franchisedepth.gumroad.com, along with single-industry packs (food, home services, fitness).

Also on GitHub and Kaggle. Compiled 2026-11-01. FDD year of the latest filing per brand: 15 from 2026, 7 from 2025, 3 from 2024.

Quick start

Each table is browsable in the viewer above (pick it from the subset menu). To query everything at once, download the repo and use the SQLite file:

import sqlite3
import pandas as pd

con = sqlite3.connect("franchisedepth.sqlite")
pd.read_sql("""
    SELECT brand_name, total_investment_low_usd, closure_rate_pct, growth_pct
    FROM v_brand_summary
    ORDER BY closure_rate_pct
""", con)

More questions with their SQL are in example_queries.sql. No code? Open franchisedepth.xlsx or brand_summary.csv in Excel or Google Sheets, or start with README.txt.

Tables

Table	One row per	Rows
brands	brand: identity, FDD year, source registry, hand-checked flag	25
fees	brand: Item 5 initial fee, Item 6 royalty and marketing fund, Item 7 total investment	25
fee_notes	brand: tiers, minimums and discounts as printed	25
item19_figures	Item 19 figure: metric, cohort, outlet count, average, median, min, max, period, FDD page	388
comparable_revenue	brand: one annual revenue figure picked by a fixed rule, for comparing brands	25
outlets_by_year	brand × year × outlet type (Item 20 Table 1)	219
turnover_by_year	brand × year: opened, terminated, non-renewed, reacquired, ceased, transferred (Table 3)	75
closure_by_year	brand × year: closure rate on the method below	75
outlets_by_state	brand × year × state	2,366
agreement_terms	brand: Item 17 term, renewal, transfer fee, non-compete	11
brand_metrics	brand: total outlets, closure rate, outlet growth, all-in fee	25
alternatives	brand × similar brand	250
peer_ranks	brand × measure: rank against peers	171
value_ratios	brand: revenue relative to investment and to royalty	25

table_docs and column_docs inside the database describe every table and column. AGENTS.md has the same in prose, with units and what NULL means.

Files

• franchisedepth.sqlite: every table, four ready-made views and the column docs.
• csv/, parquet/: each table as a file.
• franchisedepth.xlsx: every table as a tab, plus a Compare tab (pick up to 5 brands).
• brand_summary.csv: one row per brand with the key figures.
• brands.jsonl, profiles/, profiles_by_industry/, all_profiles.md: one document per brand, for retrieval pipelines and AI uploads.
• AGENTS.md, AI_GUIDE.md, AI_INSTRUCTIONS.txt: guides for analysts and AI assistants (ChatGPT, Claude, Gemini, Grok, Copilot, Perplexity).
• mcp/franchisedepth_mcp.py: a read-only MCP server for Claude Desktop, Cursor and other MCP clients.

Methods

• Closure rate = franchised terminations, non-renewals and ceased operations ÷ franchised outlets at the start of each year, over the latest three years of Item 20. Where a cause is blank, it is filled from the table's start, opened and end counts. Transfers and reacquisitions are not closures.
• Outlet growth = change in total outlets (franchised + company-owned) between the first and last year-end of the latest three years of Item 20.
• All-in fee = royalty % + marketing-fund % of gross sales (Item 6).
• NULL means not disclosed, never zero. Revenue is not profit. Figures describe the FDD year shown for each brand, not the present.

Connect it to Claude Desktop (MCP)

1. Install uv (https://docs.astral.sh/uv/).
2. Add this to Claude Desktop's config (Settings > Developer > Edit config), with your own paths:

{"mcpServers": {"franchisedepth": {"command": "uv", "args": ["run", "--with", "mcp", "python", "C:/path/to/mcp/franchisedepth_mcp.py", "C:/path/to/franchisedepth.sqlite"]}}}

3. Restart Claude Desktop and ask, for example: "Which brands here have the lowest closure rate, and what does each cost to open?"

The server opens the database read-only and refuses anything but SELECT queries.

Licence

CC BY-NC 4.0: free to share and adapt for non-commercial use with credit to FranchiseDepth (franchisedepth.com). Commercial use needs a paid licence; see LICENSE.txt. Information only, not financial or legal advice; verify figures against the FDD before relying on them.