Skip to main content
Glama
startup-gtm-skill

startup-gtm-skill

Startup Channel Atlas ๐Ÿ‡ฎ๐Ÿ‡ณ

How 350 of India's top startups (2010โ€“2025) actually acquired their users โ€” coded by marketing channel and growth stage, and queryable by an LLM.

Most "growth advice" is anecdote. This is a structured, evidence-derived dataset built to answer one question:

Which marketing channels are best โ€” for a given industry, audience, era, scale/stage, product type, and budget?

Each of the 350 companies is coded for the channels it used at three stages (0โ†’1, 1โ†’10, scale), what worked, what failed, and situational variables (sector, model, era, trust burden, virality, CAC ceiling, category creation-vs-capture). It deliberately includes ~80 acquired / distressed / shut-down companies so you can learn from failure, not just survivorship.

Headline finding: channels are stage-locked more than sector-locked. TV was the entry channel for ~0 of 350 companies; it โ€” like IPL and celebrity โ€” is a scale-stage play. 0โ†’1 is won on earned media, partnerships, SEO/content, product-led loops and community.


Quickstart โ€” pick your lane

๐Ÿง  Option A โ€” Ask any LLM (ChatGPT, Claude, Gemini)

The fastest path. Point your model at this repo and let SKILL.md be its brain.

  1. Open your AI chat.

  2. Paste the link to this repo (or paste the contents of SKILL.md).

  3. Say: Read โ€” the model adopts the channel-selection rules.

  4. Ask, e.g. "I'm a seed-stage D2C skincare brand for tier-2 women in 2024, ~โ‚น50L/month. Which channels first, and who proves it?"

SKILL.md explicitly tells the model not to dump the whole CSV into context โ€” it reasons over the rules and pulls only the rows it needs, so answers stay cheap and grounded.

โšก Option B โ€” Give it a real brain (MCP server, for Claude Desktop / Cursor / Claude Code)

Runs a local server that exposes the dataset as query tools. No hosting, no token bloat โ€” the model calls find_channels(...) and gets back a small, ranked, evidence-backed result.

Run with uv (recommended):

uvx --from git+https://github.com/lan-club-live/startup-gtm-skill startup-gtm-skill-mcp

Add to Claude Desktop (claude_desktop_config.json โ†’ Settings โ–ธ Developer โ–ธ Edit Config):

{
  "mcpServers": {
    "startup-gtm-skill": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/lan-club-live/startup-gtm-skill", "startup-gtm-skill-mcp"]
    }
  }
}

Or from a clone:

git clone https://github.com/lan-club-live/startup-gtm-skill && cd startup-gtm-skill
uv run startup-gtm-skill-mcp          # or:  pip install -e . && startup-gtm-skill-mcp

Then just ask Claude/Cursor a channel question โ€” it will call the tools automatically.

Tools exposed: find_channels, get_company, search_evidence, channel_lens, list_filters, dataset_summary.

๐Ÿ“Š Option C โ€” Just the data (Excel / Pandas / your own build)

import pandas as pd
df = pd.read_csv("data/startups.csv")
d2c = df[df["Sector"].str.contains("d2c", case=False, na=False)]
print(d2c[["Company", "0โ†’1 Channels", "Scale Channels", "What Worked Most"]])

Or open data/startups.csv in Excel/Sheets and pivot away.


Related MCP server: Meta Ads MCP Server

What's in the box

startup-gtm-skill/
โ”œโ”€โ”€ data/
โ”‚   โ”œโ”€โ”€ startups.csv          # 350 companies ร— channels-by-stage + variables + what worked/failed
โ”‚   โ”œโ”€โ”€ channel_matrix.csv    # company ร— 24 canonical channels (E=0โ†’1, G=1โ†’10, S=scale)
โ”‚   โ””โ”€โ”€ economics.csv         # 257 sourced CAC / ROAS / ad-spend datapoints (nothing estimated)
โ”œโ”€โ”€ SKILL.md                  # the lightweight "brain": rules + decision trees + token discipline
โ”œโ”€โ”€ src/startup_gtm_skill/
โ”‚   โ”œโ”€โ”€ query.py              # pure, dependency-free query logic (importable, testable)
โ”‚   โ””โ”€โ”€ server.py             # stdio MCP server wrapping query.py
โ”œโ”€โ”€ pyproject.toml            # packaged for `uvx` / `pip install`
โ””โ”€โ”€ README.md

Dataset columns (startups.csv)

Rank, Company, Sector, Subsector, Model, Status, Era Scaled, Founded, Valuation ($mn), Funding ($mn), 0โ†’1 Channels, 1โ†’10 Channels, Scale Channels, First 1000 Users, Budget Band, Economics Datapoints, Paid vs Organic, Signature Moves, Sponsorships, Regulatory Events, Retention Channels, Trust Burden, Virality, Purchase Freq, CAC Class, Category Play, Core Audience, Geo Tier, What Worked Most, What Failed/Wasted, Confidence, Template

Methodology & honest limits

  • Derived, not disclosed. Channel mixes are inferred from public evidence โ€” press, DRHPs/filings, founder interviews, agency case studies, search footprints. Companies rarely publish their channel mix; this reconstructs it.

  • Confidence-tagged. Every row is high/medium/low. Evidence density thins past rank ~250 โ€” use the tag rather than trusting all rows equally.

  • Quality loop. After the first 100 companies, an evaluation answered CMO-style questions from the data and adversarially judged them; the schema was then upgraded (calendar years, channel side, sourced economics) and flagged profiles were re-researched. See SKILL.md.

  • Scope. Top 350 Indian startups by valuation, 2010โ€“2025. India-centric; not a guide to other geographies or to today's channel costs without adjustment.

Contributing

Corrections and additions welcome โ€” this is a living dataset. Open an issue or PR with a company row plus sources. Keep the confidence tag honest.

License

Data: CC-BY-4.0 ยท Code: MIT. Attribution appreciated โ€” link back to this repo.

Available Tools

6 tools
channel_lensB

Inverse view for one channel: how common it is, its stage split, which sectors use it most, and the companies behind it. Accepts a partial channel name.

Inputs: channel (required).

dataset_summaryA

Orientation call: how many companies, split by status/era/confidence/sector. Cheap. Use this first to understand what the dataset covers before querying.

Inputs: no named parameters.

find_channelsA

PRIMARY TOOL. Given a marketing situation, return the channels comparable Indian startups actually used โ€” ranked by prevalence and split across growth stages (0โ†’1, 1โ†’10, scale), with named example companies as evidence.

Pass any subset of: sector (a group like 'Fintech' or a raw sector), business_model (B2C/B2B/D2C/SaaS/marketplace/B2B2C), era ('2010-2016', '2016-2019', '2020-2021', '2022-2025'), status, trust_burden (low/medium/high), virality (low/medium/high), cac_ceiling (low/medium/high/enterprise), category_play (creation/capture). Optionally set stage to rank by that stage.

Always report the returned cohort_size and honor any 'warning' about small cohorts.

Inputs: era, top, stage, sector, +6 more.

get_companyA

Full channel profile for one company (fuzzy name match): channels by stage, what worked, what failed, economics, sponsorships, variables, confidence, sources.

Inputs: name (required).

list_filtersA

Return the valid values for every filter accepted by find_channels (sector groups, business models, eras, statuses, trust/virality/CAC/category levels, stages, and the 24 canonical channels). Call this if unsure what to pass.

Inputs: no named parameters.

search_evidenceA

Keyword search across the narrative evidence (what worked / what failed / signature moves / ignition). Use for 'when did X fail', 'who used Y for Z'. Returns short snippets with company names โ€” not full rows.

Inputs: query_text (required), limit.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelogโ€ฆ

    TDQS

    A4.1/5.0

    Scored across 6 tools

    Disambiguation5/5

    Each tool serves a distinct purpose: dataset_summary orients, list_filters enumerates valid inputs, find_channels is the primary search, channel_lens provides the inverse view, get_company gives deep company profiles, and search_evidence does keyword-level narrative lookup. There is no functional overlap.

    Naming Consistency4/5

    All names are lowercase snake_case, but the pattern is mixed: most tools start with a verb (list, find, get, search), while two are noun phrases (channel_lens, dataset_summary). This is a minor deviation that doesn't harm usability.

    Tool Count5/5

    Six tools is well-scoped for a specialized GTM research skill. Each tool earns its place and together they cover orientation, discovery, deep dive, and evidence search without redundancy.

    Completeness5/5

    The domain is read-only GTM channel analysis. The set covers the full workflow: understand the dataset, know valid filters, find channels, inspect a channel in reverse, get company detail, and search evidence. No critical operations are missing.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      MCP server for accessing India's Ministry of Statistics and Programme Implementation (MoSPI) data APIs. Enables natural language queries for economic, demographic, and social indicators via a 4-tool workflow.
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Python MCP server that exposes Meta Marketing API data, providing tools to manage ad accounts, campaigns, and analytics through natural language interfaces.
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Production-ready MCP server integrating Google Search Console, GA4, and PageSpeed Insights for SEO and analytics intelligence, enabling natural-language queries to Google analytics data.
      -
    • F
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that exposes a LangChain agent with short-term memory to analyze real e-commerce data through predefined SQL tools, enabling natural language queries on sales, customers, and logistics.
      -