Tam Sam Som Builder

Skill: tam-sam-som-builder

Purpose

Provide a realistic market size estimate that an indie developer can actually use for decisions. Most TAM estimates are useless because they use top-down analyst numbers designed to impress VCs, not to inform a solo builder deciding where to spend the next 6 months. This skill uses triangulated bottom-up methodology — multiple independent estimation approaches cross-checked against each other — anchored to real signals from market_insights/ trend data.

Input

• Idea slug (or market research slug for pre-idea deep dives)
• memory/ideas/<slug>/idea.md (business_model, buyer)
• Search volume figures from the web-search market_insights file (Approach A) — there is no separate keywords file
• memory/ideas/<slug>/competitors.json (competitor scale and pricing signals; B2B review counts and customer counts)
• memory/ideas/<slug>/pricing.json (target price — if available)
• memory/market_insights/<niche>-*-<YYYY>-<MM>.md (trend analysis files — use all available platform files for this niche)

Using Market Insights

Trend analysis files from memory/market_insights/ provide critical calibration data. Extract the following from each available file's YAML frontmatter and narrative:

If overall_verdict across platforms is "cold", flag the entire estimate as speculative and note that market demand is unvalidated.

Methodology

Lane selection

business_model = b2c or prosumer uses Approaches A–C and the B2C capture-rate table. b2b-smb and b2b2c use Approach D as primary, Approach C as the cross-check, and the B2B capture-rate rows. Record the lane in the output.

Approach A — Search Volume (primary when the web-search insight file reports volumes)

TAM = monthly_search_volume × 12 × intent_conversion_rate × annual_price

Where:

• monthly_search_volume = total across all relevant keywords reported in the web-search market_insights file (or researched directly; cite the source)
• intent_conversion_rate = % of searchers who have genuine purchase intent (see benchmarks below)
• annual_price = from pricing.json target WTP, annualized

Intent conversion benchmarks by search type:

Approach B — Community Size Proxy (primary when market_insights available)

When keyword data is weak but trend analysis reveals active communities, estimate from community engagement:

TAM = active_community_members × platform_multiplier × annual_price

Where:

• active_community_members = sum of engaged users across platforms (subreddit subscribers, TikTok hashtag creators, App Store review volume)
• platform_multiplier = ratio of total interested population to active community members (see below)

Platform multipliers (how many silent interested people per active community member):

Approach C — Competitor Revenue Proxy (supplementary)

Estimate from competitor data in competitors.json:

TAM = sum of estimated annual revenue across all mapped competitors × market_coverage_factor

Where:

• Estimate competitor revenue from: estimated_users × competitor_price × 12 × estimated_conversion_rate
• market_coverage_factor = 1.3–2.0× (competitors don't capture the full market). Use 1.3× for saturated markets, 2.0× for markets with few competitors.

Approach D — ICP Count × ACV (primary for the B2B lane)

TAM = icp_count × annual_contract_value
SAM = icp_count_after_filters × annual_contract_value
SOM_year_1 = SAM × penetration_year_1

Where:

• icp_count = number of organisations matching the buyer definition. Sources, in order of preference: LinkedIn Sales Navigator company filters (industry × headcount × geography), industry directories and association member counts, government business statistics (Census/NAICS, Companies House, Eurostat SBS), trade-press population estimates (cite the article and its method).
• annual_contract_value = primary tier price × 12 from pricing.json, plus documented expansion.
• Filters for SAM: geography and language, the qualifying behaviour (for example "has 3+ active clients", "uses n8n or Make"), and reachability through the channels in distribution.json.
• penetration_year_1 from the B2B rows of the capture-rate table below.

Always state the ICP count source and its date; a market size built on an ICP count with no source is a guess and must be flagged as such.

Triangulation

Run all available approaches and compare:

• If estimates agree within 2× → high confidence. Use the geometric mean.
• If estimates disagree by 2–5× → medium confidence. Use the most conservative estimate and note the range.
• If estimates disagree by > 5× → low confidence. Flag assumptions that cause the divergence.

Always report which approaches were used and their individual estimates in the output.

SAM Filtering

SAM narrows TAM to the segment actually reachable by the app. Apply these filters in order:

Geographic & Platform Filters

iOS market share by region (for iOS-only apps):

For Android-only or cross-platform, apply the inverse or use 100%.

Segment Filters

Beyond geography and platform, apply any filters that narrow the market to people who would actually consider this specific app:

• Niche focus (e.g., "fitness" → "climbing-specific fitness")
• Prerequisite behavior (e.g., "already tracks workouts" → subset of fitness app users)
• Tech-savviness requirement (if the app requires unusual setup, filter out casual users)

SAM should typically be 10–40% of TAM. If SAM > 50% of TAM, the filters are too loose. If SAM < 5% of TAM, the niche may be too narrow for viable economics.

SOM Estimation

SOM is what an indie developer can realistically capture. This is where most estimates go wrong — indie builders don't have the resources to capture meaningful market share in crowded categories.

SOM Capture Rate Benchmarks by Category

Use the lower end of the range when:

• market_saturation from competitors.json is "high"
• Founder is beginner tier (from user_profile.md)
• No distribution advantage identified

Use the upper end when:

• Founder has an existing audience or distribution edge
• Strong ASO opportunity or viral loop exists
• Market is growing fast (trend_velocity = "rising-fast")

SOM Calculation

SOM_year_1 = SAM × capture_rate_year_1
SOM_year_3 = SAM × capture_rate_year_3 × growth_multiplier

Growth Rate Adjustment (from market_insights)

Trend velocity from market_insights directly affects the year-3 projection:

If multiple platform files have different velocities, use the median velocity.

Reality Check Layer

Before finalizing, run these sanity checks:

Market Size Verdict Thresholds

Based on SOM year 1 (the number that actually matters for an indie developer deciding whether to build). The thresholds apply to both lanes; in the B2B lane SOM is expressed as year-1 ARR.

Process

1. Load all available inputs: idea.md, competitors.json, pricing.json, and all matching memory/market_insights/<niche>-*-<YYYY>-<MM>.md files.
2. Extract calibration data from market_insights (trend velocity, top signals, monetization evidence, overall verdict).
3. Run Approach A (search volume) if the web-search insight file reports search volumes.
4. Run Approach B (community size proxy) if market_insights contain community signals.
5. Run Approach C (competitor revenue proxy) if competitors.json has user/pricing data.
6. Triangulate: compare estimates, determine confidence, select final TAM.
7. Apply SAM filters (geography, platform, demographics, niche).
8. Estimate SOM using category-appropriate capture rate benchmark.
9. Apply growth multiplier from trend velocity.
10. Run reality checks. Adjust if any are triggered.
11. Determine market size verdict from SOM year 1 thresholds.
12. Write output.

Output

Write to memory/ideas/<slug>/market_size.json:

{
  "idea_slug": "",
  "estimated_at": "YYYY-MM-DD",
  "lane": "b2c | b2b",
  "methodology": "bottom-up | community-proxy | competitor-proxy | icp-count | triangulated",
  "estimation_approaches": [
    {
      "approach": "search-volume | community-proxy | competitor-proxy | icp-count",
      "tam_estimate": 0,
      "key_assumptions": []
    }
  ],
  "icp_count": { "value": null, "source": "", "as_of": "", "acv": null, "penetration_year_1_pct": null },
  "triangulation_confidence": "high | medium | low",
  "tam": {
    "value": 0,
    "currency": "USD",
    "period": "annual",
    "assumptions": []
  },
  "sam": {
    "value": 0,
    "filter_criteria": [],
    "sam_to_tam_ratio": 0
  },
  "som": {
    "year_1": 0,
    "year_3": 0,
    "capture_rate_year_1_pct": 0,
    "capture_rate_year_3_pct": 0,
    "growth_multiplier": 1.0,
    "growth_multiplier_source": ""
  },
  "market_insights_used": [],
  "trend_velocity_observed": "rising-fast | rising | stable | declining",
  "monetization_evidence_found": true,
  "reality_checks_triggered": [],
  "market_size_verdict": "large | medium | niche | micro-niche",
  "sources": [
    { "url": "https://", "title": "", "accessed": "YYYY-MM-DD", "used_for": "" }
  ]
}

icp_count is filled in the B2B lane and left null in the B2C lane. sources must include the ICP count source and every competitor or pricing page used in Approach C.

Notes

• When used in the market-deep-dive workflow, pricing.json may not exist yet. In that case, use the median competitive price from competitors.json or a category benchmark ($3–$7/mo for typical B2C subscription apps).
• When used in idea-validation (if the orchestrator includes it), the output feeds into idea-scoring's Monetization dimension. The market_size_verdict and SOM values are used alongside pricing and CAC data to assess overall monetization viability.
• Market_insights files have a stale_after date. If all available files are past their stale date, flag the estimates as potentially outdated and recommend re-running trend-analysis before making a build decision.