Agentify 3000: [[2026-10-04-starter-story-pat-walls-assessment|Starter Story]] index (phase 1)
Founder ask (2026-10-04): which businesses are agentification opportunities, and which are content foils. Phase 1 = dataset only.
Data: [[data/starter-story-index.csv]] (+ .json twin). Scripts: data/scripts/. Notes: [[IMPLEMENTATION-NOTES]].
Phase 2 (2026-10-04, scoring): [[data/businesses-scored.csv]] (2,566 deduped businesses, archetype + 5 job scores + agent-carryable share + moat, .json twin) · [[archetypes]] (scheme and per-archetype table) · [[opportunities]] (top 20 + 5 [[DESIGN-scribble-works|Scribble Works]] adjacent) · [[foils]] (30 named foils + 5 [[DESIGN-denominator|Denominator]] post angles). Phase 2 scripts in data/scripts/phase2/. Suspect-revenue total is now 66 (phase-2 rule added).
What was collected and how
- Sitemap → 3,299 unique
/stories/<slug>URLs. Each page was fetched once, one at a time, at about 1.2 s per request with a browser user agent. Total time was 68 minutes. Every request returned 200. There were no 403 or 429 responses and no retries. - The parser reads only the logged-out header widget: name, website, location, start date, revenue/mo, founders and employees, plus
ss_category, JSON-LD headline/author/date/description, and the public video summary and Stack list. The idea widget fields (startup cost, model, channels, tools) are idea-level aggregates, not facts about the specific business. - When the header has no revenue, the parser reads it from the headline (annual figures divided by 12).
revenue_sourcerecords which.
Coverage
| Metric | Value |
|---|---|
| Story URLs in sitemap | 3,299 |
| Fetched (HTTP 200) | 3,299 |
| Parsed business rows | 3,261 (38 non-business pages excluded: /data reports, idea listicles) |
| Distinct websites | 2,442 (many businesses have an original story plus follow-ups) |
| Rows with monthly revenue | 85.8% (2,681 from header, 118 from headline) |
| Revenue flagged suspect | 65 |
Content types: story 1,577 · follow_up 1,115 · starter 409 · video 112 · breakdown 48.
Field fill rates: name 96.0 · description 100 · category 93.7 · revenue 85.8 · team_size 95.0 · location 91.7 · year_started 91.8 · website 96.0 · founder 96.4 · startup_cost 22.4 · model 22.0 · tools 22.6 · channels 22.4 (all %). The last four come from the idea widget, which only about a fifth of stories carry, and they are idea-level values.
Revenue distribution (USD/month, n=2,799)
| p10 | p25 | median | p75 | p90 | p99 | max |
|---|---|---|---|---|---|---|
| $3K | $10K | $40K | $194K | $533K | $5.0M | $71.6M (suspect) |
The 65 suspect values are excluded from the clean figures (n=2,734): median $40K, p75 $180K, p90 $500K.
Buckets: <$1K 96 · $1K-10K 532 · $10K-50K 832 · $50K-100K 329 · $100K-1M 815 · ≥$1M 195.
Top 25 categories (Starter Story ss_category, 50 distinct)
other 489 · marketing 282 · (none) 204 · saas 176 · fashion 170 · home 133 · health 130 · food 120 · productivity 105 · beauty 104 · technology 94 · content 91 · finance 87 · coaching 71 · design 67 · travel 60 · accessories 60 · writing 54 · ecommerce 54 · education 50 · games 49 · seo 47 · outdoor 44 · art 43 · gifts 42
Known gaps
- Revenue is self-reported and not verified. Header revenue is current and can differ from the headline (Pallyy: $2K in the headline, $90K in the header).
revenue_suspect= below $100/mo, at least $1M/mo with 5 or fewer people, or more than 10x off the headline. - Startup cost, model, tools and channels are idea-level and only about 22% filled. Business-specific tools are gated, so they were not collected.
- "other" + none = 21% of rows, so the category field alone is weak. Phase 2 needs a better taxonomy (from description or idea).
- Breakdowns and some videos have no business name (about 4% of rows lack one).
- Follow-ups duplicate businesses: dedupe on
websitebefore counting businesses. - The page does not show an as-of date for revenue.