Market Analysis and Research Portfolio Example
How the decision to let clawdbot go to OpenAI hurt Anthropic’s revenue
Layer 1 — Social Media Sentiment
| Element | Phase | What Was Covered | Gap Remaining |
|---|---|---|---|
| Hacker News thread volume | Phase 2 | 245+ points documented | Full thread taxonomy, comment sentiment, longevity not analyzed |
| GitHub star velocity (OpenCode) | Phases 1–3 | 18K stars in 2 weeks, peak single-day 2,087 on Jan 12 | Sentiment in issue threads not analyzed |
| OpenCode GitHub issue #6930 | Phase 2 | 147+ reactions noted | Full comment corpus not analyzed |
| X: OpenCode creator “Are you not entertained?” post | Phase 3 | Documented as competitive theater | Reach, impressions, RT/reply ratio not quantified |
| X: Rob Zolkos developer post | Phase 2 | Cited as incident surfacing event | No follower count, reach, or amplification analysis |
| Mid-task disruption developer quotes | Phase 2 | Two direct quotes from affected users | No volume estimate of how many similar posts existed |
Coverage depth: ~30% of Layer 1
Layer 2 — Press Coverage
| Element | Phase | What Was Covered | Gap Remaining |
|---|---|---|---|
| VentureBeat, WinBuzzer, Sherwood News coverage | Phase 2 | Named as outlets that covered incident | No headline sentiment analysis, article framing, or longevity |
| RSP rollback coverage | Phase 4 | Documented as macro-context event | No press corpus analysis on how it re-anchored to Jan enforcement |
Coverage depth: ~15% of Layer 2 — almost entirely uncovered
Layer 3 — Brand Equity
| Element | Phase | What Was Covered | Gap Remaining |
|---|---|---|---|
| Trust deficit as structural concept | Phases 2 + 4 | Characterized qualitatively | No SOV data, NPS proxy, or survey data analyzed |
| “Safe AI” brand erosion via RSP rollback | Phase 4 | Identified as self-negating benefit | No brand equity quantification attempted |
| GitHub Copilot deprecation as distribution loss | Phases 2–3 | Documented event + 1.1–1.2M user estimate | No SOV impact, no developer survey signal |
Coverage depth: ~20% of Layer 3 — quantification entirely absent
Layer 4 — Narrative Framing
| Element | Phase | What Was Covered | Gap Remaining |
|---|---|---|---|
| iOS/Android “walled garden” framing | Phase 3 | Named and contextualized (Android 70% market share) | No frame persistence analysis, no propagation mapping |
| “Safety theater” frame | Phase 4 | Identified implicitly via enforcement → RSP rollback sequence | Not named as a frame, not analyzed for spread |
| “Bait and switch” / mid-task disruption | Phase 2 | Documented structurally as mental model violation | Not analyzed as a narrative frame with propagation |
| Escalation pattern (trust erosion across 3 events) | Phase 2 | Timeline documented | Not framed as a compounding narrative arc |
Coverage depth: ~25% of Layer 4 — framework identified but zero frame analysis executed
Layer 5 — Competitive Marketing Capture
| Element | Phase | What Was Covered | Gap Remaining |
|---|---|---|---|
| OpenAI “open alliance” positioning | Phases 3–4 | Documented + named tools converted | No messaging analysis, no landing page/blog post corpus |
| Codex growth as competitive displacement | Phase 3 | WAU tripling, 1.6M WAU, Mac app 1M downloads | No marketing attribution analysis |
| OpenAI extending Codex to OpenHands, RooCode, Pi | Phase 3 | Documented | No competitive messaging content analysis |
| Cursor / GitHub Copilot as Anthropic-authorized alternatives | Phase 4 | Named as channel control instruments | No competitive positioning or messaging comparison |
Coverage depth: ~35% of Layer 5 — events documented but zero marketing/messaging analysis executed
Summary: What’s Been Done vs. What Remains
| Layer | Coverage Depth | Primary Gap |
|---|---|---|
| Social sentiment | ~30% | Volume quantification, reach/impression mapping, platform corpus beyond HN/GitHub |
| Press coverage | ~15% | Almost entirely unexecuted — full corpus needed |
| Brand equity | ~20% | No quantitative brand metrics at all |
| Narrative framing | ~25% | Frames named but no propagation/persistence analysis |
| Competitive capture | ~35% | Events documented; messaging/content analysis absent |
Net assessment: Phases 1–5 generated the evidence substrate for the marketing analysis but executed almost none of the marketing analysis itself. The overlap is real but shallow — what exists from prior phases is input data, not marketing conclusions.
Ready to execute Layer by Layer? I’d recommend starting with Layer 4 — Narrative Framing first since the frames are the load-bearing interpretive structure everything else hangs on — same logic as why Phase 1 baseline anchored the ROI model.
Leave a Reply