Phase 0 · Foundation Assessment · BEA-29 Marketplace Observatory
The permanent base layer for BEA-29. Scoped to PDF Phase 3 (Commerce & Marketplace Intelligence) shipped in a single cut. Sister pillars (organic search, AI visibility) are owned by BEA-04 and BEA-11; this dossier deliberately does not re-litigate those.
1. Success criteria (measurable, time-bound)
These are written as decision-grounded, not impression-grounded. The signal that matters is whether a commerce operator changes a pricing, channel, or enforcement decision because of what this dashboard surfaces.
- [ ] MAP-enforcement decisions: Within 8 weeks of pilot, at least 3 documented enforcement actions (cease-and-desist, listing takedown, distributor recall) cite a BEA-29 MAP-violation flag as the trigger. The lens already surfaces 36 violations on day one · the test is whether the operator acts.
- [ ] Buy-box response time: Median time from owner→3P buy-box flip to operator acknowledgement drops below 4 hours. Today the loop is "someone notices in a meeting"; the target is "Slack alert from this dashboard."
- [ ] Channel-price drift caught: At least 5 instances per quarter where the price grid catches an unauthorized discount on Amazon/Walmart/Target before Beats finance reconciles the channel-margin report.
- [ ] Shelf-rank visibility: Beats top-3 share on Amazon for the 27 seeded category keywords is tracked weekly; any 2-week sustained drop ≥10pp on a priority keyword triggers a merch-team review.
- [ ] Honest data hygiene: DataForSEO spend stays under the $100/mo hard cap. Cost-cents column is honest (validated after the 6/23 calibration fix). MAP-violation count is reproducible day-over-day on stable ASINs.
2. Decision-maker profile (Three Ledgers)
- Public ledger: "A real-time view of Beats commerce across Amazon, Apple, Best Buy, Target, Walmart, Beats Direct, and Costco · price, stock, buy-box ownership, third-party sellers, and Amazon shelf rank. Replaces three weekly spreadsheets and a 9pm Sunday email."
- Shadow ledger: Patrick O'Brien (and whichever commerce-ops lead owns enforcement) gets measured on margin protection and brand integrity. The MAP-violation signal is the highest-leverage finding: every below-MAP listing is either grey-market or counterfeit, both of which compress brand pricing power. The dashboard's quiet job is to make those visible without the operator having to ask.
- True ledger: A thin star-schema warehouse with three live-call ingest scripts, four read-side lenses, and a $100/mo DFS budget. Standard+webhook ingest is reserved for v1.1 when volumes outgrow the live cap. Non-Amazon channel coverage in v1 is degraded (price grid shows Amazon-only); Apple / Best Buy / Target / Walmart need scrapers or licensed feeds · flagged as v1.1 work, not v1.
- Owner / sponsor: Patrick O'Brien. Same sponsor as BEA-04 and BEA-11 · the cross-app sibling switcher in the sidebar is a deliberate UX bet that the same operator works across all three.
3. Constraint inventory
| Constraint | Type | Notes |
|---|---|---|
DataForSEO ToS allows Merchant Amazon endpoints; clearance recorded in data-fabric/src/clearance/sources.ts as dataforseo-amazon cleared by Patrick |
hard | Same gate as BEA-04's Labs clearance. App refuses to boot the adapter without the clearance set. |
| No external data egress of Apple-adjacent commerce data to third-party SaaS | hard | Mirrors BEA-04's posture. Forces in-house warehouse (Supabase per-project) over a Klue/Crayon equivalent. |
MAP price source is dim_product.map_price_usd, seeded manually |
soft (will harden) | Needs a quarterly review ritual; if Beats raises MAP, the dashboard's violation flag silently inverts. Documented in the runbook (v1.1). |
| Live merchant endpoints have a $100/mo hard cap; sweeps sized accordingly | hard | First sweep cost $0.19 (post-calibration). Headroom is wide; the cap is a circuit breaker not a budget. |
| Non-Amazon channel coverage (Apple/Best Buy/Target/Walmart) requires retailer-specific adapters | soft | Out-of-scope for v1; v1 ships Amazon-rich + other channels stub. The price grid UI is designed to gracefully degrade. |
| Single Supabase project per BEA (per-project isolation, NOT shared) | hard | Convention across BEA-02/04/11; beats-commerce (ref odqrjlhpodmqwqpmmoff). Each app stays independently runnable. |
4. Scope boundary
- In scope: Amazon-first commerce intelligence · 7 priority Beats ASINs, 11 competitor flagships, 27 Amazon-search keywords. Four lenses (Price Grid · Buy-Box · Seller Landscape · Shelf Rank). One shared adapter (
@b-combinator/competitive-data-fabricsourcedataforseo-amazon) reusable from any sibling BEA. - Out of scope (v1): Non-Amazon retailer coverage beyond a placeholder column in the price grid. Standard+webhook async ingest. Slack/email alerting (the dashboard is read-only; the operator pulls). Per-region (CA/UK/JP) marketplaces. Returns and rating signals (BEA-02 owns that surface).
- Recursion budget acknowledged: max 3 per phase. None used in v1.
5. Assumption log
| # | Assumption | Confidence | Validated? |
|---|---|---|---|
| 1 | DataForSEO Merchant endpoints return enough seller detail to compute MAP violations reliably | 0.80 | Yes · first sweep surfaced 36 violations across 7 ASINs with explicit per-seller prices |
| 2 | Beats ASINs in the strategic-dashboard PDF would resolve as-is | 0.60 | No · 4 of 7 needed corrections (Studio Pro / Solo 4 / Powerbeats Pro 2 / Studio Buds+). Documented in learning log |
| 3 | map_price_usd from the PDF reflects current MAP policy |
0.70 | Partial · assumed correct at seed time; needs quarterly review |
| 4 | Live mode stays under the $100/mo cap at full cadence (daily listings + weekly sellers + weekly rank for the priority set) | 0.85 | Trending · first sweep was $0.19 |
| 5 | A single operator (Patrick) can act on all three sibling dashboards (BEA-02 / 04 / 11 / 29) without drowning | 0.65 | TBD · the sibling switcher in the sidebar is the bet |
Gate: success criteria defined, constraints mapped, decision-maker identified → gatesPassed: ["0-foundation"] in manifest.json.