One-stop inventory of every pipeline the Territori dashboard depends on, grouped by the panel it feeds. Each entry names the endpoint or vendor, cost, refresh cadence, owner, and status. This page is for internal use only — it is not linked from the product surface.
Authoritative parcel boundaries, acreage, ownership, and zoning attributes for Clallam County (Port Angeles, Sequim, Forks, Joyce, Neah Bay). Accessed via ArcGIS REST FeatureServer — paginated GeoJSON / JSON pulls. Core source for the parcel overlay on the dashboard map and the acreage band filter.
Jefferson County (Port Townsend, Quilcene, Brinnon, Hoh) parcel data. Needed because the 60-minute-drive catchment from the park extends into Jefferson. Format parity with Clallam is good — same schema family. Combined with Clallam to form the unified "ONP parcel layer."
Southern Olympic Peninsula counties. Smaller share of the 60-min catchment but required for completeness on the Hood Canal and coastal corridors south of Aberdeen. De-prioritized for MVP — stub for phase-2.
Zoning classifications joined to parcels to filter out industrial, institutional, and shoreline-restricted parcels. Also flags parcels in WDFW-designated critical areas and tribal land. Drives the "Constraints" popover in the refine bar.
Open-source routing on OSM data. We stand up a single t4g.small node against a Pacific-Northwest OSM extract, compute 15/30/60-minute isochrones from the park gateway, and cache GeoJSON polygons. Zero per-call cost. Janice's parcel catchment is one market — pre-computing the three rings once covers all users in ONP.
Kept as a cross-check on OSRM output and a paid fallback once we need ferry-aware routing or traffic-adjusted isochrones. Not in MVP spend.
Forward + reverse geocoding for operator addresses, market anchor points, and user-entered parcel addresses during onboarding. Rate-limited public instance is fine at MVP traffic; we self-host if we cross 1 req/sec.
Murray Cox's open monthly Airbnb scrape — listing-level attributes, location, amenities, review counts, calendar snapshots. Covers Port Angeles / Forks / Sequim / Port Townsend. Adequate for the 412 STR count, moat coverage, and keyword-overlap scoring at MVP fidelity. ADR is inferred from "price" field (nightly list), not realized — good enough for positioning, we upgrade when we need booked-ADR truth.
Holds realized ADR, occupancy, and RevPAR at daily granularity. We upgrade here once a paying customer requires booked-rate fidelity (most likely on a Defending brief). Not in MVP spend.
Hotel inventory, demographics, and spend enrichment at the tract level. Gives us the 38-hotel count without Amadeus spend, plus overnight-visitor demographic overlays we'll need for the Narrative layer. Free trial for MVP; evaluate paid tier when we expand beyond ONP.
NPS Statistics portal gives monthly visitation to ONP (trailing 10 years) by entry station. data.gov backfills census, commuting, and tourism-economic overlays. Feeds demand-side context for the dashboard and for the narrative in the Editorial Diagnosis.
Light scrape to fill gaps for lodges and boutique operators not in Inside Airbnb or ESRI. Operator allow-list, capped request rate, respects robots.txt. Only public-facing listing pages, no auth-gated content.
STR-level realized ADR, occupancy, and RevPAR over trailing 13 months. Segment cuts: coastal / rainforest / Port-Angeles-core / trailhead-adjacent. Also feeds price-dispersion calculation (inverse contributor to substitutability).
Hotel segment ADR and occupancy benchmarks at the DMA and submarket level. Triangulation partner for Amadeus-derived hotel pricing.
Rising and seasonal hospitality queries scoped to the Olympic Peninsula — "adaptive hike olympic," "stargazing lodge olympic," "rainforest cabin hoh," etc. Google Trends gives trajectory; DataForSEO Labs gives volume + related-keyword fan-out at ~$0.07 per query. 214 tracked seed queries, refreshed weekly.
Deeper backlink + SERP feature data. We upgrade when the first paying Defending customer wants the backlink map for their moat category. Not in MVP spend.
For Defending / Repositioning mode users who connect their own GSC — real query data for their property and peer set. Never used as a primary source; enhances the personalized view only.
Runs the 50 seeded queries against ChatGPT weekly. Captures whether the answer names an operator, which operators, in what position, and with what framing. Feeds AEO operator-mention-share.
Same 50 queries, run against Claude. Compared head-to-head with ChatGPT to detect engine-specific moats (e.g. operators that rank in Claude but not ChatGPT).
Perplexity Sonar probes — valuable because Perplexity cites sources, so we get the actual URL cascade that led to the answer (operator site, OTA listing, press mention, AllTrails review).
No official API. We probe via headless Chromium on the 50 queries, capture the AI-Overview block HTML, and parse cited domains. Rate-limited, flakier than the other three engines — kept as a secondary signal.
Recent STR-relevant permits (new dwellings, ADUs, cabins) across the five jurisdictions that fall inside the 60-min ONP drive catchment. Used to flag in-progress supply that hasn't hit listings yet, to estimate near-term supply pipeline, and to trigger the "new competitive permit in your band" alert — the #1 live-dashboard signal Janice called out. Each portal publishes a searchable log; we scrape weekly and diff against the prior week.
Registered STR operators at the county level. Triangulates AirDNA coverage and surfaces operators that are newly permitted but not yet visible on OTAs.
Current municipal STR caps, owner-occupancy requirements, and pending moratoria. Stored as structured policy records and surfaced on whitespace prescriptions where the play depends on STR licensing.
Jamestown S'Klallam, Lower Elwha Klallam, Makah, Hoh, and Quileute lands. Flagged as non-developable on the parcel overlay and referenced in the Multi-gen + Indigenous-story whitespace play (which requires tribal partnership framework).
Guest reviews from Airbnb, Vrbo, Booking.com passthrough. Classified into 12 review axes (service, cleanliness, location, value, design, food, wilderness access, accessibility, family fit, pet experience, noise, bed quality). Drives the review-axis-overlap score.
Covers lodges and tour operators not well-served by AirDNA/OTAs — gear rentals, guided services, ranger-led programs. Needed for the M15 Gear/Guided moat coverage.
Qualitative signal for trail-access claims (M01) and shoulder/winter usability (M14). Hand-curated pulls, not API-driven at MVP.
legal@territori.ai. Personally identifiable operator data (owner names that aren't in public parcel records) is not permitted on the named-unclaimed or named-weakness surfaces without a contractual basis.