PlotScrub — GeoJSON scrubber for small EUDR importers

1. One-liner

Cleans messy supplier coordinates into a TRACES-ready deforestation due-diligence filing that won’t bounce at the border.

2. Trend signal — why now?

The EU Deforestation Regulation (EUDR) goes live for large and medium operators on 30 December 2026 — six months out. Any importer placing coffee, cocoa, wood, rubber, soy, palm, cattle, or their derived products (chocolate, furniture, tyres, paper) on the EU market must file a Due Diligence Statement (DDS) per consignment in TRACES NT, and that DDS must carry valid geolocation for every production plot: a GeoJSON polygon for plots over 4 ha, a point for smaller ones, in the right Coordinate Reference System, with closed and non-self-intersecting geometry.

The documented failure mode is consistent across every vendor write-up: bad geolocation is the single biggest reason DDS submissions fail TRACES validation. Self-intersecting polygons, points submitted where polygons are required, wrong/missing CRS metadata, corrupted GeoJSON (one bad character makes the whole file unreadable), legacy hand-drawn farm maps that don’t survive conversion, and missing plot IDs that break batch traceability. One bad polygon bounces the filing — and goods don’t clear customs without an accepted DDS.

The broad derived-product scope swept in thousands of small businesses who didn’t know they were covered: a furniture importer, a specialty chocolate maker, a paper buyer. They have no GIS team and no €12K/yr compliance budget. The incumbent tools (osapiens, TraceX, IntegrityNext, Coolset) are enterprise supply-chain-mapping suites built for large operators — Coolset’s pricing starts around €12,000/yr. The EUDR due-diligence software market is projected to grow from ~$1.1B (2026) to ~$1.8B (2036). Money is moving; the long tail is unserved.

Provenance:

• Signal 1 (Demand): Geolocation is documented as the #1 reason DDS submissions fail TRACES validation; “10 most common DDS filing errors” content exists from multiple vendors — https://tracextech.com/dds-filing-errors/ — 2026-02
• Signal 2 (Feasibility): GeoJSON geometry validation + TRACES schema rules are deterministic and public; AI can parse messy supplier inputs (PDF maps, KML, lat/long lists) into clean polygons — https://tracextech.com/geojson-data-capture-validation-eudr-compliance/ — 2026-02
• Signal 3 (Economic): EUDR due-diligence software market ~$1.1B (2026) → ~$1.8B (2036); funded incumbents target large operators at €12K+/yr, leaving SMBs unserved — https://www.coolset.com/academy/best-6-eudr-compliance-tools-for-2026-supply-chain-due-diligence — 2026-03 Category: Regulatory arbitrage

3. The opportunity

Every EUDR vendor sells the same thing: an enterprise platform to map your entire supply chain, score deforestation risk, and orchestrate due diligence across thousands of SKUs. That’s the right product for Nestlé. It’s wildly over-built and over-priced for a 20-person coffee roaster importing from 40 cooperatives.

The narrow, painful, repeated job those incumbents bury inside a €12K suite is this: take the geolocation data a supplier actually sent you — which is always a mess — and turn it into a GeoJSON that TRACES will accept on the first try. Suppliers send coordinates as Excel columns, WhatsApp screenshots of a map pin, KML exports from Qfield, hand-typed lat/long pairs, or a PDF with a plot sketch. Someone at the importer has to wrangle all of that into a single valid GeoJSON FeatureCollection with correct CRS, closed rings, no self-intersections, plot IDs that match the batch, and the right HS code — then submit and pray it doesn’t bounce.

PlotScrub does only that job, does it for €99–€499/mo instead of €12K/yr, and is self-serve so a non-technical compliance person can use it the week they realize they’re in scope. The wedge is “fix my filing before it bounces,” not “transform your supply chain.”

4. Target market

• Primary customer: EU-based importers and traders of EUDR-covered goods, €2M–€50M revenue, 5–50 staff. Specialty coffee roasters, bean-to-bar chocolate makers, furniture and flooring importers, tyre/rubber distributors, paper and packaging buyers, small timber merchants. The person who buys is the ops/compliance/quality manager or the owner-operator — not a procurement VP.
• Why they buy: “I import from 30 suppliers, half of them sent me coordinates in a format I don’t understand, my customs broker says the DDS has to be perfect, and if it bounces my container sits at the port. I am not buying a €12,000 platform to solve a spreadsheet problem.” Their pain is per-consignment and tied to a hard money event — held goods.
• Rough TAM reasoning: Tens of thousands of EU SMBs import covered commodities. Even the derived-product tail alone (furniture, chocolate, paper, tyres) is large. Capturing a few thousand at €200/mo average is a $5M+ ARR business without ever touching an enterprise.
• Why now for them: Deadline 30 Dec 2026 for medium operators; they discovered they’re in scope only after the Dec 2025 scope clarifications; their first filings will fail unless someone validates the geometry first.

5. Product sketch (MVP)

• Drop-anything coordinate intake: upload Excel/CSV, KML/KMZ, GeoJSON, shapefile, or a photo/PDF of a map or coordinate list — PlotScrub extracts and normalizes the plots.
• Geometry doctor: detects and auto-fixes the killers — self-intersecting polygons, unclosed rings, points-where-polygons-required (over 4 ha), wrong/missing CRS, decimal/format errors — and shows a before/after map.
• TRACES schema pre-flight: validates the full DDS payload against the live TRACES NT rules (HS code present, plot IDs unique and linked to batch, supplier DDS references, EORI/VAT, quantities/units consistent) and lists every blocker in plain language.
• One-click export: produces the exact GeoJSON + DDS field set ready to paste/upload into TRACES NT (no direct API dependency required for v1).
• Supplier collection links: send a supplier a simple link/form to drop their plot coordinates straight into your file — no GIS knowledge needed on their end.
• Consignment library: every DDS you’ve built, re-runnable next shipment (suppliers repeat) so the second filing takes minutes.
• Bounce log: paste the TRACES rejection message and PlotScrub maps it to the exact field to fix.

6. AI angle — what’s load-bearing

Two layers. The deterministic layer (geometry validation, CRS correction, TRACES schema checks) is rules, not AI — and must be exact, because a wrong “fix” is worse than no fix. The load-bearing AI is the front door: turning the genuinely unstructured mess suppliers send — a photographed hand-drawn plot map, a PDF table with merged cells, a WhatsApp screenshot of a pin, free-text “the farm is 2 km north of X village” — into structured candidate coordinates a human can confirm. Without that, the importer is back to manual transcription, which is exactly the labor the product exists to kill. Remove the AI parsing and you’re just another GeoJSON linter; the messy-input ingestion is the reason a non-technical buyer can actually use it.

7. Localization angle (if any)

This is a localization play in reverse — it’s EU-regulation-native, not a generic global tool. The moat is fluency in the EU’s specific machinery: TRACES NT field rules, the EUDR GeoJSON spec, HS code mapping for covered/derived products, EORI/VAT formats, and the moving scope (printed books dropped Dec 2025, leather contested). Multilingual matters on the supplier-collection side (Portuguese, Spanish, French, Vietnamese, Bahasa for upstream cooperatives), but the buyer-facing product is EU-English/German/French first.

8. Business model — path to $1M–$5M ARR

• Pricing: Starter €99/mo (up to ~25 plots/consignment, 5 consignments/mo); Pro €299/mo (unlimited plots, supplier links, bounce log); Plus €499/mo (multi-user, audit export for their customs broker). Annual prepay discount.
• ACV: ~€2,400 blended.
• Rough math to $1M ARR: ~350 customers × €240/mo × 12 ≈ $1.0M. Reachable within the addressable SMB tail.
• Rough math to $5M ARR: ~1,700 customers, or push ACV up via per-consignment overage and a “broker/consultant” reseller tier (compliance consultants filing on behalf of many importers — clear expansion wedge).
• Expansion path: seats (broker/consultant accounts manage many importers), volume tiers as consignment count grows, and adjacent EU traceability mandates (CBAM, packaging/PPWR, digital product passport) reuse the same “validate-before-you-file” engine.

9. Go-to-market wedge — first 100 customers

• Customs brokers and freight forwarders are the channel. They’re the ones telling importers “your DDS must be perfect” and they hate fielding bounced filings. Partner with 20–30 EU customs brokers; they refer every covered-goods client. Revenue share or flat referral fee.
• Scrape the in-scope SMB lists by commodity: EU specialty coffee roaster directories, bean-to-bar chocolate maker associations, furniture/flooring importer trade bodies. A few thousand named businesses. Personalized email: “Your first EUDR filing will bounce on geolocation — here’s a free check of one consignment.”
• Free single-consignment validator as the hook: let anyone paste one supplier’s coordinates and see exactly what would bounce in TRACES, gated to paid for export and multi-consignment. The “your file is broken, here’s why” moment sells itself.
• Ride the consultant/advisory ecosystem: EUDR compliance consultants are booked solid and would rather charge for advice than wrangle GeoJSON. Reseller tier turns them into a sales force.
• Content on the exact rejection errors: “Why your TRACES DDS bounced: the 10 geolocation errors” ranks for the panicked search the week before a deadline.

10. Build complexity — justification

Medium. The geometry/CRS validation uses off-the-shelf geospatial libraries; the TRACES schema rules are public and finite; the AI extraction layer is standard vision/LLM parsing with human confirmation. The hard part is not technology — it’s getting the TRACES NT field rules and EUDR GeoJSON spec exactly right and keeping them current as the Commission tweaks scope and FAQs (they did so repeatedly through 2025–2026). A technical founder with a compliance advisor ships a credible v1 in 3–4 months. Direct TRACES API submission can wait for v2; v1 exports a validated payload the user uploads themselves.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Compliance-assist tool; importer remains the legal filer.
Ethical — no harm / dark patterns	✅	Helps real deforestation due diligence; no dark patterns.
Market exists (evidence above)	✅	Hard deadline, documented rejection pain, funded incumbents, growing market.
1–5 person team can build this	✅	Geospatial libs + LLM parsing + public schema.
Launchable with <$50K / ₹40L	✅	Self-serve SaaS, no capital-intensive infra.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	15/20	Per-consignment pain tied to held goods, but it’s a “make-it-pass” chore, not daily hair-on-fire revenue loss.
Demand evidence	15	12/15	Strong: documented #1 rejection cause, funded market, in-scope SMBs surprised — but mostly vendor-sourced, not raw customer quotes.
Build feasibility	15	11/15	Deterministic core + AI intake are doable; staying current with TRACES rule changes is real ongoing work.
Distribution clarity	15	11/15	Customs brokers + named commodity directories + free-check hook = concrete. Conversion unproven.
Revenue mechanics	15	11/15	Pricing sits well under incumbents; ACV reasonable; broker reseller tier is upside.
Time to first revenue	10	7/10	Self-serve with a sharp wedge; 6–10 weeks to first paid once v1 ships.
Defensibility	10	5/10	Execution + regulatory-fluency moat only; a competitor could copy the validator. Speed and broker relationships are the edge.
Total	100	72/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

Key assumptions to validate (3–5)

1. Assumption: Small importers will pay €99–€499/mo for a focused validator rather than absorb it into an existing broker relationship for free. How to test: Offer the free single-consignment check to 50 in-scope importers; measure how many hit the paywall and convert.
2. Assumption: Geolocation rejection is painful enough before the deadline that medium operators act in 2026 rather than waiting. How to test: Interview 20 covered-goods importers on whether they’ve attempted a test DDS and what bounced.
3. Assumption: Customs brokers will refer rather than build/bundle it themselves. How to test: Pitch 10 EU brokers on a referral deal; measure interest and any “we’ll just do it ourselves” pushback.
4. Assumption: Messy supplier inputs are parseable accurately enough that the AI layer saves real time vs. manual entry. How to test: Collect 30 real supplier coordinate files; measure extraction accuracy and human-correction time.

Risk flags

1. Regulatory timing risk: Micro/small operators don’t bind until 30 June 2027, and the Commission has postponed and rescoped EUDR repeatedly. A further delay or carve-out softens urgency for part of the market.
2. Platform dependency: TRACES NT field rules and the EUDR GeoJSON spec can change; the product must track them or it ships wrong “fixes.” This is also the moat, but it’s maintenance you can’t skip.
3. Incumbent down-market move: osapiens/TraceX/Coolset could launch a cheap self-serve tier. Mitigant: their motion is enterprise sales; a focused SMB self-serve product with broker distribution can win the tail before they bother.
4. Commoditization: A pure GeoJSON linter is copyable. The defensible layer is messy-input parsing + broker channel + consignment library lock-in, not the validation itself.

14. Structured verdict

Score:                  72/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder (geospatial comfort) + EUDR/customs-compliance advisor
Time to revenue:        6–10 weeks after a 3–4 month v1 build
Capital to launch:      $8–15K ($/€ for geospatial APIs, LLM usage, EUDR advisory hours)
Top 3 assumptions to validate first:
  1. Importers pay for a standalone validator vs. lean on broker for free — free-check paywall conversion test
  2. Medium operators act in 2026 (not waiting) — 20 importer interviews on test-DDS attempts
  3. Customs brokers refer rather than self-build — pitch 10 brokers on referral terms
Kill criteria:
  - Abandon if <8% of free single-consignment users hit the paywall and convert across 50 trials
  - Abandon if a further EUDR postponement pushes the medium-operator deadline past 2027
  - Abandon if 3+ of 10 brokers say they'll bundle/build it themselves rather than refer

15. Next step — 1-week validation sprint

• Day 1–2: Build a throwaway landing page + a manual “free EUDR geolocation check” offer. Collect 30 real supplier coordinate files from in-scope importers (coffee roaster / furniture / chocolate forums, LinkedIn outreach).
• Day 3–4: Hand-run those 30 files through open geospatial tools + a TRACES schema checklist. Record: how many would bounce, what the top errors are, and how long a fix takes manually. Simultaneously DM 10 EU customs brokers about referral interest.
• Day 5: Decide go / no-go. Falsifiable bar: GO only if (a) ≥60% of the 30 files have at least one blocking geolocation error, AND (b) ≥3 of 10 brokers express concrete referral interest. If either fails, the pain or the channel isn’t real enough — stop.