TareTally — EPR packaging-declaration engine for cross-border Amazon sellers

1. One-liner

Calculates packaging weight per material per EU country and hands sellers a ready-to-file EPR declaration packet.

2. Trend signal — why now?

The EU packaging regime just turned from “paperwork some sellers ignored” into “marketplace-enforced or you lose money automatically.” Three hard dates stack up in 2026:

• WEEE EPR numbers were mandatory on Amazon by 31 Dec 2025; packaging, batteries, tyres, oils numbers by 31 Mar 2026 (Lovat Compliance / Amazon Seller Central).
• PPWR (Regulation 2025/40) — most provisions apply from 12 Aug 2026, and from that date marketplaces must verify that every seller shipping into any EU country meets packaging-EPR obligations. No size exemption. (ecosistant, Intertek, Lovat.)
• Sellers who don’t enter valid EPR numbers in time get auto-enrolled in Amazon’s “Pay on Behalf” — Amazon pays the eco-fee and adds a service surcharge, debited directly. Non-compliance can also mean delisting or account suspension.

The pain underneath isn’t getting a registration number — it’s the recurring declaration: weight in kilograms of each packaging material placed on each national market per reporting period. Amazon’s own policy study found 64 unique registration fields across 10 EU markets (11 in Belgium/Spain to 20+ in Sweden), 55% of fields country-specific, dual government+PRO registration in Italy/Spain/Poland, national e-ID logins in Spain/Sweden/Poland/Italy, Belgium still on offline email forms, and every portal in the local language only (Amazon policy study, 2026).

The repeating math is the killer. As one compliance vendor put it: “Forty SKUs across three material layers in five countries produces roughly 600 data points per reporting cycle” (Minefield Navigator). Spreadsheets rot between cycles; auditors catch the gaps by comparing declared tonnage to shipment volumes.

Provenance:

• Signal 1 (Demand): Amazon’s own study — 64 unique registration fields across 10 EU markets, small businesses struggle, 55% country-specific — https://www.aboutamazon.eu/news/policy/new-study-why-small-businesses-struggle-with-eu-packaging-rules-and-how-to-fix-it — 2026
• Signal 2 (Feasibility/Economic): PPWR provisions apply 12 Aug 2026 with marketplace enforcement; sellers auto-enrolled into Amazon “Pay on Behalf” + surcharge if non-compliant — https://www.ecosistant.eu/en/epr-registration-number-amazon/ — 2026
• Signal 3 (Economic): Compliance providers (Lovat) already charging €190–250/yr per category for registration+reporting; full-service ARs charge more — money clearly moving — https://vatcompliance.co/blog/complete-epr-guide-for-amazon-sellers/ — 2026 Category: Regulatory arbitrage

3. The opportunity

The existing market splits into two camps, and both leave the smallest cross-border sellers exposed:

1. Full-service Authorized Representatives / consultants (Deutsche Recycling, ecosistant, EU Compliance Partner) — they’ll register you and be your legal AR, billed per-country, per-category, with manual onboarding. Great margin, slow, and priced for brands that can absorb €500–2,000+ in service fees across markets.
2. Registration-and-reporting subscriptions (Lovat from €190/yr) — cheaper, but they hand you a portal and still expect you to know your packaging weights by material and country.

Nobody owns the genuinely painful, repeating middle: turning “here are my products and my sales-by-country” into “here is the exact tonnage of PET, paper, glass, aluminium I placed on each national market, formatted for each registry, recalculated every cycle.” That’s the 600-data-points-per-cycle problem. It’s not a legal problem — it’s a data-extraction-and-mapping problem that vision and classification models make cheap for the first time. We take that slice, stay out of the regulated AR role, and either let sellers file themselves or export a clean packet to whatever AR they already use.

The incumbents won’t chase it because their business model is per-country service revenue; a self-serve data engine cannibalizes the manual hours they bill for.

4. Target market

• Primary customer: Non-resident e-commerce sellers shipping into 2–6 EU countries — UK sellers post-Brexit, US Pan-EU FBA sellers, India/Turkey/China D2C exporters — with 50–2,000 SKUs and €100K–€5M EU revenue, no in-house compliance staff.
• Why they buy: “I have an EPR number but I have no idea what to actually declare, and if I get it wrong Amazon either suspends me or starts charging me a surcharge.” The declaration is annual-or-quarterly and never finishes — new SKUs land monthly.
• Rough TAM reasoning: Amazon EU alone has well over a million active third-party sellers; a large share sell cross-border, and PPWR’s no-size-exemption rule pulls in even occasional cross-border sellers from Aug 2026. A target wedge of even 20,000–50,000 small non-resident multi-country sellers at €30–80/mo is a comfortably >$5M-ARR pond.
• Why now for them: The 12 Aug 2026 marketplace-enforcement switch turns a soft obligation into a hard, automated penalty. Sellers who shrugged at EPR through 2025 are forced to act on a calendar they can see.

5. Product sketch (MVP)

• SKU packaging capture — seller photographs or describes each product’s packaging (box, mailer, void fill, label, tape); the system identifies material type and estimates weight per component.
• Material-weight ledger — every SKU resolves to grams of paper/cardboard, plastic (by polymer where required), glass, aluminium, composite — the unit the registries actually want.
• Sales-by-country ingest — pull units-sold-per-marketplace from Amazon/Shopify reports; multiply by per-SKU packaging weight to get national tonnage per material per period.
• Per-country declaration packets — output formatted to each registry’s fields (LUCID/Germany, Citeo+ADEME/France incl. Triman & Info-tri, CONAI/Italy, Ecoembes/Spain, etc.), with the country-specific fields pre-filled.
• Audit trail — versioned record of declared tonnage vs. shipment volumes, the exact comparison auditors run, exportable as proof.
• Recalc on change — new SKUs and each new reporting cycle re-run automatically; flags when a country’s threshold or fee tariff changes.
• AR hand-off export — one-click packet a seller can email to their existing Authorized Representative instead of filing themselves.

6. AI angle — what’s load-bearing

Remove the AI and this collapses back into the spreadsheet that’s already failing sellers. The load-bearing work is twofold: (1) vision + classification to turn a photo of a parcel into a structured material-and-weight breakdown — polymer identification, multi-layer decomposition, weight estimation — without the seller hand-measuring 600 components; and (2) mapping intelligence to translate one canonical packaging record into each country’s idiosyncratic field set and fee tariff (55% of fields are country-specific; portals are local-language-only). Both were manual-labour problems until cheap vision and language models made per-SKU automation viable. The declaration math itself is arithmetic — the AI is what makes the input and the per-country translation tractable at SKU scale.

7. Localization angle (if any)

This is a localization play — but localization across EU member states, not a single-country wedge. The moat is precisely the per-country quirks: France’s Triman + Info-tri sorting symbols and ADEME/Citeo split; Germany’s LUCID + dual-system provider; Italy/Spain/Poland’s dual government-plus-PRO registration and national e-ID logins; Belgium’s offline forms. A generic global “packaging compliance” tool that doesn’t encode these dies at the first declaration. Pricing also localizes: a €30–80/mo tier works for sellers who can’t justify a €1,500 AR engagement.

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

• Pricing: Tiered by SKU count + number of countries. Starter €29/mo (≤100 SKUs, 2 countries); Growth €79/mo (≤500 SKUs, up to 5 countries); Pro €199/mo (≤2,000 SKUs, all EU + quarterly cycles). Add-on: “declaration done-for-you” review at €49/country/cycle.
• ACV: ~€700–900 blended (most land on Growth, many add a country or a done-for-you review at deadline crunch).
• Rough math to $1M ARR: ~1,300 customers × ~€65/mo avg × 12 ≈ $1.05M.
• Rough math to $5M ARR: ~5,000 paying sellers averaging ~€85/mo, plus done-for-you add-ons at each EU reporting deadline. Plausible given the multi-million seller base and forced-compliance calendar — but requires expanding country coverage and surviving incumbent response.
• Expansion path: more countries per account, more SKUs, more EPR streams (batteries, WEEE, textiles EPR landing 2025–2027), and a per-cycle “audit-ready packet” upsell. ACV grows as sellers expand EU footprint.

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

• Amazon EU seller forums + the EPR threads — Seller Central EU discussion threads on EPR/LUCID and the m2epro / vatcompliance comment sections are full of confused sellers right now. Answer the “how do I calculate packaging weight” question with a free SKU-weight estimate, convert.
• Deadline-timed cold outreach — scrape sellers visibly shipping into FR/DE who lack a valid EPR number on their listings (Amazon surfaces compliance gaps); send a personalized “you’re about to get Pay-on-Behalf’d on 12 Aug” Loom + a free first-country declaration. Time it to the Aug 2026 enforcement date.
• Prep-center & 3PL partnerships — EU prep centers and Pan-EU enablement agencies field this question from every client. Revenue-share referral; they hate doing the packaging math themselves.
• AR/consultant white-label — the cheaper ARs (the Lovat tier) need a data engine behind their portal. Sell them the calculation layer wholesale; they keep the AR relationship, we get volume.
• Vernacular YouTube/WhatsApp for India & Turkey exporters — these sellers are heavily underserved in their own language on EU compliance; a “how to not get suspended on Amazon EU” explainer is cheap, targeted distribution.

10. Build complexity — justification

Medium. The vision/classification for packaging-material-and-weight extraction is off-the-shelf model work with a calibration dataset you can bootstrap from common packaging types. The genuinely laborious part is encoding each country’s registry field set, fee tariff, and format — that’s domain grind, not research, and it’s also the moat. Marketplace report ingestion (Amazon/Shopify) is standard integration. A 2–3 person team ships a 2-country MVP (Germany + France, the highest-volume, highest-pain pair) in ~3–4 months, then adds countries one at a time. No novel infra, no model training breakthrough required.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	We prepare data/declarations; the seller (or their accredited AR) files. We do not act as the regulated AR.
Ethical — no harm / dark patterns	✅	Helps sellers comply with environmental law; no dark patterns.
Market exists (evidence above)	✅	Forced-enforcement calendar, paying incumbents, Amazon’s own study.
1–5 person team can build this	✅	2–3 people, country-by-country rollout.
Launchable with <$50K / ₹40L	✅	Off-the-shelf models + integration work; no capex.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Hair-on-fire near the deadline: get it wrong and Amazon suspends or surcharges you automatically. Recurring, not one-off.
Demand evidence	15	12/15	Amazon’s own study quantifies the struggle; multiple paid incumbents; forced enforcement date. Strong, though most evidence is vendor/marketplace-sourced rather than raw seller threads.
Build feasibility	15	11/15	Models off-the-shelf; the work is per-country encoding. Real but bounded grind; 3–4 months to credible v1.
Distribution clarity	15	11/15	Named channels (seller forums, deadline outreach, prep-center referral, AR white-label) with a hard catalyst date. Conversion math still unproven.
Revenue mechanics	15	11/15	Pricing benchmarked below incumbents; ACV reasonable; $1M path needs ~1,300 customers — achievable but requires real acquisition.
Time to first revenue	10	7/10	A 2-country MVP can pre-sell against the Aug 2026 deadline; revenue in 6–10 weeks of a usable build is realistic.
Defensibility	10	6/10	Soft moat: accumulated per-country encoding, packaging-weight dataset, and workflow lock-in via versioned audit trails. Copyable, but the country-grind plus a deadline head-start protects 6–12 months.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — needs someone who can ship vision/classification and either has or quickly acquires EU EPR domain knowledge (or a compliance advisor).

Key assumptions to validate (3–5)

1. Assumption: Sellers will trust an estimated packaging weight enough to declare on it. How to test: Run 20 real SKUs through a manual version, compare to seller-measured truth, show the error band; ask 15 sellers if it’s “good enough to file.”
2. Assumption: The data layer can stay out of the regulated AR role and still be valuable. How to test: Interview 10 sellers + 3 ARs — do sellers want a packet they file themselves, or do they only want full-service? If only full-service, the wedge narrows.
3. Assumption: €79/mo is payable by a €100K–€5M cross-border seller. How to test: Pre-sell 10 annual seats at the Aug-2026-deadline pitch before building all countries.
4. Assumption: Per-country encoding is grind, not a moving target that breaks monthly. How to test: Track LUCID/Citeo/CONAI format changes over 8 weeks; estimate maintenance load.

Risk flags

1. Platform/regulatory dependency: The entire urgency rides on Amazon’s enforcement and PPWR timelines. If the May-2026 proposal to suspend AR obligations for micro/small enterprises passes (decision due 1 Oct 2026), some urgency softens — though the declaration obligation itself survives.
2. Incumbent response: ecosistant/Lovat/Deutsche Recycling could bolt on a cheap self-serve calculator. Our defense is speed, seller-grade UX, and the AR-white-label channel.
3. Liability creep: Sellers who get fined may blame the tool. Mitigate with clear “you/your AR file; we prepare and show our math” positioning and a visible audit trail.
4. Maintenance drag: 27 member states’ formats drift; coverage breadth is a treadmill. Start with the FR/DE pair that covers most volume and pain.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder (vision/ML + integrations) with an EU EPR compliance advisor
Time to revenue:        6–10 weeks to first paid (pre-sold against Aug 2026 deadline); ~3–4 months to 2-country MVP
Capital to launch:      $8–15K (model inference, integration dev, a compliance advisor's time)
Top 3 assumptions to validate first:
  1. Estimated packaging weights are accurate enough to declare on — bench 20 SKUs vs. measured truth
  2. The data-layer wedge (not full AR) is what sellers want — interview 10 sellers + 3 ARs
  3. €79/mo is payable — pre-sell 10 annual seats before building all countries
Kill criteria:
  - Abandon if <3 of 15 deadline-targeted sellers pre-pay after a working FR/DE demo
  - Abandon if packaging-weight estimates can't get within an audit-defensible band on common packaging
  - Abandon if an incumbent ships an equivalent self-serve calculator before the 2-country MVP and locks the forums

15. Next step — 1-week validation sprint

• Day 1–2: Manually build the FR + DE declaration for 5 real seller catalogs (borrow them) by hand — prove the 600-data-points pain is real and that our material-weight ledger maps cleanly to LUCID + Citeo/ADEME fields, Triman included.
• Day 3–4: Take that hand-built output to 15 cross-border sellers (Amazon EU forums, prep-center intros) with a deadline pitch; offer a pre-sold annual seat at €79/mo.
• Day 5: Decide go/no-go on a falsifiable bar: ≥3 of 15 sellers pre-pay for the FR/DE packet, and packaging-weight estimates land within an audit-defensible band on at least 4 of 5 catalogs. Below either threshold → no-go.