Sourcing Vintage Goods Globally

The problem

You started reselling six months ago. The first three months felt like a gold rush — Goodwill bins had Carhartt and Pendleton you flipped for $80, the local estate sales were full, and the dashboard graph kept going up and to the right. Then the slope flattened. By month four, the same three thrifters you see every Saturday at the same bins were buying the same lots. By month five, the thrift store managers were pulling the good stuff for their own eBay store before it hit the floor. By month six, you were driving 45 minutes to a suburban Goodwill you used to skip and coming home with half a tote of mediocre denim.

This is not a personal failing. The local thrift ecosystem in any metro area can support a finite number of resellers, and the math is unforgiving — working harder locally cannot break the ceiling because the ceiling is supply-side, not effort-side. And the day you decide to break out by buying a $500 “vintage Japan mystery lot” off some Instagram seller who DMed you first is the day you wipe out two weeks of margin in a single bad decision. The lot arrives, three pieces are damaged, half are the wrong era, and you absorb the loss and tell yourself you “learned something.”

The reseller who actually breaks the ceiling does not gamble. They build a system: pick a niche, score it against four numbers, learn the English-language sources first, graduate to Japanese auctions through a proxy, run every lot through landed-cost math before bidding, and authenticate from photos before the wire transfer. That is the system this book lays out, and it is the system that takes the median global-sourcer from $2,500 a month to $6,000 a month within 90 days.

What most people get wrong

They double down on local sourcing instead of opening a new vein. The three escalations new resellers reach for — radius expansion, category expansion, retail arbitrage at TJ Maxx — all fail predictably. The 60-mile thrift run pays for the gas with a single Levi’s lot the first time and runs at $15/hour the fifth time. Buying ceramics because the estate sale had a nice set means your denim-calibrated photo template, audience, and listing voice now have ceramics sitting in them. Listing TJ Maxx clearance as “vintage-style” gets your seller account a strike. Going harder at the same supply runs out of moves. The fix is not effort — it is opening a sourcing vein the resellers in your zip code do not have.

They forget landed cost and celebrate item cost. A $30 Levi’s lot on Yahoo Japan looks like a steal. A $30 lot plus 7-15% proxy fee plus Japan-side shipping plus international freight plus payment-processor spread is closer to $66 landed for a 10-item lot — $6.60 per item — and that is before you factor in the 30% of items that turn out unusable. The reseller who lives by a landed-cost spreadsheet outperforms the reseller who lives by “it feels cheap” by a factor of 2-3x in annual profit, not because the spreadsheet bids better, but because the spreadsheet makes you walk away from the bids that look like wins and lock up your capital at 8% margin for 60 days.

They trust the photos. Every reseller has a “$500 lot” story — bid the cap because the photos looked perfect, then the lot arrived with damage that was not visible, sizing inconsistent with the measurement table, or the bad pieces buried behind the good ones in the photo set. You cannot prevent every bad lot. You can build the discount into your max bid (do not bid above the price at which the lot still works if 30% of items are unusable), ask for additional photos before bidding on anything over $100, and authenticate from photos using the three-layer framework — provenance, object, anomaly — that catches 80% of the obvious problems.

A working approach

The full system is eight steps in the order they actually need to happen. Skipping ahead breaks the chain.

STEP 1 — Pick the niche. Score candidates against four numbers
         (sell-through, ASP, comp depth, friction). Run one or two.
STEP 2 — Learn English-language sources first. Bank & Vogue,
         RagHouse, Beyond Retro Wholesale, eBay wholesale lots,
         Whatnot bulk. Two test lots before Japan.
STEP 3 — Open a proxy. ZenMarket for beginners, Buyee for
         volume, FromJapan for heavier orders. Fund with Wise
         to save 1-2% on FX.
STEP 4 — Bid Yahoo Japan + Mercari Japan. Use the proxy's
         bid-scheduler — never manually bid at close.
STEP 5 — Run every lot through landed-cost math before bidding
         the cap. Below 40% margin at the comp median, walk away.
STEP 6 — Authenticate from photos: provenance, object, anomaly.
         Stay out of designer handbags, high-end watches, sneakers.
STEP 7 — Comp research before listing across eBay, Vinted, Depop,
         Mercari, Whatnot. Median + condition adjustment.
STEP 8 — Run the 72-hour pipeline. Sort A/B/C/D. Photograph in
         batch. List from templates. Pack-and-ship daily.

A 2024 community survey of 412 resellers who reported sourcing 50%+ of their inventory from non-US lots showed the median monthly revenue moving from $1,800-$2,600 (local peak at months 4-6) to $3,500-$6,200 (months 7-12 with global added) to $6,000-$14,000 (months 13-24, global-first and niched). Monthly hours sourcing actually drop alongside revenue — from 45-65 at local peak to 30-45 once the niche, the supplier scorecard, and the landed-cost discipline are in place. Global sourcing is not “work more for more money.” It is “work less for more money,” but only after the system is built.

Niche selection: the four-number test

The single biggest mistake new global sourcers make is buying interesting things. The first time you scroll Yahoo Japan and see a $40 lot of mid-century ceramics, a $60 box of vintage cameras, a $25 stack of 1980s Comme des Garçons sweatshirts, and a $30 trove of indigo-dyed boro patches in the same search, the urge is to buy one of each. Do not. A reseller running five niches at once runs five photo workflows, five listing templates, five buyer audiences, and five trust-rebuilding processes after every bad lot. They run no niche well. eBay’s internal data shows resellers in one or two categories have 38% higher sell-through than resellers in three or more.

The four-number test is the niche filter. For each candidate niche, pull sell-through rate (percentage of active listings that sell within 30 days — under 20% is oversupplied, over 40% is room), average sale price (median of the last 90 days), comp depth (50+ comps means mature, under 10 means either emerging or dead), and friction-to-flip (minutes to photograph, list, and ship the median item). Multiply STR by ASP by comp-depth, divide by friction. The categories with the best ratios are where you should be. Vintage denim wins for a reason — STR 35-45%, ASP $60-$220, deep comp pool, 18-minute listing time. Mid-century Japanese ceramics is the higher-ASP-higher-friction sibling. Vintage workwear, entry-tier watches (Seiko 5, Citizen, Orient), and mid-century European homeware round out the five niches that consistently score well for global sourcers.

Yahoo Japan + Mercari Japan + Catawiki: the workflow

Yahoo Japan Auctions is the eBay of Japan and the dominant secondhand marketplace in the country. Mercari Japan is the Vinted of Japan — peer-to-peer, lower per-item prices, more individual items. Catawiki is the pan-European specialty-auction platform with category experts vetting every lot. Between them, you cover 90% of the global sourcing opportunity for vintage resellers.

Yahoo Japan requires a proxy because registration needs a Japanese phone number and bank account. The auction-close discipline is the part most new sourcers get wrong: auctions close on Japan time, which is 7-9 a.m. US Eastern. Trying to manually bid at the close means you lose half your lots to snipers. Use the proxy’s bid-scheduler, set your max bid in advance, and walk away. You win or lose on price, not on whether you are at the keyboard at 4 a.m. The same discipline applies to eBay UK (closes 3-5 p.m. ET) and Catawiki (2-4 p.m. ET on themed weekly auctions, usually Friday or Sunday).

Mercari Japan is structurally similar to Yahoo Japan but faster — most listings are Buy-It-Now, and most Japanese sellers ship within 24 hours of payment. The only meaningful delay is the international leg. Batch four to six items at the proxy before shipping internationally and you drop per-item shipping cost by 60%. Catawiki is the easiest of the three to start with because it is in English, every lot is reviewed by a category expert before listing, and shipping is professionally handled — the trade-off is higher per-item pricing and a 9% buyer’s premium. Many resellers run Catawiki as their entire European leg for the first 6-12 months and graduate to eBay UK once their authentication instincts are sharper.

ZenMarket, FromJapan, Buyee: the proxy mechanics

The proxy service is the most important relationship in your global-sourcing operation, and the place where new resellers lose the most money to bad choices. ZenMarket is the most beginner-friendly — clean English interface, supports Yahoo Japan, Mercari Japan, Rakuten, Amazon Japan, service fee 300-500 yen per item plus 5-10% commission, free Japan-side warehousing for 45 days; best for first-time bidders. FromJapan has a more bureaucratic interface but very experienced support, 300 yen per item plus 10% on item cost, slightly cheaper international shipping for heavier parcels; best for ceramics and multi-item lots. Buyee is backed by Tenso at massive scale and integrated directly with Yahoo Japan, 300 yen flat per item but international shipping sometimes 10-15% pricier; best for high-volume sourcers who want one service for everything.

All-in proxy markup typically lands at 7-15% of total order cost once you add service fee, commission, and Japan-side handling. That number is the line item you keep in your landed-cost spreadsheet on every lot. The other hidden cost is FX: most proxies embed a 1.5-2.5% spread when converting USD to JPY through their default payment methods. Funding with a Wise transfer or ACH wire to a multi-currency account saves 1-2% — on a $2,000 monthly sourcing budget, that compounds to $240-$480 a year you keep instead of give to the proxy.

The payment chain itself is escrow by design. You never pay the Japanese seller directly. The proxy is your buffer — if the item never arrives at their warehouse, they refund. If it arrives damaged, they negotiate with the seller. Yahoo Japan’s 30-day seller-evaluation system gives sellers strong incentive to resolve disputes; community data shows 78% of disputed wins resolve with full or partial refund without escalation. The discipline is to never ask the proxy or the seller to under-declare customs value to reduce duties — it is illegal in the US and EU, the proxy can lose its forwarding license, and if the shipment is lost, insurance only covers the declared value.

Landed-cost math: the $800 de minimis sweet spot

US import law currently exempts shipments valued at $800 or less per person per day from formal customs entry and duties. This is Section 321 de minimis, and it is the single most important number in cross-border sourcing math right now. If your shipment lands at under $800 declared value, you typically pay $0 in duties and the package is released to you without formal customs paperwork. The sweet spot for batched proxy shipments is $200-$700 declared value — below de minimis, high enough to amortize international freight.

The six cost components of every international lot are item cost (local currency), proxy commission plus per-item fee, domestic shipping in the origin country, international shipping, US duties (zero under de minimis), and miscellaneous fees (currency-conversion spread around 2%, payment processor 1-2%, optional insurance 1-3%). The bonus landed-cost-calculator.csv is the working template — columns for item cost, currency, proxy fee percent, shipping, duties, fees total, computed landed cost in USD, target sell price, and margin percentage. The reseller who runs this on every lot before bidding the cap is the reseller whose annual P&L works.

Two side notes on duties for the shipments that do cross $800: cotton clothing duties run 8-16%, wool clothing 18-25%, ceramics 4.5-8%, watches 3.1-6% plus a small per-movement fee, and genuinely antique items (pre-1925 at time of import, with documentation) qualify for 0%. The Section 321 threshold has been the subject of significant political discussion and could change — check current US import law before relying on de minimis as your structural sourcing assumption.

Authentication from photos: three layers, fifteen seconds each

You are not going to authenticate items at the level of a sworn expert. That is not the job. The job is to catch the 80% of obvious fakes, mis-described items, and condition issues before you bid. The framework has three layers. Provenance signals — who is the seller, how long have they been selling vintage, do they have positive feedback in your niche, do their photos have a consistent style, is the lot context coherent or is it a strange mix that signals a flipper offloading mistakes? Estate auction houses through LiveAuctioneers or Invaluable provide auction-catalog descriptions with provenance — much higher trust than open marketplaces. Object signals — for vintage Levi’s: red tab placement (Big E vs. small e), care tag style, selvedge edge color, button stamp number on the back of the waistband. For studio ceramics: maker’s mark cross-referenced against published catalogs, foot ring construction (hand-thrown shows tool marks), glaze pooling, weight. For entry-tier watches: dial printing under macro (refinished dials show fuzzy lettering), lume patina (1970s with fresh-white lume is suspicious), crystal type, crown signature, movement shot if the seller will provide one. Build one cheat sheet per brand you regularly source. Anomaly signals — listing photographs the item on a hanger that obscures the tag, “vintage-inspired” or “vintage style” in the title (deliberately ambiguous, usually not actually vintage), seller cannot answer specific dating questions, price is dramatically below comps with no clear reason, listing avoids showing the spine, the staples, the foot ring, or the movement. For any item over $100, ask the seller for additional photos before bidding — the standard request format works in any language. Sellers who refuse are not always being deceptive, but the refusal is itself a signal.

Comp research across eBay, Vinted, Depop, Mercari, Whatnot

Pricing is where most resellers leak the most money — either under-pricing by 15-40% on items they could have priced higher, or creating ghost listings that sit forever on items priced too high. The bonus comp-research-template.md walks the five-step flow: identify the item with precision (brand, item type, era, country of manufacture, size in seller-country sizing then in US, color, condition, any rare feature), search the three primary comp platforms with sold-listings-only filtered to 90 days, take the median, adjust for condition, set your list price 5-15% above the median if you have above-median condition or provenance.

Platform-specific notes. eBay (US) is the primary comp database for nearly every vintage category — sold listings deep enough for any item that has moved more than five times in 90 days. Vinted has surged for Y2K, athleisure, and Gen Z fashion comps; if you source 2000s-era clothing, Vinted comps often run 20-40% above eBay because the buyer base skews younger. Depop is the curated-aesthetic platform — vintage and Y2K with photo-styled listings, comps run higher than eBay but with a steeper condition penalty for any visible flaw. Mercari (US) is closer to eBay with slightly faster sell-through on small accessories and collectibles. Whatnot is the live-stream auction platform that has become the dominant comp source for trading cards, comics, and certain sneaker subcategories — if you source there, Whatnot comps diverge meaningfully from eBay and you should weight them accordingly. The 8-12 comps you record per item give you a defensible list price: above-median condition with provenance, price at the 60th-70th percentile; median condition, median price; below-median, 25th-40th percentile with every flaw disclosed. Vague condition descriptions (“good vintage condition”) are the single biggest reason items sit. “Light wear at hem, no holes, no stains, original buttons intact, fits true to vintage 32x32” tells a buyer everything and gets the question-free purchase.

Where this scales

The article walked the eight-step system. The book covers it in operating detail: the niche scoring sheet for the five categories that consistently work for global sourcers, the complete English-language source map (Bank & Vogue, RagHouse, Beyond Retro Wholesale, B-Stock, Whatnot Bulk, eBay Wholesale Lots, Faire, LiveAuctioneers) with min-buy quantities and lead times, the proxy account-opening workflow, the Japanese vocabulary file (40-60 reseller-relevant terms to read 80% of a Yahoo Japan listing natively after three months), the full landed-cost spreadsheet with the duty-rate table, the 20-question supplier vetting scorecard plus the red flags that mean walk away, the five-station listing pipeline that takes a lot from box to listed in 72 hours, and the reorder rhythm that keeps a 30-day forward inventory pipeline running without binge-sourcing.

The chapter on what to do when something goes wrong is the one most reseller books skip. Item never arrives at the proxy — most refund within 14-21 days. Item arrives different than described — the proxy will photograph on arrival if you request it in the bid notes, then negotiate with the seller for partial refund. Item damaged in international shipping — file a claim with the carrier within 21-30 days, photos before fully opening, unboxing video usually required. Customs holds the shipment — almost always resolves within 5-10 business days. These are not edge cases. They happen. The discipline is having the playbook before the problem, not after.

Included with the book

• Landed Cost Calculator (CSV) — the working spreadsheet with example rows across JPY, GBP, and EUR sourcing at different price tiers. Drop your own bids in, see the margin number, walk away when it does not work.
• Supplier Vetting Scorecard (markdown) — the 20-question pre-order email template, the red-flag checklist, and the annual review framework you run on every supplier you keep.
• Comp Research Template (markdown) — the five-step worksheet for any single item, with the eBay / Vinted / Depop / Mercari / Whatnot platform map and the condition-adjustment math.

Get the full picture

For wholesale lots and bulk vintage inventory beyond what Japanese auctions and rag houses cover, profit.deals indexes reseller-grade lot suppliers across consumer categories — a complementary sourcing vein once your global pipeline is running.

Readers of this also chose

Questions readers ask

I am still in my first six months. Should I wait to start global sourcing?

No. Run them in parallel. Local sourcing keeps cash flow steady while Japan lots ship (2-4 weeks) and European lots ship (1-3 weeks). The global lots build the inventory depth that lets you raise prices on your best categories. Hybrid is the right answer for the first half year.

How much capital do I need to start?

About $500-$800 for the first two English-language test lots, plus $200-$400 for your first Yahoo Japan or Mercari Japan items through a proxy. The Section 321 de minimis math works best when your first international shipments are sub-$800 declared value, so you do not need to capitalize a giant first batch. Treat the first 90 days as paid learning.

What if Section 321 de minimis gets lowered or removed?

The math shifts toward larger consolidated shipments where duty cost amortizes across more items. The book covers the duty-rate table by category and the DDP vs. DDU shipping incoterms so you can adjust without re-learning the system. The proxy and authentication mechanics do not change.

Do I need to speak Japanese?

No. The proxies translate the interface. Auto-translate gets you to about 80%; the book teaches the 30-40 reseller-relevant Japanese terms (condition vocabulary, damage descriptions, Edo-era units) that cover the remaining 20%. After three months of regular sourcing you read most of a listing natively.

What if I need a refund?

Checkout runs on Lemon Squeezy. The standard refund window applies. You keep the PDF either way.