From November 1, 2026, tax-free shopping in Japan works differently: you pay the tax-inclusive price in store and get the consumption tax refunded after you leave the country. Official sources publish the rules — never the payout. This page collects real refund records travelers posted on Xiaohongshu, Threads and Instagram, and converts them into payout rates and refund speeds on one consistent basis. Samples are still scarce; every figure carries its source and sample size.
Payout rate = amount actually received ÷ tax due. Only records with full amounts use this basis, so they can be compared side by side.
Tax due = tax-inclusive amount × 10 / 110The original post gave only a percentage, with no purchase amount, so it cannot be converted to the tax basis. These figures are tabled separately and clearly labelled — never mixed with the tax basis — and serve only as a relative reference within one author's single trip.
The tax-basis dataset grew on Nov 13: 7 fully documented records, plus Sample T's Alipay set from Xiaohongshu (converted at the mid-rate) and two partial records that are shown but not counted. Airport cash now spans 76.3–85.1% (n=4). The surprise is the card channel: JPrefund × credit card produced the best confirmed payout in the dataset (94.8%, credited in 2 days) and an 84.6% record with a full fee chain. The 71.8% JPrefund × WeChat record (Nov 11) still stands at the low end of the verified set.
What was deducted: a refund handling fee (1% of the pre-tax amount) plus an extra fee for collecting cash at the airport. Process: pick up a 5%-off coupon at the tax-refund counter before shopping, then return to the counter after purchase (Cantonese-speaking staff available). Refunds to a card or e-wallet require registering on the J-taxrefund website first; cash needs no registration.
The higher rate comes down to the fee: Matsumoto Kiyoshi's refund handling fee is just 0.33% (per the author, quoting store staff). Pay at the tax-free register; staff scan your passport, and the receipt carries a J-taxrefund QR code.
Customs terminal passed in about 10 seconds; the cash-machine queue took 10+ minutes. Doubt flagged: of the ¥400 drop from ¥1,944 to ¥1,544, rounding down explains only ¥44 — the machine gave no breakdown for the rest, so its composition (machine fee / operator fee) is unknown. The author also notes machines inside the restricted area had no queue.
Fee chain, verified against the author's chat record: the drugstore's fee (~16% of its ¥972 tax ≈ ¥156) comes off first, then the machine charges ¥100 per record (2 records = ¥200). Staff guide travelers to the machines before the duty-free shops; the interface is multilingual.
The fee was 0.5% of the pre-tax amount (¥6,500). Purchased Nov 4, credited Nov 6 (2 days) — to a different card than the one used to pay, and the receiving card even refunded ¥9 of overseas transaction fee (1.5% of ¥616) on top. Relayed by Threads @linshibi from a traveler's case, with bill and merchant-record screenshots; counted as a first-hand measurement. Same operator as the 71.8% WeChat record below — channel and store fee together decide the outcome.
Caveat: this 84.6% is computed from the full fee chain, not a confirmed credit — when the author posted (Nov 12) the refund had not yet landed; the authorization SMS quoted 3–14 business days. Card refunds via JPrefund carry up to three fees (processing, transaction, Inter Remit); the first two may be absorbed by the store, the Inter Remit Fee always applies. The author mentions a second case at 95%, but its numbers were not published, so it is not counted.
From a screenshot of the author's jprefund.com purchase history: “Total Fees ¥330” against ¥1,170 of tax due. The fee breakdown is collapsed in the screenshot and the store is not named, so what the ¥330 consists of is unknown. Caveat: this is a small purchase (¥1,170 of tax ≈ ¥12,870 spent) — if any fee component is fixed, small amounts inflate the percentage; that explanation is plausible but unconfirmed. On top of the fee, the author reports a supplier exchange-rate loss of about 4 points on WeChat. Commenters on the same post report similar figures — WeChat and Alipay deductions of about 29% of the tax due — putting JPrefund × WeChat/Alipay deductions in the 28–29% range. Not comparable with Sample A's author-reported 9.5% for JPrefund (different basis).
How the ≈71% is derived: Global Blue's displayed refund was exactly 7.00% of spend in all four records — a pre-fee estimate (the same 7.00% pattern now appears in 9 records across Samples M, R and T). The actual CNY credits imply a conversion rate of ≈0.0388 CNY/JPY, about 7.7% below the mid-rate (≈0.0421); valuing the tax due at the mid-rate, each record works out to 71.0–71.2% of tax due. This is a converted figure — treat it as an estimate, not a JPY-denominated measurement.
Sample T · JPrefund × Alipay (3 records with JPY detail): refunds were 8.64% of Sales at Arc'teryx Gotemba and 6.91% at Alpen TOKYO (×2), at an implied rate of ≈0.0409 (≈2.8% FX loss — better than Global Blue's 0.0388). Whether “Sales” is pre- or post-tax is not stated, so two readings stand: 86.4% / 69.1% of tax due (if pre-tax) or 95.0% / 76.0% (if tax-inclusive). Either way, the same operator charged different rates at different stores.
Sample S · JPrefund × Alipay: fees ¥3,447 and refund ¥14,462 are visible, but the tax line is cropped out of the screenshot. If tax = fees + refund (the layout Sample D verifies), the implied fee rate is 19.2% — an inference, not a measurement. The refund was still stuck at “remitting” 5 days after acceptance.
Sample Y · JPrefund × card (Hands Sapporo): Sales ¥13,640 → refund ¥1,178, credited to the author's card. The Sales basis is unstated: 95.0% of tax due if tax-inclusive, 86.4% if pre-tax. Not counted either way.
Xiaohongshu author 米奇大作战 tested four refund operators on a single trip, all refunded to WeChat, recorded on day 2 after departure. The percentages are the author-reported basis (no purchase amounts published, so not convertible); the ranking holds only among these four.
| Refund operator | Author-reported payout | Hassle | Speed | Field notes |
|---|---|---|---|---|
| J-taxrefundDon Quijote · operated by J&J Tax Free | 10% | ●●●●● | ●●●●● | Highest of the four. Register once and new purchases update automatically; almost no exchange-rate loss. The author calls it the hassle-free pick. |
| JPrefundKaruizawa Prince Shopping Plaza · operated by Smartdetax | 9.5% | ●●●●● | ●●●●● | Second-highest rate, but the author's lowest speed score (1/5): still processing 2 days after departure. Single sample; the final arrival time is unknown. |
| Global BlueA shopping mall (unspecified) | 7.7% | ●●●●● | ●●●●● | Fastest payout, but face verification failed repeatedly — about 20 scans. The refund exchange rate of 3.88 involved a loss. |
| PIE VATGrandberry Park, Minami-Machida · Montbell | 7% | ●●●●● | ●●●●● | Lowest of the four and the most hassle (5/5): upload receipts yourself, photograph your passport and entry sticker, and confirm your departure airport in the app after leaving Japan. |
Speed depends less on the channel than on the refund operator: the fastest had landed before boarding; the slowest was still processing 2 days after departure.
Collect in person at Narita, Haneda, Kansai and five other airports on the day you leave; Self Tax Refund counters are at Narita, Haneda, Naha and elsewhere. The trade-off is an extra cash fee (see the measured 76–85% above).
Fastest in Sample A: already arrived before boarding or on landing; author speed score 5/5. Single sample only.
Sample A records no exact day count; speed score 4/5, second tier alongside PIE VAT.
Still processing when recorded on day 2 after departure; speed score 1/5. This is a single sample (n=1) with no follow-up record — how long it ultimately takes is unknown.
Sample U: purchased Nov 4, ¥616 credited Nov 6. Sample T: all nine refunds (Global Blue + JPrefund, Alipay) from Nov 4–7 purchases had arrived by Nov 8. Sample V's card refund, by contrast, was quoted 3–14 business days by SMS and had not landed when posted.
Accepted Nov 8, 12:06 and still “remitting” when posted on Nov 13 — although JPrefund's own page says e-wallet refunds normally arrive within minutes to 24 hours. The author had no support channel to ask. A stall point, not a loss: the outcome is unknown.
Matsumoto Kiyoshi charges 0.33%, Tobu 1% — every store differs, and this is the main reason payout rates diverge within one channel. Ask at the refund counter before you buy.
An Instagram user (Nov 6, 2026) photographed an in-store J-TaxRefund sign listing a 7.80% service fee. That is the posted rate and it roughly matches measured payouts; the exact base it applies to follows each store's notice.
Sample A recorded a payment rate of 4.22 vs a refund rate of 3.88 — about a 10% gap. The claim that cash “takes about another 10% in fees” is second-hand in the original post, not the author's own measurement — reference only.
Cash Japan charges ¥100 per record and pays out rounded down to the nearest ¥100 (Samples W, X). Reference data F (Threads @link7311, Nov 12): six Global Blue records totaling ¥11,414 came out as ¥10,050 — the machine took 12.0%; and not every store supports cash machines at all.
Processing fee + transaction fee + Inter Remit Fee (Sample V: ¥86 + ¥0 + ¥35). The first two may be absorbed by the store; the Inter Remit Fee always applies. Sample U's store charged 0.5% of the pre-tax amount and the refund landed at 94.8%.
Enter this trip's total tax-inclusive spending and pick a refund channel. All three channels now have tax-basis measurements: airport cash n=4 (76–85%), card via JPrefund n=2 (84.6–94.8%, one not yet credited when posted), and wallet points from JPrefund × WeChat (71.8%) and Global Blue × Alipay (≈71%, converted).
You still show your passport, and the ¥5,000 (pre-tax) per-store minimum is unchanged from the old system. If you want the refund to a card or e-wallet, register on that operator's website or app first.
Around 200 confirmation terminals at roughly 36 international airports — or do it in advance on Visit Japan Web (menzei.nta.go.jp) over the airport customs Wi-Fi and skip the queue.
After confirmation, the store's refund operator pays to your chosen channel. There are 12 operators nationwide, each with its own registration and app — currently the biggest complaint from travelers.