Why translated copy underperforms in Japan

Localization for Japan is the rewriting of commercial copy so that it answers the questions a Japanese buyer asks, in the order they ask them, with the proof they need to stop asking. Translation is the conversion of the words that your existing argument is currently made in. These are different jobs, they cost different amounts, and only one of them changes a conversion rate. If you have a translation invoice, a Japanese-language site, arriving traffic and an order count that does not move, you have almost certainly bought the second job and been billed for it as if it were the first.

What is localization for Japan?

Localization for Japan is a rewrite of what a page argues, in what order, and with what evidence. Translation is a faithful conversion of the argument you already made for a different market, which is why a technically flawless Japanese page can still fail.

The failure is structural. An English product page typically opens with a positioning line, moves to two or three benefits, adds a lifestyle image, and puts specifications in a collapsed tab near the bottom. That sequence assumes the reader is deciding whether to want the product. A Japanese page that converts usually assumes the reader has already decided they want something in the category and is now deciding which seller to eliminate. It front-loads dimensions in centimetres, material composition, model number (型番), country of manufacture, what is in the box, delivery window, who pays import duty, and the returns term in days. Persuasion comes after that, if at all.

So the rewrite touches six layers, and a per-character translation quote touches one of them.

Specification detail that reads as excessive in English is doing work in Japanese. It is not padding and it is not a cultural preference for thoroughness. It is the removal of the buyer's remaining reasons to hesitate, one line at a time.

How do Japanese buyers decide?

Across most consumer categories, the decision runs as elimination rather than attraction: a shortlist is assembled quickly, then sellers are struck off for missing information. Completeness of pre-purchase information and speed of response outrank brand storytelling for the majority of segments — a statement that is counterintuitive to most Western marketers and worth testing on your own pages before dismissing.

Tendencies differ sharply by segment, so describe yours rather than the country:

• Rakuten Ichiba regulars — heavy users of the points economy, skewing older and more regional. They read shop-level reviews as much as product reviews, and they check shipping cost and payment options before they check the product.
• Amazon.co.jp buyers — decide on delivery-date certainty and the return path. Ambiguity about who handles a failed delivery is fatal here in a way it is not elsewhere.
• Enthusiast categories (audio, outdoor, tools, camera) — will read a very long page willingly, compare specification tables across sellers, and treat a missing figure as a defect in the seller.
• Beauty and apparel buyers discovered through Instagram or LINE, roughly 20 to 34 and metro-weighted — need Japanese-language review volume and sizing expressed the Japanese way (着丈, 身幅, 肩幅, 袖丈; shoes in cm; rings in 号).
• B2B procurement staff at small and mid-size manufacturers — look for a company overview page (会社概要), a contactable phone number, and documentation their accounts department can file. Their internal process, not their preference, drives this.

If part of your audience is Japanese travellers buying from you while abroad rather than importing from you, the discovery and decision path is different and covered separately in how Japanese travellers plan and book a trip and which channels actually reach Japanese travellers abroad. Selling goods into Japan and selling to visitors outside it are separate builds; do not merge the copy.

What proof does a Japanese buyer need before ordering?

Enough proof that no unanswered question remains at the point of clicking buy. In practice that means seller identity, physical evidence of the product, and explicit terms — in Japanese, on the page, not behind a link.

The proof set that most often has to be added rather than translated:

• Seller identity. A company overview page carrying legal entity name, address, and a contact route. Japanese distance-selling rules generally require a specified disclosure page (based on 特定商取引法, the Act on Specified Commercial Transactions); whether and how it applies to a seller based outside Japan is something to confirm with a Japanese lawyer, not something to copy from a competitor's footer.
• Dimensional evidence. Diagrams with measurements in cm, a scale shot against a recognisable object, box dimensions, and assembled weight. For furniture and interiors, room-size context in 畳 or 帖 does more than a styled lifestyle photo.
• Landed cost clarity. Whether duty and consumption tax are collected at checkout or on delivery. Vagueness here produces checkout abandonment that looks like a payment problem in your analytics. Your import and tax treatment must be confirmed by a licensed Japanese tax professional and a customs broker; publishing a guess is worse than publishing nothing.
• Returns in numbers. How many days, who pays return shipping, what condition is required. "Hassle-free returns" translates into Japanese as a sentence that commits to nothing.
• Japanese-language reviews. Review volume in English does not substitute. For newly launched pages, a Q&A block written from real inquiries carries part of the same load.
• Payment options by name. クレジットカード, コンビニ決済, 代金引換, 後払い, carrier billing, and the wallets you actually accept. Availability of some of these to overseas merchants varies by provider; verify before you list them.

What does customer service have to look like?

It has to be visible before the order, fast after it, and written in a consistent register. Among most consumer segments, a reply on the same business day is the working expectation; for B2B procurement, a reply by the next business morning is usually sufficient if it is complete.

Three things go wrong in translated setups, and none of them are about politeness in the abstract.

First, keigo drift. Marketing pages come back in a breezy consumer register, the FAQ in a stiff machine-generated formal register, and transactional email in whatever the platform shipped by default. The buyer reads this as three different companies. Pick one register — です・ます with light honorifics is safe for most direct-to-consumer selling — and enforce it across pages, email, packing inserts and support replies.

Second, untranslated machinery. The homepage is localized; the checkout error states, address form labels, furigana field, postcode auto-fill, shipping-confirmation email and refund notice are not. These are the surfaces a hesitant buyer meets at the moment of highest doubt.

Third, an unpublished response window. State your hours in Japan Standard Time, state what happens at weekends, and state a response window you can beat from your own time zone. A published twenty-four-hour window that is always met outperforms an unstated ambition to reply immediately.

How much does proper localization cost?

Budget by page value, not by word count. Per-character translation pricing rewards brevity, which is precisely the wrong incentive in a market where longer, denser pages tend to convert better — so the pricing model itself quietly works against you.

A workable structure for localization for Japan on a catalogue of any size:

• Tier one: the money pages. Your top ten to thirty SKUs by traffic, plus shipping, returns, company overview and contact. These are rewritten by someone who can argue in Japanese, not translated. This is where the real spend belongs.
• Tier two: the long tail. Structured specification templates, filled from your product database, in correct Japanese. Consistent and complete beats individually crafted here, and it can be generated rather than written.
• Tier three: transactional surfaces. Email templates, form labels, error states, support macros. Small volume, disproportionate effect.

Marketplace costs sit alongside this and change the maths. Rakuten Ichiba, Amazon.co.jp and Yahoo! Shopping each charge some combination of commission, plan fee and optional advertising, and each publishes its own current schedule; treat any figure you read in an article, including ranges, as something to verify against the operator's published terms on the day you plan.

For reference, AiBou's indicative bands: a focused fix from $1,800, typically three to five days, which suits a rewrite of a small set of high-traffic pages plus the transactional surfaces behind them; systems design and build from $9,000, typically three to four weeks, where the catalogue is large enough that the long tail has to be generated and maintained rather than hand-written; fractional DX lead from $1,800 a month with a three-month minimum, for ongoing operation. Contracts are denominated in Japanese yen at the exchange rate fixed on the signing date. These reflect a Tokyo cost base, not a discount.

Can you launch with translated copy and fix it later?

Yes on your own site, cautiously; no on marketplaces. The difference is whether the first version of your copy leaves a permanent record.

On your own domain, a translated launch is a legitimate diagnostic. You are buying data: which pages get traffic, where scroll depth collapses, which step of checkout loses people, what inquiries arrive. Keep paid traffic small while you do it, because you are paying full price for clicks onto a page you already expect to underperform. Instrument before you spend — add-to-cart rate by page, checkout step drop-off, inquiry-to-order ratio, and mobile share, which in most Japanese consumer segments is the majority of sessions.

On Rakuten Ichiba, Amazon.co.jp or Yahoo! Shopping, the calculation changes. Early conversion and early reviews feed ranking, and a listing that accumulates poor ratings, high return rates or unanswered questions carries that record forward. Recovering a damaged listing costs more than rewriting it would have. If you are opening a marketplace storefront, rewrite the listings first; there is no cheap correction later.

What usually goes wrong?

The same eight failures account for most of the cases where a translated website in Japan draws traffic and takes no orders. None of them are language errors.

1. The brief was "translate this." The agency did exactly that, correctly, and had no mandate to change the argument. The output is not defective; the specification was.
2. Pages are too short. Source pages were written for a market that rewards economy. Nothing in the translation process adds the specification density that removes hesitation.
3. The brand story sits at the top. Founding year, values, craft narrative — above the fold, where the specification table should be. Move it below the proof. Keep it; do not lead with it.
4. Units and sizing left as-is. Inches, pounds, US or EU apparel sizes, MM/DD dates. Each one asks the buyer to do work, and some will not.
5. Text baked into images. Hero banners and infographics in English, unsearchable and unmodifiable, with Japanese text set without line-break control (禁則処理), producing breaks that look careless.
6. Landed cost left ambiguous. Duty and consumption tax treatment implied rather than stated. Verify your position with a licensed Japanese tax professional and a customs broker, then state it plainly.
7. Support surfaces untouched. Confirmation emails, error states, refund notices and packing inserts still in English after the marketing pages were done.
8. No measurement. Without page-level funnel data you cannot tell a copy problem from a shipping-cost problem from a payment-option problem, and you will rewrite the wrong thing.

A useful test if you do not read Japanese: give your Japanese page to a Japanese-speaking colleague or contractor and ask them to list every question they would still have before ordering. If the list runs past three items, the page is not finished. The exercise costs an hour and is more diagnostic than most audits.

How AiBou approaches localization for Japan

AiBou is worth considering here when the problem is that nobody in your current chain both knows what a Japanese buyer needs to see and can change the page. A translator supplies language, a consultant supplies a recommendation, and the change waits on a developer; AiBou is one operator holding the commercial judgment and the build.

This work sits in the knowledge gap — what Japanese buyers expect, which no translation supplies — and in the handoff gap between whoever advises and whoever implements. The background behind it is four and a half years at Japan's largest e-commerce support firm, running marketplace strategy and advertising for more than thirty manufacturer brands, and nine years in Japanese commerce and retail operations.

Catalogue-scale work needs tooling rather than headcount. In one e-commerce engagement I built a pricing and promotion audit across roughly 2,000 products that reduced an eight-hour manual pass to ten minutes; the same approach applies to generating and maintaining specification content across a long tail.

Indicative rates: a focused fix from $1,800 over three to five days for a defined set of pages and the transactional surfaces behind them; systems design and build from $9,000 over three to four weeks where the catalogue needs generated content and ongoing maintenance.

A first conversation covers which pages carry your traffic, what the current Japanese copy does with it, and which failures are structural rather than linguistic. It costs nothing: info@aibou365.jp.

Frequently asked questions

Is machine translation ever acceptable for the Japanese market?

For internal documents, supplier email and low-stakes help content, yes. For product pages, checkout copy and anything a buyer reads while deciding, no — not because the grammar fails, but because machine translation preserves an argument built for a different buyer. The problem is the source, not the engine.

Should Japanese product pages be longer than the English originals?

In most consumer and enthusiast categories, yes. The added length should be specification, dimensional evidence, delivery and returns terms, and answers to real inquiries — not more prose. Length that is not doing decision work is just length.

Do I need a Japanese address and phone number to sell into Japan?

Buyers in several segments, particularly B2B procurement and older marketplace regulars, look for them, and their absence costs conversions. Whether either is legally required for your model depends on your structure and what you sell; confirm that with a Japanese lawyer before deciding how to present it.

How many SKUs should I rewrite before I judge the result?

Enough to carry a readable share of your traffic — usually the top ten to thirty products, plus shipping, returns and company pages. Rewrite those, leave the rest untouched, and compare. If the rewritten set does not move, the problem is upstream of the copy.

Should I hire a Japanese copywriter or a translation agency?

Neither alone solves it. A copywriter who has never operated a Japanese storefront will write elegant copy in the wrong order; a translation agency will convert your order faithfully. What localization for Japan requires is someone who can specify the page structure and proof set first, then have it written.

Does a .jp domain improve conversion?

It can help with trust among some segments, but it is well down the list against review volume, delivery certainty and a visible company page. Registration of .jp addresses generally involves a local presence requirement; check current rules with an accredited registrar before planning around it.

How do I tell if my Japanese copy is bad if I cannot read Japanese?

Ask a Japanese-speaking reader to list every question they would still have before ordering, and check whether the register is consistent across pages, email and support replies. Then look at your own funnel data by page. Between those two, you will find the failure without reading a word.