How to Localize a Product Title for Coupang Search (Beyond Translation)

Why a translated title almost never ranks

A translation answers the question "what is this product called?" A Coupang title has to answer a different question: "what does a Korean shopper type when they want this product?" Those two answers overlap less than you'd expect.

Coupang's search matches on tokens inside your title, plus category, attributes, and behavioural signals (click-through, conversion, delivery speed). If the exact string a shopper types never appears in your title, you are relying on Coupang to infer the connection — and it often won't, especially for a new listing with no sales history to learn from.

The classic failure: a US seller lists a "Stainless Steel Insulated Tumbler 500ml" and a translator returns 스테인리스 단열 텀블러 (stainless insulated tumbler). Correct Korean. But Korean shoppers overwhelmingly search 보온병 (thermos/vacuum flask) or 텀블러 with modifiers like 대용량 (large capacity) and 빨대 (straw). The title is accurate and invisible.

• Translation optimises for accuracy; localisation optimises for retrieval.
• New listings have no behavioural data, so the title carries almost all the matching weight early on.
• Being "technically correct" in Korean and being searched are unrelated properties.

Step 1: Build the demand-side vocabulary, not the product-side vocabulary

Before writing anything, collect the words buyers use. Start from the top sellers in your exact Coupang category — not the whole category, the leaf category — and pull apart their titles token by token. Patterns repeat fast. If eight of the top ten include 무선 (wireless) and 충전식 (rechargeable), those are demand words, not decoration.

Then split your candidate keywords into three buckets: the head term (what the product fundamentally is), attribute modifiers (size, material, colour, count, capacity), and use-case modifiers (누구/어디/언제 — for babies, for camping, for the office). Korean shoppers modify heavily; a bare head term is a low-intent query with brutal competition.

This is the step where not reading Korean hurts most, because you can't feel which of two near-synonyms carries volume. Tools that surface Korean search volume with English glosses — SeoulRank is built for exactly this gap — let you compare 텀블러 vs 보온병 vs 물통 side by side instead of guessing from a dictionary.

• Mine the top 20 listings in your leaf category and tally repeated tokens.
• Check Coupang's own autocomplete and the related-search chips under results — they are demand signals, free.
• Separate head term / attribute modifiers / use-case modifiers before drafting.
• Note which competitors are 로켓배송 (Rocket Delivery); their titles reflect what already converts at scale.

Step 2: Choose between the loanword and the native word

Korean frequently has two live options for the same concept: a Sino-Korean or native word, and an English loanword written in Hangul. 냉장고 vs 리프리저레이터 is an easy call, but most cases are genuinely contested — 손목시계 (wristwatch) vs 워치, 가방 vs 백 (bag), 이불 (bedding/duvet) vs 컴포터.

The rule of thumb: fashion, beauty, tech accessories, and fitness lean loanword; household goods, food, baby, and traditional categories lean native. But the rule of thumb is not evidence. Check volume for both, and when they're close, include both — Korean titles routinely carry a synonym pair because it costs a few characters and doubles match surface.

Also watch romanisation variance. English brand and material names get transliterated inconsistently (실리콘 vs 씰리콘 for silicone). Use the spelling the market uses, not the one that's phonetically defensible.

• Test loanword vs native word for volume before committing; don't assume.
• If both terms have real volume and the title has room, include both.
• Match the market's transliteration spelling even when it looks 'wrong'.
• Keep the Latin-alphabet form of your brand and model number in the title too — shoppers do type those.

Step 3: Structure and word order

The workable Coupang title pattern is: brand → key modifier(s) → head noun → spec/size/quantity → colour or variant. So: 브랜드 + 대용량 + 스테인리스 + 텀블러 + 500ml + 블랙. Front-load the terms you most want to rank for; the head noun should never be buried at the end.

Keep the title readable by a human, because click-through rate feeds back into ranking. A wall of thirty comma-free keywords looks like spam to shoppers and often gets suppressed or trimmed by Coupang's own display anyway. Aim for a title a Korean person would read out loud without wincing.

Spacing (띄어쓰기, word spacing) matters more than most foreign sellers realise. Korean search tokenises around spaces, so 무선청소기 and 무선 청소기 can behave differently. Where a compound is commonly written both ways, use the spaced form for the primary term and let the compound appear naturally elsewhere in the listing.

• Head noun in the first half of the title, never trailing.
• Include the numeric spec exactly as shoppers type it: 500ml, 2개입, 30cm — not spelled out.
• Don't repeat the same token three times; repetition doesn't stack and reads as spam.
• Avoid superlatives and unverifiable claims (최고, 1위) — they invite listing sanctions.
• Keep option-level differences (colour, size) in the option, not stuffed into the parent title.

Step 4: Cover the search variants you can't guess

Every category has vocabulary that no dictionary will hand you: abbreviations, community slang, and portmanteaus. Korean shoppers clip words aggressively (아이폰 케이스 → 아케 in some contexts) and coin compounds around use cases (차박 for car camping, 홈트 for home training). Missing one of these can mean missing an entire buying segment.

Seasonal and event vocabulary is the other blind spot. Terms tied to 설날 (Lunar New Year), 추석 (Chuseok harvest holiday), 어린이날 (Children's Day), and gift-set framing (선물세트) spike hard and predictably. If your product is giftable, having 선물 or 선물세트 in the title during those windows is often the single highest-leverage edit you can make.

Your Q&A section and product reviews are a free corpus. Run them through translation and note the nouns customers actually use to describe your item — that language is more reliable than anything you'd pick from outside.

• Harvest vocabulary from your own reviews and buyer questions.
• Map the gifting calendar for your category and pre-edit titles two to three weeks ahead.
• Watch for use-case compounds (차박, 홈트, 캠핑) that define a whole shopper segment.
• Check whether your category has a dominant abbreviation before you write a formal-sounding title.

Step 5: Change one thing, then measure

Title edits on Coupang are not free. Rewriting a converting listing can reset the signals it has accumulated, so treat titles like a controlled experiment: change one variable — swap the head noun, add one modifier, adjust spacing — and hold everything else steady for at least a week or two.

What you're measuring is rank position for your target keywords, not just sales. Sales move for a dozen reasons; rank on a specific query tells you whether the title change actually improved matching. Track a fixed set of 15–30 target keywords daily so you can see a change land within days rather than inferring it from noisy revenue data a month later. Daily Coupang rank tracking in English is the whole reason SeoulRank exists, but any consistent method beats checking manually and remembering wrong.

Two guardrails. First, don't edit during a sales spike or a promotion — you'll never separate the effects. Second, keep a changelog with the date, the exact old and new title, and the keywords you were targeting. Six months in, that log is the most valuable Korean-market asset you own.

• One variable per test; document the before-and-after string exactly.
• Judge outcomes on keyword rank first, revenue second.
• Give each test 7–14 days before deciding.
• Never test on a listing that's mid-promotion.
• Roll winning patterns across your catalogue rather than re-testing every SKU from scratch.