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

Why translated titles underperform on Coupang

A Coupang title is not a sentence. It is a keyword container with a strict reading order, and Korean shoppers scan it in roughly a second. When you run an Amazon-style English title through machine translation, you get grammatically correct Korean that contains almost none of the words buyers actually type.

The gap is usually one of three kinds: the translator picks a dictionary word where shoppers use an English loanword, it picks a loanword where shoppers use the native Korean word, or it produces a phrase that is technically right but has essentially zero search volume. Coupang's search engine matches tokens in your title against the query; if the token isn't there, you are invisible regardless of price or reviews.

Localization means rebuilding the title from Korean demand upward, not translating downward from English.

• Translation optimizes for meaning; localization optimizes for the exact strings buyers type.
• Coupang shows a truncated title in grid view — the first ~25-30 characters carry most of the click decision.
• A title that ranks and a title that converts are not always the same string; you are balancing both.

Step 1: Build a Korean keyword set before you write anything

Start from how Koreans name the category, not from your English product name. For almost every product there is a 대표 키워드 (daepyo keyword — the 'head' category term), a set of 세부 키워드 (sebu keyword — specific/long-tail modifiers), and a set of 속성 (sokseong — attribute) words like material, size, color, and use case.

Collect candidates from Coupang's own autocomplete, the related-search chips under results, the 'best' category pages, and the titles of the top 10-20 listings for your category. Top sellers have already done thousands of dollars of testing; their shared vocabulary is your shortlist. Then validate volume — this is where a Korean keyword tool that shows English glosses, like SeoulRank, saves you from guessing which of five synonyms is the real one.

Watch especially for synonym pairs where only one has traffic. 물병 vs 텀블러 vs 보틀 (water bottle / tumbler / bottle) are three different demand pools for overlapping products, and the right pick can differ by subcategory.

• Pull 3-5 head terms, 10-20 long-tail terms, and every attribute word you can support honestly.
• Record each keyword's volume and your realistic chance of ranking, not volume alone.
• Note which keywords appear in competitor titles 8+ times out of 10 — that's a near-mandatory token.
• Check seasonal terms (여름 summer, 겨울 winter, 선물 gift) that spike predictably.

Step 2: Choose loanword vs. native Korean the way shoppers do

Korean has two vocabulary layers for many product concepts: a Sino-Korean or native word, and a Konglish loanword written in Hangul. Which one wins is a matter of category convention, not logic.

Fashion, beauty, and tech skew heavily toward loanwords: 케이스 (case), 클렌징 (cleansing), 니트 (knit), 백팩 (backpack). Kitchen, household, and health goods often keep native or Sino-Korean terms: 수세미 (scrubber), 건조대 (drying rack), 영양제 (supplement). Machine translation routinely picks the wrong layer.

Also decide your romanization policy. Many brands are searched in Latin letters, in Hangul transliteration, or both — 나이키 and Nike both get typed. If your brand has any Korean recognition, include both forms once each.

• Test both forms in Coupang autocomplete; the suggested form is the dominant one.
• Never include both synonyms for the same concept unless both have real volume — it eats characters.
• Transliterate your brand into Hangul and keep that spelling consistent across every listing and ad.
• Avoid inventing a Konglish term; if autocomplete doesn't know it, nobody searches it.

Step 3: Order and spacing — the mechanics that quietly cost rank

The conventional Coupang title order is roughly: brand → product line/model → core category keyword → key attributes → quantity/option. Putting the category keyword too late pushes it out of the visible truncated title on mobile and weakens relevance signals.

Korean spacing (띄어쓰기, ttuieossugi — word spacing) matters more than sellers expect. Search tokenization treats 무선이어폰 and 무선 이어폰 (wireless earbuds) somewhat differently, and shoppers type both. The safe approach is to use the spacing that autocomplete returns for your main term, and let a secondary phrase elsewhere in the title cover the other variant naturally.

Avoid special characters as separators beyond simple spaces. Brackets, slashes, and emoji add noise, sometimes break tokenization, and can trip category policy checks.

• Front-load the single keyword you most want to rank for, right after the brand.
• Keep the title readable — a comma-less keyword dump suppresses click-through even when it ranks.
• Don't repeat the same keyword; repetition does not increase weight and wastes visible space.
• Put size, count, and color at the end where shoppers expect them (e.g. 2개입 'pack of 2').

Step 4: Handle options, counts, and units the Korean way

Units and counters are a frequent source of silent mismatch. Koreans search 2개 or 2개입 for a two-pack, use 매 for sheets, 장 for flat items, 켤레 for pairs of shoes, and 세트 for sets. A title saying '2 pack' in English or '2 팩' captures far less traffic than 2개입.

Sizing is another trap. Clothing often uses both letter sizes and numeric Korean sizing; shoes use mm (250, 270) rather than US or EU numbers. If you only list EU 42, a large share of shoppers will never match your listing. Put the Korean-standard unit first and the foreign equivalent in the option name or detail page.

• Shoes: lead with mm sizing. Apparel: include the Korean numeric size alongside S/M/L.
• Volume in ml, weight in g/kg — avoid oz and lb entirely in titles.
• Use 입 / 개입 for pack counts; it reads as native and matches search behavior.
• Keep option names short and consistent — they appear in filtered search results.

Step 5: Test, measure, and iterate on real rank data

Treat your title as a hypothesis. Launch with your best keyword set, then track daily position for each target keyword rather than relying on a single overall sales number. Rank movement tells you which token is working; revenue alone does not.

Change one variable at a time — swap a synonym, move the head keyword forward, or add an attribute word — and give it one to two weeks before judging. Coupang's ranking responds to click-through and conversion over time, so same-day readings are noise. Daily Coupang rank tracking in English (SeoulRank does this) makes the signal legible without reading Korean dashboards.

Keep a simple log: date, exact title text, target keywords, and starting rank. After three or four cycles you'll have a category-specific playbook that beats any generic translation workflow.

• Track 5-10 keywords per SKU, mixing one head term with several long-tail terms.
• Expect long-tail wins first; head-term rank usually follows review volume.
• Re-check autocomplete quarterly — Korean search vocabulary shifts with trends faster than English.
• If rank is good but clicks are weak, the problem is readability, not keywords.