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

Why translated titles underperform on Coupang

A translated title answers the question "what is this product called in Korean?" A localized title answers a different question: "what would a Korean shopper type into the 쿠팡 (Coupang) search bar to find this, and what would make them tap it over the nine listings above it?" Those are rarely the same string of words.

Coupang's search matches your title tokens against the query, then ranks on behavioral signals — click-through rate, conversion rate, sales velocity. So a title has two jobs at once: contain the exact query terms shoppers use, and read like a native listing so it earns the click. Machine translation reliably fails the first job and often fails the second.

The most common symptom: your listing is technically indexed, gets impressions on a long-tail query nobody searches, and never accumulates the sales signal needed to surface on the head term.

• Translation optimizes for accuracy; localization optimizes for retrieval and clicks.
• If your title contains zero high-volume Korean queries, no amount of ad spend fixes it.
• Korean shoppers frequently search in a category + attribute + use-case pattern, not brand-first.

Start from real queries, not from your English title

Do not translate and then check volume. Pull the Korean queries first, then build a title that contains the winners. For a stainless water bottle, candidate head terms might include 텀블러 (tumbler), 보온병 (thermos/insulated bottle), 물병 (water bottle), and 스테인리스 텀블러 (stainless tumbler). These are not synonyms in shopper intent — 텀블러 skews to daily-carry café culture, 보온병 skews to hot-liquid retention and outdoor use. Choosing wrong puts you in the wrong competitive set.

Check what the top-ranking listings for each term actually are. If page one of 보온병 is all 1L+ camping flasks and your product is a 500ml desk tumbler, you will lose on conversion even if you rank.

This is the step where an English-language Korean keyword tool matters — SeoulRank exists so you can see volume, competition, and the English gloss for each candidate without guessing at Hangul you can't read. Whatever tool you use, the rule is the same: pick the head term by data, not by dictionary.

• Collect 20–40 candidate Korean terms per product before writing anything.
• Separate head terms (category) from modifiers (material, size, use case, target user).
• Eyeball page one for each head term to confirm intent match.
• Note which terms your direct competitors are ranking for — that's your realistic target set.

Konglish, pure Korean, and Latin script: know which one shoppers type

Korean product vocabulary is layered. Many categories have a loanword (Konglish) form, a native Korean form, and sometimes an English-script form, and shoppers do not use them interchangeably. 노트북 (notebook = laptop) dominates over any native alternative. But 냄비 (pot) beats a transliterated "팟". For beauty, 세럼 (serum) and 앰플 (ampoule) coexist as distinct product types, not translations of each other.

Latin script does get typed — brand names especially. Korean shoppers often search "Nike 운동화" or the Hangul transliteration 나이키. Safe practice for a recognizable brand is to include both scripts, since Coupang won't reliably bridge them for you.

When in doubt, the transliteration convention used by the market leader in your category is the convention. Copy it. Inventing a novel transliteration of your brand name guarantees zero matches.

• Include your brand in Hangul and Latin script if both are plausible queries.
• Never invent a transliteration — check how Korean media or existing sellers already spell it.
• Watch for near-homographs: 세럼/앰플/에센스 are different product classes to a Korean shopper.
• Some categories use English abbreviations natively (USB, LED, PD, ANC) — leave those in Latin script.

Structural conventions Korean shoppers expect

Korean titles have a recognizable rhythm: brand → product line/model → key attribute → variant (color/size/count) → category noun. The category noun typically lands near the end and is the token most likely to be the head query. Front-loading the category noun instead is acceptable, but burying it after a long spec dump is not.

Spacing matters more than beginners expect. Korean has 띄어쓰기 (word spacing) rules, and search behavior splits along them. Shoppers type both spaced and unspaced forms of compound terms, so hedging by including the spaced form in the title while a variant appears elsewhere in your listing data is reasonable. Don't jam everything together into one unreadable string — that reads as spam and hurts CTR.

Units and sizing follow local convention. Volume in ml/L, weight in g/kg, length in cm, and multi-packs written as 2개입 or 3개입 (2-pack / 3-pack). Apparel sizing may need both the letter size and a Korean numeric equivalent depending on category.

• Keep the title readable — Coupang truncates on mobile, so the first ~25–30 characters carry the click.
• Use 개입 / 세트 (set) conventions for bundles instead of "x2".
• Metric only. A title with inches or ounces signals a lazy import listing.
• Avoid stacking the same head term three times; repetition doesn't compound and looks like spam.
• Put the strongest differentiating modifier before the category noun, not after.

The mistakes machine translation makes every time

Even good MT output fails in predictable ways on product titles, because titles are noun stacks with no grammatical context to disambiguate. Reviewing for these specific failure modes catches most damage without needing to read Korean fluently — you can spot-check by searching each translated term on Coupang and seeing whether the results match your product.

That reverse-check is the single highest-value QA step available to a non-Korean speaker: paste your candidate term into Coupang search. If the top results aren't your product category, the word is wrong.

• Over-formal or literary register: translated titles often read like a manual, not a listing.
• Wrong sense of a polysemous English word ("case", "light", "set", "stand").
• Marketing adjectives translated literally — "premium quality" renders as filler that no one searches.
• Grammatical particles left in, which break token matching; Korean titles are largely particle-free.
• Missing the actual category noun entirely because MT chose a descriptive phrase instead.

Test, measure, and iterate on the title

Treat the title as a variable, not a one-time deliverable. Launch with your best data-backed version, then track daily rank on the three to five target keywords you actually care about. Rank movement plus CTR tells you whether the problem is retrieval (you don't appear) or persuasion (you appear and nobody clicks).

Change one thing at a time and give each version enough days to accumulate signal — swapping the head term and the modifier order simultaneously teaches you nothing. Daily rank tracking in English makes this loop workable when you can't read the SERP yourself; that's the core of what SeoulRank does, but the discipline matters more than the tool.

Revisit titles seasonally. Korean search vocabulary shifts with trends, and a modifier that was dead in March can become a high-volume query by June.

• Baseline your rank on target keywords before any edit.
• One variable per test; 7–14 days minimum before judging.
• Not ranking at all = keyword missing or wrong category. Ranking but no clicks = title reads foreign or thumbnail mismatch.
• Log every title version with its date so you can roll back.