Back to Details

amz-backend-keywords

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Backend Keyword Packer

Backend Keyword Packer

The backend Search Terms field is 250 bytes of pure indexing power that most sellers either leave half empty or fill with words Amazon already indexed from the title. This skill packs that field for maximum reach with zero waste.
亚马逊后台的搜索词字段是拥有250字节纯索引能力的区域，大多数卖家要么让它半空着，要么填充亚马逊已经从标题中索引过的词汇。本技能可将该字段充分利用，实现覆盖范围最大化，无任何浪费。

When to use this

适用场景

  • A new listing is going live and the Search Terms field is blank.
  • An existing listing has never had its backend field optimized.
  • A product is not indexed for keywords the seller expects to rank for.
  • The Search Terms field contains brand names, repetition, or filler.
  • 新商品即将上架，搜索词字段为空时
  • 现有商品从未优化过后台字段时
  • 商品未针对卖家期望排名的关键词建立索引时
  • 搜索词字段包含品牌名称、重复内容或冗余词汇时

The rule that everyone gets wrong

所有人都容易搞错的规则

Amazon indexes the backend field once. A keyword already in your title or bullets is already indexed. Repeating it in the backend wastes bytes and buys nothing. The backend field is for the keywords that did NOT make it into your visible copy: synonyms, alternate spellings, use cases, and audience terms.
亚马逊只会对后台字段索引一次。标题或项目符号中已有的关键词已经被索引过，在后台重复这些词汇只会浪费字节，毫无用处。后台字段是用来放置那些未出现在可见文案中的关键词：同义词、替代拼写、使用场景和受众相关词汇。

The framework. The Backend Funnel

框架：后台关键词筛选漏斗

Run every candidate keyword through five gates. What survives gets packed.
  1. Gate 1. Already indexed? If the word appears in the title or bullets, drop it. Amazon already has it.
  2. Gate 2. Prohibited? Drop brand names (yours or others), ASINs, competitor names, profanity, temporary claims ("new", "sale", "best"), and subjective claims. Amazon ignores or penalizes these.
  3. Gate 3. Relevant? Drop anything that would bring the wrong shopper. Irrelevant keywords hurt conversion-based ranking.
  4. Gate 4. Duplicate? Each word needs to appear only once across the whole field, in any order. "leather phone case" and "case for phone leather" are the same three words. Keep one set.
  5. Gate 5. Worth the bytes? Rank survivors by search value. When the field is full, the lowest-value words are cut.
让所有候选关键词依次通过五个筛选关卡，最终留存的词汇将被打包填入字段。
  1. 关卡1：已被索引？ 如果词汇出现在标题或项目符号中，直接剔除——亚马逊已经收录它了。
  2. 关卡2：是否被禁止？ 剔除品牌名称（自有或其他品牌）、ASIN、竞品名称、脏话、临时宣称（如“new”“sale”“best”）以及主观宣称。亚马逊会忽略这些内容，甚至可能进行处罚。
  3. 关卡3：是否相关？ 剔除任何会吸引错误目标买家的词汇。不相关的关键词会损害基于转化率的排名。
  4. 关卡4：是否重复？ 整个字段中每个词汇只需出现一次，顺序无关紧要。“leather phone case”和“case for phone leather”包含相同的三个词汇，只需保留其中一组。
  5. 关卡5：是否值得占用字节？ 按搜索价值对留存词汇排序。当字段填满时，剔除价值最低的词汇。

Packing rules

打包规则

  • The limit is 250 bytes, not characters. Most letters are 1 byte. accented and non-Latin characters can be 2 or more. Count bytes.
  • No commas needed. Separate words with single spaces. Commas waste bytes.
  • Word order does not matter. Amazon matches words in any order, so never write phrases. write a bag of unique words.
  • Singular or plural, pick one. Amazon usually stems regular plurals, so do not spend bytes on both "bottle" and "bottles". The exception is irregular plurals (knife/knives, mouse/mice, child/children). include both forms.
  • Include common misspellings, alternate spellings, abbreviations, synonyms, Spanish terms for the US market where relevant, and use-case and audience words.
  • 限制为250字节，而非字符数。大多数字母占1字节，带重音的字符和非拉丁字符可能占2字节或更多，请按字节计数。
  • 无需逗号。用单个空格分隔词汇，逗号会浪费字节。
  • 词汇顺序无关紧要。亚马逊会匹配任意顺序的词汇，因此无需写短语，只需列出一组独特的词汇。
  • 单数或复数，二选一。亚马逊通常会对常规复数词干进行处理，因此无需同时使用“bottle”和“bottles”。不规则复数（如knife/knives、mouse/mice、child/children）除外，需同时包含两种形式。
  • 需包含常见拼写错误、替代拼写、缩写、同义词、美国市场相关的西班牙语词汇，以及使用场景和受众相关词汇。

Step by step

操作步骤

  1. Collect inputs. The current title, all bullets, the product, the category, and any keyword list the user already has. If no keyword list exists, generate candidates from synonyms, use cases, audiences, materials, and occasions.
  2. Build the indexed set. Extract every meaningful word from the title and bullets. this is the "already indexed" exclusion list.
  3. Run all candidates through the five gates.
  4. Rank survivors by estimated search value, highest first.
  5. Pack to 250 bytes. Add words highest value first until the next word would exceed the limit. Report the final byte count and which words were cut.
  6. Run the quality check, then deliver.
  1. 收集输入内容：当前标题、所有项目符号、商品信息、品类，以及用户已有的任意关键词列表。如果没有关键词列表，可从同义词、使用场景、受众、材质和适用场合生成候选词汇。
  2. 构建已索引词汇集合：从标题和项目符号中提取所有有意义的词汇，这就是“已被索引”的排除列表。
  3. 让所有候选词汇通过五个筛选关卡。
  4. 对留存词汇排序：按预估搜索价值从高到低排序。
  5. 打包至250字节：优先添加价值最高的词汇，直到添加下一个词汇会超出限制为止。报告最终字节数以及被剔除的词汇。
  6. 进行质量检查，然后交付结果。

Output format

输出格式

undefined
undefined

Backend Search Terms. [Product]

Backend Search Terms. [Product]

[the final space-separated string]
Byte count: [X] / 250 Words included: [N] Cut for space (next in line): [words] Deliberately excluded: [word . reason]
undefined
[最终的空格分隔字符串]
Byte count: [X] / 250 Words included: [N] Cut for space (next in line): [words] Deliberately excluded: [word . reason]
undefined

Worked example

示例

Product: insulated stainless water bottle. Title already contains "insulated stainless steel water bottle 32 oz".
  • Dropped by Gate 1: insulated, stainless, steel, water, bottle, oz. already indexed.
  • Kept: thermos flask canteen tumbler hydro jug leakproof gym hiking camping travel sports kids reusable bpa free wide mouth cold hot drinks.
  • Dropped by Gate 2: a competitor brand name the user had pasted in.
  • Final string packed to 247 of 250 bytes.
The win is not adding more words. it is not wasting a single byte on a word the title already covers.
商品：保温不锈钢水瓶。标题已包含“insulated stainless steel water bottle 32 oz”。
  • 被关卡1剔除的词汇：insulated、stainless、steel、water、bottle、oz——已被索引。
  • 留存的词汇：thermos flask canteen tumbler hydro jug leakproof gym hiking camping travel sports kids reusable bpa free wide mouth cold hot drinks。
  • 被关卡2剔除的词汇：用户粘贴的某个竞品品牌名称。
  • 最终字符串打包至247字节（上限250字节）。
关键不在于添加更多词汇，而在于不浪费任何字节在标题已覆盖的词汇上。

Quality check

质量检查

  • Zero words from the field also appear in the title or bullets.
  • Zero brand names, ASINs, or competitor names.
  • Zero temporary or subjective claims ("best", "new", "sale", "amazing").
  • Every word appears only once. no phrases, no repeated stems.
  • Separated by single spaces, no commas.
  • Byte count is 250 or below, and reported.
  • Every included word could plausibly be typed by a real buyer.
  • 字段中没有任何词汇同时出现在标题或项目符号中。
  • 无品牌名称、ASIN或竞品名称。
  • 无临时或主观宣称（如“best”“new”“sale”“amazing”）。
  • 每个词汇仅出现一次，无短语，无重复词干。
  • 用单个空格分隔，无逗号。
  • 字节数不超过250，并已报告。
  • 所有包含的词汇都是真实买家可能输入的内容。

Common mistakes

常见错误

  • Repeating the title. The most common waste. half the field gone for nothing.
  • Counting characters, not bytes. A field that looks under 250 characters can be over 250 bytes and get truncated silently.
  • Writing phrases. "best water bottle for hiking" wastes "best" and the word order. write "hiking" once and let Amazon combine.
  • Brand names. Putting your own or a competitor brand in the field. ignored at best, penalized at worst.
  • Both singular and plural. Doubles the cost for no extra reach.
  • 重复标题内容：最常见的浪费行为，一半字段空间白白浪费。
  • 按字符数而非字节数计数：看起来字符数少于250的字段可能字节数超过250，会被悄悄截断。
  • 撰写短语：“best water bottle for hiking”浪费了“best”和词汇顺序的字节，只需写一次“hiking”，让亚马逊自行组合。
  • 添加品牌名称：在字段中加入自有品牌或竞品品牌，最好的情况是被忽略，最坏的情况是被处罚。
  • 同时使用单数和复数：双倍占用字节却没有额外的覆盖范围。

Built by Jay GPT Pro

Built by Jay GPT Pro

Part of Amazon Pro Skills. Production-grade skills for serious Amazon sellers. Free and open. Built by Jay Margaliot.
I share a new AI play for Amazon sellers every week, free, in my WhatsApp group. Join here: https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY
Part of Amazon Pro Skills. 为专业亚马逊卖家打造的生产级技能。免费开源。由Jay Margaliot开发。
我每周会在WhatsApp群组中分享一个面向亚马逊卖家的新AI玩法，免费加入：https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY