Why your backend search terms are being cut off

A character is not a byte

In UTF-8, which is what Amazon stores, a plain ASCII letter costs one byte. Anything with an accent costs two.

That means a string of 250 characters can be anywhere from 250 to well over 400 bytes depending on the language it is written in. English sellers never meet this problem. Sellers writing French, Spanish, Italian, Swedish or Polish meet it constantly, and usually without knowing.

Counted at build time with the same encoder Amazon stores with, not by hand.

Why nobody notices

There is no error. That is the whole problem.

How to count it yourself

Spreadsheets count characters, which is why this is easy to miss.

LEN() in Excel and Google Sheets counts characters. There is no built-in byte count in either. The quickest honest check without leaving your browser is the developer console on any page:

new TextEncoder().encode("your search terms here").length

If that number is over your field budget, the tail is being dropped. Shorten by removing whole words from the end rather than trimming characters, so you know exactly which term you gave up.

While you are in there

Two other search term habits that waste budget you now know is tighter than it looked.

• Do not repeat the titleWords already in your title, bullets or brand are indexed. Repeating them in search terms spends bytes on something you already have.
• Do not repeat yourselfSingular and plural, and word order, do not need separate entries. One occurrence of a word is enough; the byte budget is better spent on a word you have not used yet.
• No commas neededSeparate terms with spaces. Punctuation spends bytes without adding meaning.

How Utlista handles it

Briefly.

Search terms are measured and trimmed by byte rather than by character, and the cut is made between characters so a half-encoded letter can never reach a live listing. Terms over budget block the export and are shown to you with the byte count, rather than being quietly truncated the way Amazon would do it.