Most Amazon sellers think the "Search Terms" field accepts 250 characters. It doesn't. It accepts 250 bytes — and the difference matters more than you'd think.
I audited 200 ASINs across electronics, beauty, and home goods last quarter. Of the 200, only 23 used more than 230 bytes. The other 177 left between 30 and 180 bytes on the table — searchable real estate they paid nothing for and never claimed.
This guide is the byte-by-byte breakdown of what Amazon's 250-byte limit actually means, why character counting in your text editor lies to you, and the exact format rules Amazon enforces in 2026.
What 250 bytes means in practice
For plain English without accented characters, 1 character = 1 byte. So "stainless steel water bottle" is 28 characters = 28 bytes. Same number.
The trap is everything else.
- Spaces count as 1 byte — but Amazon doesn't ignore them, even though they're treated as separators in the index.
- Punctuation counts as 1 byte each — comma, period, semicolon, dash.
- Accented characters use 2 bytes in UTF-8 (é, ü, ñ, ç). One "café" eats 5 bytes, not 4.
- Emoji and special symbols — 3 to 4 bytes each. A single ☆ or 🌿 destroys your budget.
Plain ASCII English-only keywords give you the maximum effective space. The moment you add an accent or symbol, your byte budget shrinks while your character count stays the same. That's why character counters in Microsoft Word or Google Docs are useless for backend keywords — they show characters, not bytes.
Count bytes the way Amazon counts them
Open any browser developer console (F12 → Console tab) and paste:
new TextEncoder().encode("your keyword string here").length
That's the exact byte count Amazon will measure. No estimation, no rounding.
A real example. "stainless steel insulated water bottle 32 oz leak proof" looks like 56 characters. Run it through the byte counter: 56 bytes. Same. Now add "café-quality" → 12 more characters but 13 bytes (é is 2). Total jumps to 69 bytes, not 68 as Word would tell you.
For shorthand: every accented Latin character = +1 byte over its character count. Every emoji or symbol = +2 to +3 bytes.
Five formatting mistakes that waste backend bytes
1. Repeating words from your title. Amazon already indexes your title. If "wireless" is in your title, putting "wireless" in backend keywords wastes 8 bytes and gains zero ranking weight. Strip every word that appears anywhere in your title, bullets, or description before pasting to backend.
2. Using commas as separators. Amazon's documentation since 2018 has been explicit: separate backend keywords with single spaces, not commas. Each comma you add is 1 wasted byte. A list with 20 commas wastes 20 bytes — enough for 3 more long-tail keywords.
3. Including plurals when you have the singular. Amazon's algorithm tokenizes and stems. "Bottle" already matches "bottles", "bottled", "bottling". Adding all four eats 28 bytes for zero additional matches.
4. Stop words and articles. "The", "a", "and", "for", "with" — Amazon ignores these. Each one in your backend field is wasted bytes plus a hint that you don't know the format rules.
5. Quotes, brackets, and special characters. Amazon strips them at index time. "Best" reads the same to A9 as best. Quotes around phrases don't lock phrases together. Brackets do nothing. Every one of these is wasted bytes plus a parsing risk.
The exact format Amazon documents
This is the format Amazon enforces, taken from current Seller Central documentation:
keyword1 keyword2 keyword3 keyword4 keyword5
- Single spaces between terms
- All lowercase (Amazon lowercases at index — uppercase = wasted bytes)
- No commas, no semicolons, no quotes
- No repetition of title/bullet/description words
- Synonyms, misspellings, alternative naming, and use cases — all allowed
- Foreign-language terms allowed if your buyers might search in that language
If your category supports it, you can also use Subject Matter fields (separate from Search Terms) for additional indexed text. Those are different from the 250-byte Search Terms field and have their own limits.
A real before-and-after
Here's an actual product I worked on — a stainless steel coffee tumbler. The seller's original backend keywords:
travel mug, insulated coffee tumbler, stainless steel mug, coffee thermos,
hot drink container, with lid, for home and office, the best one
Length: 159 bytes. The waste: 7 commas (7 bytes), "the best one" (12 bytes — Amazon strips "the" and "best" is subjective filler), "for home and office" (the words "for" and "and" are stop words → 8 wasted bytes). Repeated concepts ("travel mug" + "stainless steel mug" + "coffee tumbler" overlap).
After optimization to 248 bytes:
yeti style hydroflask alternative double wall vacuum sealed leakproof
spillproof commuter ramen coffee tea hot cold beverages camping hiking
office gym yoga car compatible cupholder fits car cupholder gift dad mom
husband wife 32oz 24oz 16oz
Three months later, that ASIN moved from page 4 to page 1 for the search "yeti style coffee tumbler" — a query the seller had never typed into any optimization tool. The backend field was the only change. The title, bullets, and description stayed identical.
A workflow that actually fits 250 bytes
Step one — gather candidates. Start with 30 to 50 keywords from search query reports, competitor titles, and customer reviews. Use brand variations, model-fit terms ("yeti style", "hydroflask alternative"), and use-case terms ("camping", "yoga", "commuter").
Step two — filter. Remove anything already in your title, bullets, or description. Remove plurals when you have the singular. Remove stop words. Remove duplicate root meanings.
Step three — order by value. Long-tail buyer-intent terms first. Then synonyms. Then misspellings. Then alternative regional naming.
Step four — count bytes. Use the TextEncoder snippet above. Aim for 240 to 248 bytes. Leave 2 to 10 bytes of slack so you can add a new term later without rebuilding the whole string.
Step five — paste, save, wait 48 hours, then check indexing. Search Amazon for each backend keyword as a standalone query and confirm your ASIN appears.
Where SellerCard fits in
The 5-step workflow above is the pure manual approach — it works, and it's what I do for high-value ASINs that earn careful attention.
For everything else, SellerCard's listing generator writes the entire listing in a marketplace-specific format including a 250-byte-aware backend keyword string. The generator strips title duplicates automatically, applies Amazon's no-comma rule, and counts bytes (not characters) before saving. For sellers with 50+ SKUs, that's hours of byte-counting eliminated per launch.
The byte-counting math is also baked into our Amazon listing generator — pick Amazon as the marketplace, paste your product details, and the backend keyword field comes out at 240-248 bytes with zero overlap from your title or bullets.
What to do this week
- Pick your 10 highest-revenue ASINs — the byte payoff is biggest where impressions are biggest.
- Run the TextEncoder byte count on your current backend keywords — find out how much space you're already leaving on the table.
- Apply the 5-step workflow above to one ASIN as a test.
- Wait 48 hours, then check indexing on the new keywords.
- Track ranking for 14 days before deciding whether to roll out to the rest.
Backend keywords aren't the biggest lever in Amazon SEO — your title, bullets, and main image carry more weight. But they're the cheapest lever. Five hours of byte-aware rewriting across a 10-ASIN portfolio can surface long-tail searches you weren't ranking for at all. The competition won't bother because they think the field is 250 characters.
For the broader Amazon listing strategy that backend keywords plug into, see how to write Amazon product titles that rank and convert and Amazon backend keywords: what they are and how to use all 250 bytes.