A listing can be fully live, Buy Box won, plenty of inventory, and still not indexed for 40% of its target keywords. Sellers throw PPC at it and wonder why organic rank never climbs. The fix is not bidding, it is indexation. The most common causes: a restricted term in backend search, a 250-byte truncation, a brand-locked variant, or a prohibited claim quietly stripped by Amazon's content filter. This audit produces a heatmap of indexed vs not indexed for the top 30 keywords plus a root-cause diagnosis per gap.
When to use this
- 60-90 days post-launch and organic rank refuses to move
- PPC for a keyword performs well but the ASIN never shows organically
- After a backend keyword edit, verifying nothing was silently stripped
- Pre-launch sanity check, confirming indexation before scaling PPC
The framework. The Indexation Matrix
For each ASIN x target keyword pair, run a search on Amazon.com:
{ASIN} {keyword}
If the ASIN appears in results, INDEXED. If not, NOT INDEXED. If it appears only with quoted phrase fragments, PARTIAL.
Root cause diagnosis table
| Symptom | Likely root cause | Fix |
|---|---|---|
| Not indexed, term is in title | Brand-locked or prohibited claim | Review title, replace risky terms |
| Not indexed, term is in bullets | Bullet character limit or filter strip | Move to backend, simplify |
| Not indexed, term only in backend | 250-byte overflow or duplicate | Re-prioritize backend, dedupe |
| Partial (some variants only) | Variation parent indexation issue | Update parent-level keywords |
| Indexed but no rank movement | Indexed but irrelevant per A9/COSMO | Improve relevance signals in copy |
Step by step
-
Build the target keyword list. Top 30 commercial-intent terms for the product. Pull from Helium 10 Cerebro, Brand Analytics top search terms, or the seller's known target list.
-
Run the matrix. For each keyword, search "{ASIN} {keyword}" on Amazon. Log INDEXED, NOT INDEXED, or PARTIAL.
-
Build the heatmap. Visual grid, green for indexed, yellow for partial, red for not indexed. Sort by search volume descending.
-
Diagnose root causes. For every red row, walk through the diagnosis table. Identify the most likely cause.
-
Check backend search terms. Edit > Keywords. Confirm under 250 bytes, no duplicate words across title and backend, no restricted terms.
-
Check for prohibited claims. Words like "FDA approved", "best", "cure", "kills bacteria" without certification, "organic" without USDA cert. These get silently stripped from indexation.
-
Generate per-keyword fix list. Specific, actionable. "Move 'antibacterial' to backend, remove 'kills 99% germs' from bullet 2".
-
Run the quality check, then submit edits.
Output format
## Indexation Audit. [ASIN]
**Heatmap (top 30 keywords sorted by search volume)**
Keyword | SV | Status | Where it appears | Diagnosis
[rows]
Indexed: [N] of 30
Partial: [N] of 30
Not indexed: [N] of 30
**Backend health check**
- Byte count: [N] of 250
- Duplicates with title: [list]
- Restricted terms detected: [list or "none"]
**Fix list (ranked by SV impact)**
1. [Specific edit] - unlocks [keyword] - SV [N]
2. [...]
Worked example
ASIN: B0F4XYZ123, sleep mask, 30 target keywords. Matrix run results: indexed 19, partial 3, not indexed 8.
Not indexed analysis:
- "blackout sleep mask" (SV 14,800): in title but missing. Diagnosis: title is 187 chars, Amazon truncated. Fix: shorten title to 145 chars, prioritize this term in first 80 chars.
- "anti aging eye mask" (SV 5,400): in bullet 4. Diagnosis: bullet has "FDA approved" claim, full bullet got filtered. Fix: remove FDA claim.
- "silk sleeping mask" (SV 9,100): not in any field. Diagnosis: missing. Fix: add to backend search terms (uses 18 bytes).
- 5 more similar.
Backend health: 287 bytes (over limit by 37, last 4 keywords are not indexed), 3 duplicates with title wasting 22 bytes. Cleanup recovers 59 bytes for new terms.
Fix list ranked by SV impact: 8 specific edits, total SV unlock = 67,000 monthly searches. Projected organic impression lift: 35-50% within 14 days of edit.
Quality check
- Search performed on amazon.com logged out or in incognito (personalization off)
- Search includes the ASIN to force-find regardless of rank
- Backend byte count measured in UTF-8 bytes, not characters
- Restricted terms checked against current 2026 prohibited claims list
- Variation parent and child both checked if listing is varied
Common mistakes
- Searching by keyword only. Without ASIN you cannot tell if the listing is indexed but just ranked low.
- Using character count for backend. Amazon uses bytes. Non-ASCII chars count as 2-4 bytes.
- Duplicating title words in backend. Wastes the 250 bytes, no ranking benefit.
- Ignoring partial status. Partial means a variation issue, easy to miss.
- Not re-auditing after edits. Index updates take 24-72 hours. Re-run the matrix to confirm fix took effect.
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