Amazon search has two gates, and most listing advice muddles them. First, indexing: does your listing appear in the result set for a query at all, decided by the words in your title, bullets, backend search terms, and structured attributes. Second, conversion: once shown, does your listing earn clicks and purchases, because Amazon promotes what sells and demotes what browsers skip. A listing can fail either gate independently, and the fix differs, which is why diagnosis comes before rewriting.
Passing the indexing gate
- Title: the primary keyword surface, built per your category’s style guide, brand, product line, key attributes, size and count, the title craft in its Amazon dialect: spec-dense, front-loaded, no promotional fluff.
- Bullets: five lines that index AND persuade, one benefit-plus-spec per bullet, the secondary keywords worked in naturally. Keyword lists formatted as bullets convert nobody and read as spam.
- Backend search terms: the hidden field (capped by byte length, around 250 bytes) for synonyms, regional spellings, and vocabulary that does not belong in visible copy. No competitor brands, no repetition of words already indexed, both wasted and against policy.
- Structured data: category placement and every attribute field filled, the same data discipline as everywhere, plus correct identifiers, because attributes feed the filters buyers narrow with.
The blunt test: search your target phrase, and if you are absent past the first pages entirely, it is an indexing problem, add the vocabulary; if you are present but unclicked, it is a conversion problem, keep reading.
Passing the conversion gate
- Main image decides the click: product on pure white filling most of the frame, high resolution (1000px+ enables zoom), per Amazon’s image requirements. The remaining slots earn the purchase: scale, angles, lifestyle context, and an infographic for the spec buyers skim for.
- Price sits inside the click decision too: the result page shows it, and the Buy Box math governs multi-seller listings, repricing with real floors.
- Reviews compound slowly: early velocity from a genuinely good product, fast fulfillment, and Amazon’s own request tooling, never from incentives that violate policy.
- Availability protects everything: rank decays during stockouts, so the optimization program is only as strong as the in-stock discipline under it.
Category rules bound all of it: restricted categories carry their own listing requirements, and claims that trip them suppress the listing outright.
The loop that makes it optimization
One change, then two weeks of Business Reports: sessions (indexing and click-through), unit session percentage (conversion), and rank on your target queries. Machine-drafted variants accelerate the drafting; the measurement decides what ships. Sellers who skip the measuring are redecorating, not optimizing, the same test-then-roll habit as every other number.
Common questions
How often should I touch a listing?
When the data says so: falling sessions, weak conversion against category peers, or new query vocabulary. A well-ranked listing earns benign neglect.
Do backend search terms really matter?
For indexing coverage, yes, they catch the synonyms your visible copy cannot hold gracefully. They do not boost rank by repetition; coverage is the whole job.
What about A+ content?
Brand-registered sellers get richer description modules that lift conversion on considered purchases. Worth building after the fundamentals above, never instead of them.
Does my off-Amazon catalog matter here?
Consistency does: one catalog source feeding every channel keeps titles, attributes, and quantities agreeing, and the quantity part is what keeps the optimized listing sellable, sync is the unglamorous half of optimization.
Optimized listings, never dark
Complete data flowing to Amazon from one catalog, and stock truth that keeps rank compounding instead of resetting, from $49/month with unlimited orders. See pricing.