Amazon SEO Best Practices for 2025: Making A9 and Rufus Work Together
Amazon SEO best practices in 2025 come down to one uncomfortable shift: you are no longer optimizing for a single ranked list. A9 still governs classic keyword relevance and conversion signals, but Rufus, Amazon's in-app shopping assistant, reads your listing the way a shopper asks a question out loud. The workable answer is one listing that satisfies both engines — clean structured attributes plus copy that answers real questions in plain language. What follows is the routine I run with brands here in Los Angeles and across US marketplaces, from Seller Central accounts in El Segundo to FBA sellers shipping out of the Inland Empire warehouses.
Start With a Two-Engine Audit, Not a Keyword Dump
Most sellers still treat keyword research as a list-building exercise. That gets you halfway. A9 weighs relevance, conversion rate, and velocity; Rufus leans on structured attributes, A+ content, review text, and the Q&A section to assemble an answer. If your attributes are half-filled, Rufus has less to work with, and your listing underperforms in conversational results even when your A9 rank looks healthy.
So before touching bids, audit both sides. Pull your Search Query Performance report in Brand Analytics, export the top 20 converting queries, then read your own listing out loud. If a phrase in your bullets would sound strange in a customer's spoken question, rewrite it.
The Pre-Publish Checklist I Run on Every Listing
This is the same sequence I apply before a launch or a relaunch. It takes about an hour per parent ASIN and catches most of the expensive mistakes.
- Confirm category and browse node. A wrong node can suppress relevant browse traffic for weeks, and it is painful to unwind after reviews accumulate.
- Title: brand, core keyword, one differentiator, readable in the mobile truncation. Skip the comma-stuffed keyword salad.
- Bullets: five, each opening with a benefit phrase a shopper would actually say, then the supporting spec.
- Backend search terms: no repeats of visible copy, no competitor brand names, no competitor ASINs. That is a policy issue, not a clever trick.
- Attributes: fill material, size, color, target audience, compatibility, and country of origin. These structured fields feed Rufus directly.
- Images and video: main image on pure white, lifestyle shots that show scale, and a short video if your category supports it.
- Mobile check: open the live detail page on a phone and read the first three lines. Most US shoppers never see the rest.
Run the checklist, then leave the listing alone for two weeks. Constant edits reset learning and make your own data unreadable.
Writing for Rufus Without Losing Keyword Coverage
Rufus answers questions like "will this fit a 65-inch TV on a wall with studs 16 inches apart" or "is this safe for a toddler." Your listing either contains that information or it does not. The fix is not more keywords; it is specificity. Add a short compatibility paragraph, a sizing table in A+ content, and a Q&A entry for the three questions your customer service inbox sees most.
Keep your primary keyword in the title and first bullet, then write the rest for a human. Hummingbird-style semantic coverage still matters to A9, but it comes from real detail — materials, dimensions, use cases — not from repetition. Reviews and answered questions get indexed too, so respond to them with substance rather than a canned line.
Reading US Metrics Without Fooling Yourself
Everything in Seller Central is denominated in USD, and that makes the math simple but also easy to misread. Track session-to-conversion rate, ACOS, TACOS, and Buy Box percentage together. A rising ACOS during Prime Day or the Black Friday-Cyber Monday window is not automatically a problem; a falling ACOS with falling sessions usually is.
Los Angeles sellers face a specific wrinkle: local shoppers expect fast delivery and compare you against same-day options, so your conversion rate can be lower in the LA metro than in secondary markets even with identical creative. Segment your reports by region before you conclude a listing is weak. Also watch inventory placement — split FBA stock across both coasts when volume justifies it, because shipping speed feeds conversion, which feeds rank.
Compliance Details US Sellers Get Wrong
- Review practices: no incentivized or discounted reviews, and no review swaps. The FTC's endorsement guides apply to your brand's own marketing as well.
- Health and safety claims: avoid "FDA approved" style wording unless you can document it. Most supplements and cosmetics cannot.
- California specifics: Prop 65 warnings on applicable goods, and CCPA/CPRA obligations if you collect customer emails off-platform.
- Trademarks: never place competitor brand names in backend search terms. It risks suppression and, in some cases, legal letters.
None of this is glamorous, but suppressed listings and pulled reviews cost more than any optimization gain.
Test Small, Then Scale What Earns Its Spend
Amazon Tools is built around the two-engine reality: it checks your listing against both A9 ranking signals and Rufus-facing content quality, flags the gaps, and charges on performance rather than a flat monthly fee. That structure suits sellers who want to test before committing budget. Treat any tool as a diagnostic, not a verdict — you still own the decision about what to change and when.
A reasonable next step: run one parent ASIN through the audit, fix the attribute gaps and the top three Rufus-style questions, and give it two weeks of clean data before you judge the result. If the session-to-conversion rate moves and your target queries hold, expand to the next ASIN.
You can start the check at amazon-seo-geo.dsux.com. Bring one listing you already know well — that is the fastest way to see whether the diagnosis matches what you suspect.