A listing that converts
How Amazon Works · C2
Titles, bullets, images, A+ and video — the parts of the page that turn a click into an order, with the numbers behind them.
Titles, bullets, images, A+ and video — the parts of the page that turn a click into an order, with the numbers behind them.
Ranking ([C1]) and ads ([C3]) only decide who sees the thumbnail; the listing decides the two ratios that compound into revenue — click-through rate (CTR: clicks ÷ impressions) and conversion rate (CVR: orders ÷ clicks). One agency case moved CTR from 0.8 % to 1.8 % and CVR from 19 % to 30 % by changing a single main image, and revenue went from $39K to $139K a month [L17]. A brand-story-plus-A+ rebuild took another seller from $146 to $1,200 on the same calendar day, no coupon [L67]. Nothing else on Amazon is this cheap to change or this fast to pay back.
The mechanism: two ratios, one page
The funnel is impressions → clicks → orders, and each step has an owner: the main image and the first words of the title own the click; images 2–7, bullets, A+ and reviews own the order [L84]. Amazon's free Search Query Performance report shows which step is broken — a steep drop from impression share to click share on a term is a thumbnail problem; from click share to purchase share, a page problem [L38]. Fix conversion before indexing; you cannot rank or profit on a term the page does not convert [L93].
The leverage is lopsided. The main image is called "more important than price or reviews" [L52], the "only real weapon" in the search-results knife fight [L64], the "hidden sales killer" in a 100,000-listing audit [L87]; an Amazon science paper reportedly ranks CTR ahead of conversion as a ranking factor [L16], so a better thumbnail also buys organic position. The arithmetic: CTR 1 % → 1.3 % and CVR 10 % → 13 % (both +30 % relative) lifts sales ~80 % and more than doubles profit [L17].
Above the fold: title, bullets, backend
SHIO's own listing is the worked example through this section — title, bullets and gallery as a phone shows them [SS-02] — with the category leader as the benchmark [SS-04].
Title
The title carries the most indexing weight of any field; earlier words weigh more than later ones, intact phrases more than scattered words [L43]. Humans read the first five words [L84]; Amazon's "Project Amelia" reorders titles per shopper with priority on roughly the first 80 characters, because ~75 % of shoppers are on mobile [L16].
Amazon is formalising that: from 27 July 2026 non-media titles are capped at 75 characters ("Item Name") plus a 125-character "Item Highlights" field — the ~200-character budget is unchanged but split; sellers get 14 days to review Amazon's AI split, and doing nothing lets Amazon do it, which in the presenter's test cut "coffin gift box" to "gift box" [L63] (US — verify for .ae). Both fields are searchable [L91].
| Rule | Detail | Source |
|---|---|---|
| Lead with the shortest, most-searched exact phrase | Position and exact string both score | [L47] [L43] |
| Write one noun phrase, not a keyword pile | Head noun + material / use / audience / constraint / proof, ~75 chars | [L02] [L04] |
| Cover ~5 of Amazon's 15 semantic relations | What it is, used-for, used-by, used-in | [L16] |
| Repeat a keyword ≤2× in the title, ≤3× in the listing | Past 3 the ASIN can be de-indexed for the term; once is enough | [L84] [L93] [L43] |
| Keep phrases intact across the 75/125 split; fill the allowance | [L63] | |
| Highlights = benefits an image cannot show | [L01] | |
| No competitor brand names in listing text | Allowed as PPC targets only | [L54] |
Where sources disagree — brand name placement. My Amazon Guy: an unknown brand's name at the front costs CTR and belongs at the end [L84]. Helium 10's reading of the new format: brand first, product name second [L63]. Crescent Kao: omit it, Amazon prepends the brand field [L47]. With no brand search demand, follow [L84] — the first five words must say what the product is — but check what the marketplace's Item Name policy enforces.
Pick the title strategy by the weak metric: 1–2 keywords drive 70–80 % of sales → brand + product type + those keywords; strong CVR, weak CTR → lead with the strongest differentiator; new or long-tail ASIN → maximise natural keyword coverage [L91]. Superfuel claims 25–30 % impression/click lifts and shows one change moving CTR 1.5 % → 1.84 % [L91]. Low "title density" (few page-one titles with your exact phrase) can put you on page one from day one [L63].
Bullets
Amazon reportedly weights only the first ~1,000 characters of combined bullets: front-load [L84]. Sell benefits — ask "so what?" of each spec until you reach the felt outcome, lead with it, use the feature as proof [L53]; the litmus test is whether a competitor could paste the bullet onto their listing [L47]. Put a constraint or proof phrase in the first two bullets (the filters an AI agent applies before comparing) and add an audience/occasion line [L02]. Lift customers' own phrases from reviews [L02] [L71]. Rufus reads bullets to answer predicted questions, so write bullets that answer them [L89]. Brand-registered sellers get up to 10 bullets indexed [L04].
Backend and attributes
Leftover keywords go into the 500-character generic-keywords field [L43], as single words rather than phrases [L54]. A blank structured attribute is an exclusion before scoring, not a penalty [L04]; the filter tick-boxes on your search page are the checklist of attributes to state [L53]; "used with" and "used in location" are the nodes most listings miss [L96]. A free Catalog Auditor flags missing attributes and title violations [L08]. Claim only what is true — mismatches produce returns that teach Amazon to stop suggesting you [L53].
The image stack
The stack below rebuilds SHIO's gallery slot by slot; the current opening frame is the baseline to beat [SS-01].
Main image: the one lever
Constraints: white background, product only, filling ≥85 % of the canvas [L82]; props and models are a terms-of-service grey area that can suppress the listing — check whether top competitors get away with them first [L82] [L72]. Square is preferred in most niches [L52]. Professional photography or 3D renders are the baseline; phone and manufacturer photos are "under-optimised by default" [L84].
What moved CTR in the cases:
- Four upgrades in one image — colour and badges, visible ingredients, the primary keyword printed on the pack, photo → render — took CTR 0.8 % → 1.8 % at 99 % confidence within 48 hours [L17].
- Print the selling point on the pack: "22g protein per serving", "1kg", the search keyword instead of a sub-brand; recolour a cream bag bold orange so it stops "getting lost in the search results" [L52]; benefit text on the visible package drove one seller's US launch [L76].
- Sell the experience, not the sealed box: product out, sensory cue in — a protein-puff brand won a poll 23–7 [L10].
- Screen-first packaging: go dark (screens are white), keep it simple (thumbnails are small), distinctive typeface, break one category rule, sell an identity [L87]; "stop the scroll" against the 45–50 products on a results page with ingredient → benefit callouts and blank space [L72].
- Searchable benefits on the image, a badge, product filling the frame [L07]; primary selling point + purpose + three features + product in one thumbnail [L84].
Amazon's Rekognition OCR reads text on images and packaging, so image text feeds relevance as well as conversion [L16] [L72]; a free checker shows what Amazon detects [L13].
Images 2–7: one job each
Up to 10 images and 10 videos are allowed, all indexable [L75]. Build the sequence around sticking points mined from negative reviews, not around angles [L82] [L47]; every callout must pass two questions — do shoppers care, will they get it in a second? [L82]
| Slot | Job | Evidence |
|---|---|---|
| 2 | Scannable infographic: 3–4 benefits with icons (facts panel for supplements/food) | [L84] [L82] |
| 3 | How it works: cross-section render of the internal mechanism | [L82] [L68] |
| 4 | Show, don't tell: before/after or a demonstration of the key claim | [L82] |
| 5 | Lifestyle, model from the buyer demographic; alternate eye contact / looking away | [L84] [L82] |
| 6 | "Us vs. them" comparison answering the top objection | [L82] [L75] |
| 7 | Install/use in 3 steps, size reference, what's in the box | [L82] [L21] |
| 8+ | Dense spec image "for nerds and Rufus"; FAQ image of real questions; care instructions | [L75] |
No screenshots of Amazon's review UI or star icon [L21]; new images in A+, not gallery repeats [L84]; photorealistic AI-generated people must be labelled, AI scenes without people are unrestricted [L01].
Mobile-first, scan-then-read
Three-quarters of traffic is mobile and secondary images illegible on a phone are "functionally useless" [L21] — check every image on your own phone; add zoom-ins on the detail that substantiates a durability claim [L82]. Shoppers scan in F/Z patterns: design a scan layer (contrast, hierarchy, white space) first, a read layer (one point per section) second [L11]. Localising the images per marketplace was one €3M seller's "main secret" for CVR and PPC cost; Seller Central supports country-specific image sets [L76] (EU — verify for .ae).
Below the fold: A+, brand story, video
SHIO already has the below-the-fold real estate built — a "From the brand" band and a product video — which most of its .ae rivals lack [SS-08].
Shoppers click into reviews first, then scroll back up to A+ [L68]. Basic A+ (needs Brand Registry [L59]) historically lifted conversion by at least 20 % [L75]; the In Motion Hemp A+ redesign — mechanism graphic, pain-point callouts, drug-facts image — took CVR 13 % → 15 % and monthly units 497 → 548, projected +$15–20K a year, with page views slightly down [L68] [L71]. Brand Story is a swipeable carousel of up to 7–12 modules that lifts conversion, cross-sell and average order value and indexes keywords in its images [L21] [L67] [L84]; a newer "Shoppable Collections" module is replacing it as free in-listing advertising [L21] [L19].
Premium A+ is "the most effective conversion tool Amazon gives sellers" — mobile-optimised, with carousels, hotspots, Q&A and shoppable comparison tables — and four in five sellers lack access [L36]. Of 19 modules only 7–8 earn the 7 slots: Full Video, Full Image, Hotspots, Regimen Carousel, image/video carousels, Comparison Table, Q&A; skip Text, Single Image with Text, Technical Specs [L36]. Q&A built from return reasons and negative reviews cuts mismatched-expectation purchases [L36], and question-plus-answer is the format AI assistants read best [L89] [L67]. Put answers in crawlable A+ text — Amazon is moving away from alt text [L89]; images alone are not indexed for search [L47].
Where sources disagree — Premium A+ eligibility. Brand Story on every live ASIN is common to all; the approved basic A+ count is 15 [L36], "5 (up to 15 for some brands)" [L21], or "5 changes in 12 months" [L67]; approvals run monthly [L36] or weekly on Fridays [L67]. All agree revisions of one listing count and a 1–2 week wait follows [L67] [L71]. Assume 5, iterate to 15 if the module does not unlock (US/UK — verify for .ae).
Video fills a large module above reviews; reviewer and influencer video is judged more persuasive than brand-made [L21]. Diversify — comparison, testimonial, how-to, best-features [L75]. The first frame is the thumbnail and the product must solve the problem within three seconds; swap only the hook to test angles [L19]; open with the top sticking point [L82]. AI-generated video beat polished human video on ACoS [L19]; Sponsored Brands video pre-qualifies clicks [L51] ([C3]).
Variations
Merging single-unit and multipack SKUs into one parent lifted single-item CVR, Subscribe & Save uptake, AOV and daily orders while cutting repeat ad spend [L71]. Caution: since Feb–May 2026 reviews pool across children only when the declared variation theme implies no functional difference — colour, size and volume share pool; bundle-vs-standalone and flavour do not — and a mislabelled theme kills sharing for the whole family [L10]. A child launched only to inherit reviews should be disconnected once it stands alone [L39]. Virtual bundles are the safer cross-sell: a free slot under the buy box that displaces a competitor's ad [L21] [L71].
How the numbers were made: testing
| Method | Speed | Note |
|---|---|---|
| Amazon Manage Your Experiments (live A/B) | 8–10 weeks | Loser costs real sales, so sellers test timidly [L74] [L64] |
| Off-platform panel (PickFu / Helium 10 Audience) | 1–2 weeks | ~$85 tests; Yesbar +12 % traffic in two weeks; the written "why" is the durable value [L74] |
| Manual swap, watch daily CTR | Days | Read raw recent days, not the 30-day blend [L93] |
| Search-results mockup vs top 3 competitors | Pre-launch | Test the click in context [L74] [L16] |
Benchmark against competitors, not your old image — beating yourself is a local maximum [L74]. Change one variable and watch its SQP shares for 2–3 weeks [L02]; stagger title, image and text changes unless the listing is already failing [L16]. Then make it a ritual: one experiment a month, rotating main image → secondary images → A+ → video → bullets → title [L71]. Refreshed listings also compound exit value [L90].
QA against Rufus and AI agents (US — verify for .ae)
Rufus reads the whole listing — title, bullets, A+, backend, images, reviews, Q&A — and Cosmo asks whether the product solves the described problem, not whether the words match [L75] [L53] ([C1]); Rufus users are said to be 60 % more likely to buy [L96]. Loop: ask Rufus shopper questions about your product; wherever it says "the listing doesn't specify", add an image callout or bullet [L75] [L89]; log which prompts your ASIN appears for versus competitors and re-check monthly [L89]. Listings untouched ~6 months are reported to slide [L96].
Common failure modes
- Manufacturer or phone photos; images unchanged for years [L84] [L67].
- Secondary images written like manuals, illegible on mobile [L21] [L11].
- No A+, brand story, variations or video — the pattern in every plateaued-seller audit [L69] [L72] [L71].
- Titles that repeat a term 3+ times or hide the product behind an unknown brand [L93] [L84].
- Never testing the main image against competitors [L74] [L71].
- Claims the product cannot keep [L53].
What this means for SHIO
Assumptions: one filtered shower head on amazon.ae, no brand search demand yet, Brand Registry unknown. In this order.
Title (75 + 125 if .ae adopts the split; otherwise same order in one field)
- Item Name (75 chars): Filtered Shower Head, Pressure Boost, Hard Water & Chlorine Filter, 3 Modes — shortest exact phrase first [L47], one noun phrase [L02], ~5 relations: what / for hard water / for skin and hair / in the shower / 3 modes [L16]. Brand at the end or in the brand field [L84]; move it forward only once "shio" has search volume.
- Highlights (124 chars): Vitamin C + activated carbon cartridge for dry skin & hair breakage from hard water | high-pressure spray for low-flow flats — benefits a photo can't show [L01]. Verify the exact phrases against .ae search data first ([C1]); keep "filtered shower head" intact across both fields [L63].
Bullets (first 1,000 characters carry the weight [L84] [SS-02]) 1. Constraint/proof: what it filters (only what you can prove), cartridge life in days, fits a standard 1/2" hose [L02]. 2. Benefit first: softer skin, less breakage ← vitamin C + carbon as proof [L53]. 3. Pressure boost, in the buyer's own words from competitor reviews on .ae [L02] [L71]. 4. Audience/occasion: for Dubai and Abu Dhabi apartments and villas with hard water [L02]. 5. Three modes, tool-free install, what's in the box; answer the predicted questions — replacement filter, leaks, hose compatibility [L89]. Litmus each bullet: could a generic filter brand paste it? [L47]
Backend and attributes: single-word leftovers into the 500-char field [L43] [L54]; fill every attribute — material, spray settings, mounting, "used in: bathroom, shower", "used with: shower hose, replacement cartridge" [L04] [L96]; mirror the filter tick-boxes on the .ae results page for "shower filter" [L53].
Image stack
| Slot | SHIO image | Why / source |
|---|---|---|
| Main | Head + cartridge on white [SS-01], ≥85 % of frame, square, rendered; test (a) tan/blue cartridge visible against chrome competitors [SS-04], (b) benefit printed on the pack, (c) faint spray as the "experience" cue | [L82] [L17] [L52] [L87] [L10]; poll vs top 3 .ae competitors [L74] |
| 2 | Infographic, 3 icons: filters chlorine/hard water, boosts pressure, 3 modes | [L84] [L82] |
| 3 | Cross-section render of the cartridge: carbon layer, vitamin-C layer, flow path | [L82]; mechanism graphic drove 13 → 15 % [L68] |
| 4 | Show-don't-tell: spray from a standard head vs SHIO on the same supply; used vs new cartridge | [L82] |
| 5 | Lifestyle: model matching the .ae buyer, in-shower, eye contact | [L84] [L82]; label if AI person [L01] |
| 6 | Us-vs-them table vs "regular shower head" and "generic filter": pressure, media, modes, refill cost | [L82] [L75] |
| 7 | 3-step install + fits standard hose + size next to a hand | [L82] |
| 8 | Specs for Rufus: flow rate, cartridge life, dimensions, materials; FAQ image from real questions | [L75] |
| 9 | What's in the box; replacement schedule (pre-empts "stopped working" reviews) | [L21] [L75] |
| Check every slide on a phone [L21]; the objections in slots 4 and 6 must come from mining competitor reviews on .ae [L82]. |
A+ / brand story / video - Get Brand Registry if not done [L59]. Basic A+ with new images [L84] [SS-08]: mechanism, UAE hard-water problem → solution, refill logic, brand story (塩, Japanese roots, Dubai). Fill gaps from a review synopsis once reviews exist [L84]. - Work toward Premium A+ by resubmitting revisions (5, then 15) [L21] [L36] (verify on .ae); then Full Video, Hotspots on the cartridge, Comparison Table (head vs refill), Q&A [L36]. - One 30-second video: pressure claim in the first 3 seconds, then filter, then modes [L19] [L82]; add reviewer/UGC video as it arrives [L21].
Variations: keep the head as one listing; sell the replacement cartridge as a virtual bundle under the buy box, not a child ASIN — bundle-vs-standalone no longer shares reviews [L10] [L21]. A later 2-pack is a quantity variation and pools reviews [L10].
Cadence: main-image poll before any ad spend [L64]; one experiment a month through the stack [L71]; SQP weekly to see whether the gap is impression → click (thumbnail) or click → purchase (page) [L38]. Do not claim "removes 99 % of chlorine" or "cures hair loss" unless a test backs it [L53].
The concept graph behind this chapter
While building the corpus, lore filed the ideas above as pages of a knowledge graph — each link opens that concept's own page, with its definition, the videos it came from, and how to apply it. The 36 most relevant of the subject's 385 published concept pages:
Amazon A+ Content · Day Parting (Amazon PPC) · AWD (Amazon Warehousing and Distribution) · Amazon 85% Canvas Rule (Main Image) · Amazon Listing Title Structure Formula · Amazon Listing Optimization: SEO vs. Conversion Optimization · Amazon Seller 10-Step Launch Process · Six-Area Amazon Listing Optimization Framework · Amazon as the True Bottom of Funnel · "Amazon Escape" 23-Criteria Product Funnel · Advisor-Led vs. DIY Amazon Brand Sale (Fee Alignment & Risk) · Amazon Brand Storefront as Ad-Free Conversion Destination · Amazon Virtual Bundles (Buy Box Cross-Merchandising) · Five Reasons Amazon FBA Products Fail · Instructional / How-To Image · Amazon Gated-Category Documentation Requirement · Amazon Product Life Cycle (Launch, Expansion, Harvest) · Amazon Creator Connections (Influencer Affiliate Program) · 'Us vs. Them' Comparison Image · Amazon-Only vs. Omnichannel Seller Skillset Framing ("Inside vs. Outside the Box") · A+ Content New-Image Rule · Content Marketing for Amazon Launches: UGC Video & Free Lead Magnets · Multi-Use Call-Out Image · Secondary Image Set Testing · Size Reference Image · Specialist-Partner Team Structure (Sourcing + Meta + Amazon) · Before-and-After Results Image · Show-Packaging Image (Safer Alternative to Main-Image Props) · Show-Everything / Show-Variations Main Image · Master Keyword List & Listing Scorecard · Main-Image Tactic Checklist · Helium 10 Listing Builder · Brand Domination Check · Hidden SEO Image Text (Invisible Keyword Indexing) · Brand-Name Boundary: Listing Content vs. PPC Targeting · Mobile-First Image Design & Detail Zoom-Ins
Source legend
Not covered / open questions
- Whether amazon.ae has the 75/125 title split, Premium A+, Shoppable Collections, Creator Connections video, Manage Your Experiments and Rufus at all — every source is US/UK/EU; [C8] must confirm before the SHIO plan is executed.
- No CTR/CVR benchmarks for bath hardware; the 0.8 → 1.8 % and 13 → 15 % figures come from spices and hemp cream. Pull SHIO's own SQP for the baseline.
- Arabic/bilingual images and copy on .ae are untested; the localisation evidence is EU-only [L76].
- Image-text indexing rests on practitioner demonstrations [L16] [L72] and conflicts with the older "images are not indexed" view [L47].
- Nothing isolates the effect of video or Brand Story alone; the $146 → $1,200 case bundles several changes and a Prime Day tailwind [L67].
- Whether "unknown brand at the end of the title" survives an Item Name policy that expects brand first is unresolved [L84] [L63].