Summary
The page's frame is sound: clear product categories, breadcrumbs, a purchase button just below the first screen with the size preselected, and a sticky purchase bar. The problems sit around that path. The information a Japanese shopper needs before committing is spread across the page, behind icons or off the site, and the structure gives it no fixed place.
| Structural issue | What we measured | Best Japanese reference |
|---|---|---|
| 1. “Which sofa is right for me?” has no home. Comparison sits at the end of the product page, not in the navigation. | 0 menu entries for 比較 or 選び方. The three sofas appear in the menu by name only. The comparison starts at screen 10.1 of 12.1. | ACTUS: a sofa buying guide linked from the homepage |
| 2. Trust content arrives late. Delivery, returns and reviews are far from the decision point. | Product page: reviews at screen 10.5 (LOWYA: review summary at 4.1). No arrival date anywhere; return conditions on an external help centre. | LOWYA: delivery date, guarantee tags and store check above the cart button, within 2.4 screens |
| 3. The product page tells the story before the facts. Seven brand-story sections come before comparison and reviews. | 8.3 screens of story (1.8–10.1), with 「究極の座り心地」 twice and 「ふんわりやわらか」 three times. | LOWYA: description after the purchase area, reviews next |
| 4. Ways to try before buying are split up. Showroom and fabric samples sit in two places. | The showroom is a utility link in the drawer. Fabric samples are an ⓘ pill under the button. Neither is linked from the product page body. | LOWYA, Re:CENO, unico: 「店舗の展示・在庫を確認する」 next to the cart button |
A long page is not the problem in itself: LOWYA's product page is longer than Mareo's (14.0 vs 12.1 screens). What differs is what comes first. LOWYA puts the decision and trust content in the first 2.4 screens and its reviews at 4.1. Mareo does the reverse.
Page layout and hierarchy
Each bar is one page at mobile width, measured section by section. Colours show what each section is for. The question is not how long the page is but where the decision and trust content sit.
Why each page exists
The job each page should do, based on how visitors arrive and what they need at that point. The maps below are judged against it.
| Page | How visitors arrive | What they need to do here |
|---|---|---|
| Top page | Brand search, ads, social, returning visitors | Understand what Mareo is, see that it's trustworthy, and get to the right sofa quickly. Out of scope in this example: no homepage was reviewed. |
| ソファ category | Menu | Compare the three sofas and pick one. Out of scope in this example. |
| Sofa product page | Product-name search, ads, the menu | Choose size and colour, confirm the seat feel, check that it fits the room and the doorway, check delivery and returns, read reviews, buy |
| Search | Menu drawer (mobile) | Find a product, or an answer to a service question (returns, delivery, payment) |
Top page
Out of scope in this example. The review covered the product page, header, menu and search; a homepage map would compare Mareo's top page with LOWYA's in the same way.
Product page
- Mareo: buy box, then five accordions, then seven brand-story sections (1.8–10.1). The same claims appear more than once. Comparison starts at 10.1, reviews at 10.5.
- LOWYA: rating, gallery, options, price (税込), guarantee and delivery tags, delivery date by postcode, store check and cart, all within 2.4 screens. Description follows, with reviews from 4.1. LOWYA positions below screen 3 shift by a few hundred pixels between loads because of lazy loading.
Category page (ソファ)
Out of scope in this example.
Heading hierarchy (Sumi product page)
- The product name is the page's H1, as it should be.
- The seven story headings are H2s at the same level as 「モデルを比較」, 「カスタマーレビュー」 and “FAQs”, so the decision content has no more weight than the story. The four tiles add H3s, one of them in English (“Height Adjustable”).
- The accordions have no headings at all, so the specs, delivery and warranty sections are invisible to an outline or a screen-reader heading list.
Key journeys (遷移)
Eight tasks a Japanese sofa shopper typically performs, walked on mobile. Red steps mark friction. Green steps show how a Japanese site handles the same step.
T1 · “Which of the three sofas suits me?” no menu route
The three sofas differ mainly in seat feel and construction (pocket coils, feather, foam), which is exactly what a shopper needs to choose. None of it appears in the menu, and the only place the three meet is the comparison at the end of the product page, where firmness is a number out of five rather than a word.
T2 · “When will it arrive, and what if I don't like it?” 4 places to check
T3 · Buy Sumi, 3-seat, grey 2 taps to cart
The shortest path works well, and the sticky purchase bar keeps the button in reach. The issue is what the shopper is unsure about at the button (T1, T2, T6), not the number of taps. Cart and checkout weren't entered in this review.
T4 · “Can I sit on it somewhere first?” utility area only
T5 · “I have a question before buying” leaves the site
T6 · “Will the sofa fit through my door?” no answer on the page
The FAQ sends the shopper back to a table that doesn't contain the answer. Nothing says whether delivery includes carrying the sofa into the room.
T7 · From the homepage hero to a sofa out of scope
Out of scope in this example: no homepage was reviewed. In a full review this journey checks whether the hero names its destination and lands on the right option.
T8 · “What's the returns policy?” via search no answer on site
IA findings
Structural findings by layer (structure, labels, disclosure). The right-hand column links each one to the localisation review where it overlaps.
| # | Finding | Severity | Review link |
|---|---|---|---|
| Structure | |||
| IA-1 | The choosing task has no home. Comparison and buying advice aren't in the menu; the only comparison sits at screen 10.1 of the product page. | Major | Finding 10 |
| IA-2 | Try-before-you-buy routes are split. Showroom (utility link) and fabric samples (ⓘ pill), and neither is linked from the product page body. | Major | New |
| IA-3 | Homepage sections. Out of scope in this example: no homepage was reviewed. | — | — |
| IA-4 | Category page. Out of scope in this example: no category page was reviewed. | — | — |
| IA-5 | Product-page order follows the brand story, not the decision. Seven story sections (8.3 screens, repeated claims) come before comparison and reviews. | Major | Finding 12 |
| IA-13 | Search is product-only and hidden on mobile. Service questions (返品, 配送, ペイペイ) return products; the FAQ lives on an external help centre; on mobile search is only inside the menu drawer. | Major | Finding 13 |
| IA-14 | One product split across several URLs. Not observed on the reviewed page: sizes and colours are in-page options. Other product pages were out of scope. | — | — |
| Labels | |||
| IA-6 | Menu items carry no reason to choose. Sumi, Ren and Mio are names only; tags describe newness and sales (NEW, SALE). | Major | Finding 10 |
| IA-7 | The utility area mixes four kinds of link (brand, showroom, B2B, help); FAQ leaves the domain with only an icon as warning. | Minor | New |
| IA-8 | Naming drift across the page: “Free Shipping” / 送料無料, 「口コミ」 / 「カスタマーレビュー」 / “reviews”, 「Q&A」 / 「よくある質問」 / “FAQs”, and two different newsletter offers (10% and 3,000円). | Minor | Finding 6 |
| IA-15 | Hero button destination. Out of scope in this example: no homepage was reviewed. | — | — |
| Disclosure | |||
| IA-9 | Purchase-risk information is in four places (ⓘ pills, accordion, product FAQ, external help centre); the arrival date and return shipping cost are in none. One product page has 10 ⓘ triggers, so key facts need a tap each. | Major | Finding 3 |
| IA-10 | Five flat accordions, starting with 「よくある質問」; firmness sits inside the English-titled 「Specifications」, and 「保証」 holds one line. | Major | Finding 4, App. A2 |
| IA-11 | Heading outline is flat: story headings at the same level as comparison and reviews; accordions without headings; an English H3. | Minor | New (App.) |
| IA-12 | Sticky in-page anchors run in reverse (Q&A / 口コミ / 詳細) and skip the comparison. | Minor | App. A2 |
Proposed structure
Each page gets one job, and the decision and trust content moves forward. These are structural directions for the mock-up, within the current template and brand system.
TOP PAGE
Out of scope in this example: no homepage was reviewed.
ソファ CATEGORY PAGE
Out of scope in this example. The menu proposal in section 02 gives each sofa the fixed descriptors (座り心地 · 座面 · price) a category page would reuse.
PRODUCT PAGE
- Buy boxRating, name, one-line “who it's for”, price (税込), firmness scale, sizes with width and price, named colours, carry-in size, CTA with returns sub-line
- Trust blockArrival estimate, returns, warranty, fabric samples, payment, chat, “see it in the showroom”
- Review summaryHistogram, attribute filters in Japanese, picked reviews
- Compare (short)Key rows vs Ren and Mio, link to the full table
- One brand storyEach claim once
- Details (5 groups) and FAQSize and carry-in, materials and care, delivery and returns, warranty; buying questions first
- Second purchase point, cross-sellAfter the story
Additional findings to confirm
Further localisation and UI items found during this analysis. Some may have business reasons we can't see from outside the company, so we would like to confirm them with you before adding them to the localisation review.
| # | Finding | Japanese context / reference | Suggested priority |
|---|---|---|---|
| N1 | Prices never say 税込. The product page shows 「124,000円」 only, in the buy box, the sticky bar and the comparison. | Total-price display (総額表示) has been mandatory since April 2021, so the amounts are presumably tax-inclusive already. The label is still the Japanese convention that lets shoppers compare confidently: LOWYA shows 「(税込)」 next to the price, and IKEA Japan 「消費税込み」 under it; Apple Japan does not, so it is a common convention rather than a rule. | 中 |
| N2 | Claims have no 注釈 (basis, date). 「人気No.1」 on the main photo and 「購入者満足度98%」 in the story carry no footnote. | The Consumer Affairs Agency published a report on No.1 claims in 2024 setting out what a fair basis requires, after a series of enforcement orders. Apple Japan footnotes claims throughout and Dyson Japan uses ※ notes. Worth a legal check. | 高い |
| N3 | No route to see the sofa in person from the product page (IA-2). | Mareo has a showroom, but a shopper deciding on the product page never learns about it. LOWYA, Re:CENO and unico all link a store check next to the cart button. | 中 |
| N4 | Support channels are hard to find. Phone appears only in the footer; LINE only as a social icon; FAQ is external. The floating chat button states no hours. | For a 100,000–150,000円 purchase, a visible contact route reduces hesitation. LINE is a primary customer-service channel in Japan; LOWYA puts it in the menu and its call-centre hours in the footer. | 中 |
| N5 | The legal notice is labelled 「ご利用ガイド・規約」 instead of the standard 「特定商取引法に基づく表記」. | Japanese shoppers and reviewers look for the standard label (LOWYA uses it in its menu and footer). Content is probably compliant; it's a findability issue. | 低い |
| N6 | No express checkout on the product page (Apple Pay, Amazon Pay, 楽天ペイ). | Checkout not inspected, so unconfirmed. Re:CENO shows R Pay and Amazon Pay under its cart button. Relates to finding 7 (payment visibility). | Verify |
| N7 | Screen-reader labels. Not assessed beyond the appendix: seven of eight product images lack alt text and the ⓘ icons have no names. | General accessibility; appendix. | 低い |
Context we should keep in mind
Some benchmark patterns exist for reasons that don't apply to Mareo. LOWYA's dense buy box serves a large catalogue, and its postcode delivery date relies on its own logistics data, which Mareo would need before showing a date. Re:CENO's in-room delivery is part of its service, not a display choice. We cite these sites for the specific pattern, not as a whole model. Where the reason does transfer (a sofa has to fit through a Japanese doorway whatever the delivery terms), we say so in the finding.
Method
Navigation trees were extracted from the DOM. Section maps were measured at 390 × 844 using each section's position and height on the pages (Mareo, LOWYA) on 18 November 2026. Journeys were walked on the page; cart and checkout were not entered. The analysis follows a three-layer IA review (structure, labels, disclosure).