futureshop Not in Search: Four Checks
futureshop Not in Search
Four Checks
| Category | Read Time | Topic | Description | Source | Copyright |
|---|---|---|---|---|---|
| Tips | 9 min read | Japan EC platform guides for search and AI search | futureshop outputs structured data and an XML sitemap automatically, but broken commerce creator headings, duplicate URLs from multi-group registration, missing Article markup, and WordPress posts lef... | searchos.io/en/blog | © 2026 Search OS |
futureshop automatically outputs breadcrumb, product, review, and site-search structured data, plus an XML sitemap. That coverage is wider than other Japanese hosted carts, and for product pages it is a sufficient starting point. Even so, the questions keep coming: a landing page built in commerce creator is never cited by AI, Search Console fills up with "Duplicate" entries, only the WordPress-integrated posts fail to get indexed.
The cause is not a gap in futureshop's features. It clusters in four places outside the automatic output: heading structure, product URLs and canonical, the scope of structured data, and the sitemap for WordPress posts. This article puts those four checks in order. Numbers and specifications come only from the individual articles on each topic.
1. On commerce creator pages, can the headings be read as a table of contents
commerce creator builds pages by stacking parts. That makes it easy to produce a page where the "large text" you see and the HTML headings (h1 to h3) don't match. Search engines and AI search read structure from heading tags, not from appearance, so a page without headings is treated as one long paragraph, and nobody can tell where the answer is. The standard is one h1 per page, three to six h2s, and h3s only under an h2. If the page template already outputs an h1 (the shop name or page title), start the top heading inside the page at h2.
The check is simple: open the page in developer tools, list only the h1, h2, and h3 elements from top to bottom, and read them. If the list is nothing but "NEW", "Recommended", and "See more", the structure is broken.
Breakage pattern | What happens | Fix |
|---|---|---|
No h1 | The page's subject isn't conveyed | Make the template or topmost heading part the h1 |
Multiple h1s | The subject is split | Change the second and later ones to h2 |
No h2, only h3 and h4 | The hierarchy can't be read | Make the first heading of each section an h2 |
Decorative large text set as h2 | "Notice" and "NEW" become headings | Change to a text part plus CSS |
Headings exist only inside images | No heading exists | Put a text heading below the image |
Whether each part lets you set a heading level is something to confirm in the commerce creator part specifications. For parts that don't, combine them with a heading part. When you duplicate a landing page to build a new one, the old page's subject stays behind, so don't leave its headings as they are.
2. For products registered in multiple groups, does the canonical point to the URL without a group
futureshop lets you register a product in more than one group. If the URL carries group information when a visitor reaches the product through the group hierarchy, the same product ends up with three URLs: via group A, via group B, and with no group at all. Search engines decide on their own which one is real, and inbound links and search performance spread across several URLs. AI search also becomes less likely to pick a page as a source when several pages carry the same content.
To see whether this is happening in your shop, open the same product from different groups and compare the URLs. Also check whether the Search Console pages report shows "Duplicate without user-selected canonical", and where rel="canonical" in the product page source points.
State | Action |
|---|---|
The product URL contains a group ID and differs by group | Point the canonical to the product URL without a group. Align internal links to that URL too |
One product URL, but group listing pagination is duplicated | Self-referencing canonical on each listing page. Don't consolidate to page 1 |
The canonical points to a group-based URL | Fix the target. Check every product |
No duplication | Do nothing. Check "Duplicate" in Search Console once a month |
Even with the canonical aligned, the effect weakens if links from group listings and related-product parts still use the group-based URL. noindex is an instruction not to index and doesn't consolidate search performance, so consolidation is done with canonical. Register products only in groups that are meaningful classifications, handle temporary groups like "Sale" and "New" with tags or feature pages, and keep the hierarchy to three levels.
3. Does the automatic structured data match the screen, and does it reach Article
The automatic output covers "basic product information" and no more. Article for content pages is outside the automatic output, and the automatic sitemap includes only pages created in the admin panel. On top of that, invalid structured data can be worse than none, because an error can cause the whole block to be ignored. Run four kinds of URL through the Rich Results Test (product page, category page, feature page, top page) and record the detected types, errors and warnings, and whether Offer price and availability match the price and stock shown on screen.
Priority | Item | Reason | Where to implement |
|---|---|---|---|
1 | Consistency of price, stock, and options | A mismatch is a direct reason for exclusion from AI product recommendations | Product data and template |
2 | Article + BreadcrumbList on content pages | Prerequisite for comparison and buying-guide articles to be cited | Template head |
3 | FAQs and comparison tables as visible HTML | AI can't reliably read hidden information or text inside images | Body |
4 | Ongoing verification | Template updates and added apps break the output | Operations |
The easy one to miss is a product that is on sale but still carries template-derived "out of stock" text. AI can drop the product from its recommendation candidates; at a Korean ecommerce brand with roughly ₩40 billion (about ¥4 billion) in annual revenue, removing this hidden out-of-stock text was where the improvement started. For FAQs, Google restricted FAQ rich results mostly to authoritative sites such as government and medical, so prioritize FAQs as visible HTML over FAQPage structured data. When you change the template, run the Rich Results Test the next day.
4. Are WordPress-integrated posts connected to the shop through the sitemap and internal links
The sitemap futureshop generates covers only pages created in the admin panel. Posts written in WordPress live outside the admin panel, so they aren't included. Search engines can find posts by following internal links, but pages missing from the sitemap are discovered later and their updates go unnoticed longer. The fix is to register the WordPress sitemap (either WordPress's standard /wp-sitemap.xml or the one an SEO plugin generates, but only one of them) in Search Console separately from the futureshop sitemap.
WordPress placement | Search Console property | Registration |
|---|---|---|
Subdirectory on the same domain as the shop (/blog/) | Same property as the shop | Add the WordPress sitemap to the same property |
Subdomain (blog.example.jp) | Same if a domain property; separate if a URL-prefix property | Consolidate to a domain property and add it there |
Separate domain | Separate property | Register under the separate property |
A sitemap only helps discovery; it says nothing about how posts relate to products. Link from post bodies to product and group pages, from the end of product descriptions to related posts, and share the header and footer. On the post side, confirm that the canonical points to itself, that Article carries the headline, publish date, modified date, and author, that Organization appears only on the shop's top page, and that tag and date archives are noindex while they are thin.
Search OS is the layer that keeps checking whether the four points still hold today
Search OS supports existing shops, including those on futureshop. It doesn't replace building in commerce creator or the automatic output. It continuously verifies the state of page heading structure, metadata, structured data, canonical, and sitemaps, and confirms from bot logs whether search engines and AI crawlers actually fetched the pages. Places that disagree with visible text, content pages missing Article, canonicals pointing the wrong way, and broken internal links are organized as prioritized fix items.
Questions to check first
When you pull out only the headings of your main landing page, do they read as a table of contents, and is the template already outputting an h1
When you open the same product from a different group, is the URL the same, and if not, does the canonical point to the URL without a group
Do Offer price and availability on your main products match the price and stock on screen
Do feature pages and blog posts output Article and breadcrumbs
Is the WordPress sitemap registered in Search Console separately from the futureshop sitemap
Do recent posts link to related product pages, and do product pages link back to posts
Which is more recent: the day you last changed the template, or the day you last ran the Rich Results Test
Conclusion
futureshop's automatic output is a sufficient starting point, but heading structure, canonical targets, agreement with visible text, and the sitemap for WordPress posts are yours to check. In an experiment tracking 81 pages on a high-authority domain for 30 days, Google AI Mode citations fell from 59% in the first week to 26% on day 30, and pages not cited in the first week were rarely cited afterward. Whether a page is readable in the first few weeks after publishing shapes what follows, so check the four points before publishing and the day after any template change.
Related reading
Duplicate URLs from futureshop multi-group registration and canonical design
The scope of futureshop's automatic structured data and what to add for AI search
futureshop-integrated WordPress posts left out of the sitemap: how to register them separately