jp.dare.uk + cn.dare.uk — locale-pattern POC on the dare test bed (parked 2026-06-02)
DARE.CO.UK · PARKED SKETCH · 2026-06-03
Mirrored from ~/.claude/.../memory/parked_sketch_dare_uk_locale_poc_2026-06-02.md. This is a design sketch parked for future build — read for context, not as a current deliverable.
Two sandbox subdomain POCs to validate a portable locale/i18n primitive. Hosted on dare.uk (the short alias of dare.co.uk) as test bed; pattern graduates to jp.audreyinc.com / cn.audreyinc.com when commercial reason arrives. Resume when either commercial signal lands or pattern is needed by a sibling site.
Why dare.uk and not audreyinc.com
Established pattern: dare = test bed, audrey = commercial leverage (see user_dare_test_bed_to_commercial_leverage.md). New patterns prove themselves on dare’s surface first — no commerce-stack risk, no customer-visible breakage, freedom to break and rebuild. Once proven, lift cleanly to audreyinc.com when there’s a real commercial reason.
dare.uk is the short alias of dare.co.uk — same site, same content, same static-baked stack. Using the short alias here also dogfoods the short-alias pattern for a locale subdomain shape (jp.dare.uk vs jp.dare.co.uk — shorter wins).
Why this exists at all
Audrey is Chinese/Malaysian (Mandarin + English native). audreyinc.com already sells into Japan; China is not yet entered. Both markets are obvious adjacent-locale targets for audreyinc, but neither has been validated as a build-out.
The POC isn’t about driving traffic — it’s about proving the localisation primitive once on dare, so any future audreyinc locale (jp/cn/ms/de/etc.) becomes a same-day deploy when there’s a commercial reason.
The two POCs
| Subdomain (test bed) | Locale | Language | Graduation target | Discovery surface |
|---|---|---|---|---|
jp.dare.uk |
Japan | Japanese (ja-JP) | jp.audreyinc.com (audreyinc already sells in JP) |
Google (no GFW) |
cn.dare.uk |
Mainland + Mandarin diaspora | Simplified Chinese (zh-CN) | cn.audreyinc.com (future market) |
Google (TW/HK/SG/MY) + RED + (mainland needs ICP later) |
Both subdomains stay on the same dare.uk static-baked stack — no separate property, no rerouting, no commerce-stack involvement during POC phase.
POC bar — what the sandbox actually proves
| Question | Reusable downstream (on audreyinc + any sibling) |
|---|---|
How does hreflang en ↔ ja-JP / zh-CN map on a static-baked dare page? |
Any future locale |
| LLM translation quality at scale — Opus vs Sonnet vs Haiku per locale? | All future locales |
| Typography stack — Noto Sans JP / Noto Sans SC, line-height, font-display tuning? | Per-locale typography primitive |
| Subdomain SEO isolation — does cn. / jp. affect en.* rankings in GSC? | Confirms subdomain-per-locale arch |
| Deploy pipeline — fork the en build, swap translation layer, ship? | Becomes dare-locale-build.py primitive, ported to audreyinc when needed |
Minimal POC scope (when resumed)
- Pick 3-5 well-fitting dare.co.uk pages (Methods / Field Notes — editorial, not commerce-specific)
- LLM first-pass translation per locale (Opus for quality, compare with Sonnet/Haiku cost)
- Audrey eyeballs zh-CN output for slop (jp-JP needs a Japanese-language reviewer — gap to flag)
- Deploy each subdomain — hreflang round-trip, Noto Sans loaded · no commerce on dare = cleaner test, isolates the locale primitive from commerce concerns
- Document the build pattern in a new memory entry
- Stop. No RED account, no Baidu, no traffic measurement — those are commercial-mode work for the audreyinc graduation phase.
Resume conditions
Resume the sandbox when ONE of:
- Audrey-side commercial signal: existing JP customer of audreyinc asks “do you have a Japanese site?” / JP order volume justifies localised checkout / CN market entry goes from idea to plan
- Pattern demand from a sibling: any portfolio site needs i18n and we’d rather lift a proven primitive than invent one
- Audrey-time available for the jp-JP slop-check gap (Japanese-language review needs sourcing — or accept LLM-only quality bar for sandbox)
What NOT to do during sandbox
- Don’t graduate to audreyinc.com prematurely — keep the experiment on dare until pattern is proven
- Don’t seed RED / WeChat OA / Weibo (commercial-mode distribution work)
- Don’t apply for ICP licence (commercial-mode mainland-entry work)
- Don’t measure traffic / sessions / conversions (POC bar is “pattern works”, not “people came”)
- Don’t translate the whole site (3-5 pages is enough to validate the pattern)
Graduation path (out of scope for POC, here for shape)
Once dare-side POC proves the locale primitive:
1. Port dare-locale-build.py → audreyinc-locale-build.py (likely 80% reuse)
2. Stand up jp.audreyinc.com with real product/collection pages (commerce-aware)
3. Layer currency display (¥), checkout localisation, jp-specific payment rails (Konbini, JCB) — Shopify handles most of this with Markets
4. Distribution work: jp-language paid + organic, eventually a JP-specific creative direction
5. cn.audreyinc.com follows same shape if/when mainland entry becomes real (ICP + Aliyun + Alipay/WeChat Pay + Xiaohongshu/RED + WeChat OA)
Cross-references
user_dare_test_bed_to_commercial_leverage.md— the parent pattern; this sketch is a direct instance of itfeedback_portfolio_subdomain_pattern.md— subdomain-per-purpose convention this fits insidefeedback_short_alias_redirects_proactive.md— short-alias preference (dare.uk over dare.co.uk for the subdomain shape)feedback_short_subdomain_collision_expand_by_function_2026-06-01.md— caveat: jp/cn are 2-letter codes, confirm no collision with other gf.cx/portfolio uses before standing them upuser_audrey_commercial_roi_filter.md— the lens that promoted this from idea to parked sketchproject_audreyinc_beta_foundation.md— the commercial target this graduates toproject_audrey_business_identity.md— Audrey’s commercial brand identity
Status
- 2026-06-02: Parked. Sketch on file. No code, no DNS, no translation. Surface = dare.uk (NOT audreyinc.com). Resumes on any of the conditions above.