00 — Master Synthesis | BSH bshsg.com DIALux Product Strategy

§1. One-Page Decision Summary

For someone in a hurry. Read this box and you have 80% of the document.

The strongest product framing — in 3 sentences

"BSH Flat Plan" — a fixed-fee SGD 49–99 photoreal Blender-Cycles render + lux heatmap + bill-of-materials, sized for the ~30 standard HDB floor plans, delivered in 24 hours and priced as the brief the homeowner brings to their ID firm rather than as a competing service. The product wraps DIALux-grade engineering authority around the Lutron/Koizumi "named scene grid" UX, leverages BSH's existing EMA Class L1 licence as the trust moat, and converts upsell through a bundled LEW install slot (where BSH's 35-year contracting business already lives). It is the only Singapore offer that sits between the SGD 15 lighting-shop chit-chat and the SGD 250 ID-firm 3D-render-per-view at a price the market does not currently address.

The single most important "do this Monday" action

Digitise five HDB floor plans — start with the 4-room BTO (the volume sweet spot at ~50% of 2024 launches), 5-room BTO, 4-room resale (1990s), 5-room resale, and 2-Room Flexi. Use the existing bshsg.com Three.js + IES infrastructure that's already 80% built. Each plan becomes a Blender scene with locked geometry, locked furniture, and 2–3 swappable lights. Target output: 14-second photoreal render at RTX 5070 (already proven). This single 1–2 week sprint unlocks every framing in Section 5 — without it, every framing is theoretical.

The top 3 risks (ranked)

Risk 1 · ID-firm hostility

The Singapore renovation market is ID-firm-gated. If BSH positions as a replacement, IDs blacklist BSH SKUs from specs. Mitigation: position the Flat Plan as the brief homeowner hands to their ID — cooperative, not disruptive.

Risk 2 · Coohom/Foyr/Kujiale embed undercut

A VC-funded Chinese platform (Coohom has free tier, 1M+ floor plans, IPO'd HKD 35B) could land in SG with a localised version in 12–18 months. BSH window is finite. Mitigation: ship before they localise; build the LEW install moat they can't replicate.

Risk 3 · No-neutral-wire compatibility landmine

80% of pre-2015 HDB switch boxes have no neutral. Smart-switch SKUs that ignore this generate 1-star reviews and rework costs. Mitigation: mandatory "needs neutral / no-neutral compatible" badge on every smart SKU; default catalogue to no-neutral-required.

A single hero image suggestion

A two-pane side-by-side: left = a typical Singapore 4-room HDB living room at dusk with the original developer ceiling rose + single bare fluorescent batten (gloomy, flat, "ang moh standard" lighting). Right = the same room with the BSH-spec'd 4-room Japandi kit — 6 × 6W 3000K downlights at 1.2m grid + 2700K cove + dimmable dining pendant + 4000K kitchen task — rendered photoreal in Blender Cycles with the homeowner's actual sofa, dining table, and 2.6m HDB ceiling locked into the geometry. Caption underneath: "Same flat. Same furniture. SGD 49 plan." No marketing copy required.

§2. The Singapore Lens — non-negotiable filter

From Report 9 (the canonical Singapore Bible). Every foreign-market idea — every Chinese factory SKU, every Japanese aesthetic, every Indian price-point, every Korean feature — runs through these ten questions before getting onto a bshsg.com PDP or roadmap. The Lens is not aspirational; it is the gate.

Is it 220–240 V, 50 Hz? Anything 120 V or 60 Hz-only is dead on arrival. Universal-input drivers are fine.

Does it fit a 2.4 m post-false-ceiling clearance? Drops, beam-angles, cove depths must all assume tropical-HDB compression, not US 8-ft ceiling assumptions.

What is the CCT — and does the SG consensus accept it in this room? 3000K living / 4000K kitchen are the local defaults. 2700K is "old-feeling" outside cove and bedside. 6500K is "hospital".

Does it survive RH 70–90% year-round? Especially drivers and any outdoor / bath / kitchen fixture. IP65 outdoors, IP44 in Zone 2 bath.

Is it BS 1363 Type-G plug-ready (DIY SKUs) or hard-wire-ready with safety chain (LEW SKUs)? US Type-A plugs and EU Schuko are non-starters.

Does it work without a neutral wire at the switch? Or — is the requirement clearly labelled? 80% of resale HDB switch boxes are neutral-less.

Is the smart protocol Zigbee 3.0, Wi-Fi 2.4 GHz, or BLE/Thread/Matter? Z-Wave SG band is 920–925 MHz, niche. Wi-Fi 5 GHz is concrete-weak.

Does it stand up to NEA MELS labelling? Lumens, lamp life (h), Ra, CCT must be on the box. Aim for 3-tick minimum, 4-tick for hero SKUs. Climate-Voucher eligibility is gold.

Does the aesthetic align with Japandi / Modern Asian dominant 2024–2026? Or with industrial-warm, heritage-Chinese, Scandi niches. Pinterest-pretty doesn't always survive HDB scale.

Will an HDB Renovation Permit + EMA-LEW pathway accommodate it? If the install needs hacking a structural wall, embedding LED in load-bearing concrete, or hanging a 30 kg chandelier on a beam-less slab — flag it.

Worked example — applying the lens to one foreign idea

Test case: "Partner with Coohom (KuJiale's international 3D engine) — embed their browser SaaS on bshsg.com"

From Report 5 (China). Coohom claims 1M+ 3D models, 30M+ furniture library, 8K rendering, sub-1-minute renders, 720° walkthroughs, free tier. Same parent (Manycore, HKEX 00068.HK) as the platform serving 86.3M MAU across 200+ countries. Plausible to integrate via API or iframe.

Lens question	Verdict	Reasoning
Q1 · 220–240V / 50Hz	PASS	Coohom is software — voltage-agnostic. SKU library can be filtered.
Q2 · 2.4 m clearance	ADAPT	Coohom defaults to ~2.7m ceilings. Must hard-code HDB 2.6m presets + 200mm cove offset. Solvable but not out-of-the-box.
Q3 · CCT consensus (3000K living)	ADAPT	Coohom render engine accepts any CCT input but defaults to international 2700K. SG defaults must be set explicitly.
Q4 · RH tolerance	PASS	Not relevant to software. Only fixture catalogue selection.
Q5 · BS 1363 plug	PASS	Software-agnostic.
Q6 · No-neutral compatibility	ADAPT	Coohom doesn't model switch wiring. BSH must overlay this as a separate spec layer.
Q7 · Smart protocol	PASS	Visualization only — protocol-agnostic.
Q8 · MELS labelling	ADAPT	Coohom's catalogue is Western/Chinese mass-market. BSH must curate Climate-Voucher-eligible SG SKUs separately.
Q9 · Japandi/Modern Asian aesthetic	PASS	Coohom's strongest aesthetic library is exactly this (KuJiale serves the Chinese Japandi market — heavy overlap with SG taste).
Q10 · HDB Permit pathway	FAIL by default · solvable	Coohom doesn't surface HDB-banned operations (structural hacking, cove cutting in newer BTOs). BSH must add a compliance layer on top.

Final Lens verdict: ADAPT, not fail. Coohom is feasible as a backend rendering engine if (a) HDB floor-plan templates are pre-loaded, (b) SG-default CCT presets are forced, (c) catalogue is curated to BSH-stocked + Climate-Voucher-eligible SKUs only, (d) compliance overlay flags banned HDB operations. Estimated adaptation work: 4–6 weeks. Cost: under S$30K if done in-house against existing bshsg.com Three.js stack.

This is the analytical pattern Steven should apply to every foreign idea in this document. A binary "use Coohom yes/no" is not a useful question. "Coohom passes 5/10 Lens questions natively, fails or needs adaptation on 5/10, total adaptation cost ~S$30K, no other alternative passes more cleanly" — that is a usable answer.

§3. Cross-report convergence map — 7 patterns that emerged independently across 3+ reports

Where multiple independent reports — Singapore, China, Japan/Korea, India/SEA, the global competitive landscape, the portfolio study — landed on the same conclusion without coordinating, that's the strongest possible signal. Below are the seven patterns I found.

C1 · The same-room named scene grid is the strongest layperson UX

Reports that surfaced this independently: 02 (portfolio · Lutron Seaholm Penthouse 4-tile master bath grid), 04 (translation patterns · Pattern #2 same-room photo triptych + Pattern #4 mood-name façade), 05 (China · Aqara Scene Panel S1 + Mi Home 3D auto-scenes), 06 (Japan · Koizumi Virtual Lighting Simulation 4 named "life scenes" — breakfast/focus/relax/movie + Iris Ohyama 6畳/8畳 tatami sizing), 08 (SG · Lemon8 regret threads convert through narrated scenes not specs)

Synthesised claim: When a non-technical buyer sees the same room photographed/rendered three or four times with different named scenes ("Saturday breakfast", "Movie night", "Late study", "Reunion dinner"), they understand layered lighting without reading a word. This pattern beats every spec-sheet, every CCT swatch strip, every lux calculator, every isolux contour. It is the single most reproducible UX across all studied geographies. Crucially, Singapore-specific scene names (CNY reunion / Hari Raya open house / Saturday prata / kids' study) emotionally outperform Western names ("Read", "Relax").

Singapore Lens verdict: PASS — All 10 Lens questions are satisfied because the format is fixture-agnostic. Pattern is portable as-is. Must culturally adapt scene names to SG festive calendar + bilingual Mandarin/Singlish labels for older HDB buyers.

C2 · "Show the buyer their own room, not stock photography"

Reports that surfaced this: 01 (DIALux outputs · photoreal raytrace render rated Tier-A, #1 of 13), 02 (portfolio · 5/5 stars for John Cullen Leaf House 7-room journey narrative + Behance freelancers who succeed are those who render in furnished context, never empty isolines), 03 (competitive · Modsy's $72M failed startup proved demand for "my actual room"; RoomGPT did 6M+ generations with this hypothesis), 06 (KR · Ohou + Archisketch 96% of Korean apartment blueprints — homeowners pull their literal unit), 07 (India · Asian Paints Beautiful Home app proves Indian/Asian buyers will photograph their own walls + Livspace render-vs-reality side-by-side is the trust currency)

Synthesised claim: The single highest-conversion artifact in residential lighting commerce, globally, is "here is your actual room with the chosen fixtures lit at the chosen CCT". Stock photography no longer converts. AI-generated rooms convert (RoomGPT) but lose trust when buyers realise the furniture isn't buyable. The killer combo is buyer's actual floor plan + buyer's actual SKUs + photoreal render. Singapore has the unique advantage that ~30 HDB floor plans cover 80% of the population (Korea-like apartment monoculture; not the 1M unique-plan US/UK market).

Singapore Lens verdict: PASS — Singapore's standardised HDB floor plans make this more tractable than any of the source markets except Korea. BSH has already proven Blender Cycles + IES + Mac Mini server can produce 14-second renders at DIALux-grade quality. This is not technical risk — it is execution risk.

C3 · Brands sell SKUs, services sell rooms — the wedge is between

Reports that surfaced this: 03 (competitive · whitespace gap analysis — Modsy died trying to be a service; furniture sites avoided photometry; pro tools never reach consumers), 05 (China · NVC's "one household, one solution" + Yeelight Pro CAD+DIALux for consumers), 06 (Japan · Koizumi's life-scene framing + Iris Ohyama sized-by-room), 07 (India · 8/8 lighting brands ship filter-grid catalogues with zero room context; Livspace/HomeLane/DesignCafe sell rooms via 3D), 08 (SG · every Singapore retailer surveyed — Hooga, Million, TLG, Threecubes, Delight, Hipvan, Castlery — sells lighting as commodity SKUs; not one offers photometric room visualization)

Synthesised claim: There is a structural gap between brand-direct (SKU filter grid, no visualization, e.g., Havells, Hipvan, Million Lighting) and turnkey design service (full ID renovation, $50K+, e.g., Livspace, Carpenters, akiHAUS). Nobody — globally, in any of the 6 studied markets — sits in the middle as a productised "I sell SKUs but I also sell the design plan for your room" offer at consumer pricing (SGD 49–99). Soraa died trying to sell premium CRI-95 bulbs without selling the perceived difference; Modsy died trying to render real rooms at $300+ unit cost; Singapore retailers compete entirely on price. The wedge is photometrically-credible room visualization at consumer pricing, tied to a buyable BOM.

Singapore Lens verdict: PASS — This is the single biggest commercial finding. BSH is uniquely positioned: it already has the EMA Class L1 contractor licence (regulatory authority), the e-commerce surface (bshsg.com), and a 35-year customer track record. No Singapore competitor has all three. The biggest competitive threat is not domestic — it is a Coohom or Foyr Neo international expansion.

C4 · Reveal the material, hide the source — "see light, not fixtures"

Reports that surfaced this: 02 (portfolio · Light Collab The Reserve onyx + Nipek "light fixtures should not be seen" + John Cullen "floating marble" + LPA "learn from natural light" — 4 of the 18 case studies independently arrived here), 05 (China · 见光不见灯 [see light, not fixtures] is the dominant Xiaohongshu trend), 06 (Japan/Korea · 多灯分散 tatō bunsan + Korean K-drama trope "ceiling fixture off, table lamp on")

Synthesised claim: The premium aesthetic globally — and especially across East Asia (China, Japan, Korea, Singapore) — is converging on hidden sources, indirect light, glowing materials. Cove lighting is now table stakes in HDB renovation (Report 9 Section C18 — "80% of renovated HDBs have cove"). Magnetic track + recessed downlight + accent cove is the 2024–2026 dominant signature. Implication for BSH's catalogue: photograph the kitchen marble glowing, not the downlight; photograph the headboard washed warm, not the wall sconce. The lighting product is the effect on the material, not the fixture body.

Singapore Lens verdict: ADAPT — Passes 8/10 Lens questions but flags Q10 (HDB Permit pathway). The 无主灯 (mainless / no-main-light) backlash in China is real and well-documented: high infrastructure cost, glare risk, maintenance hell when drivers fail inside ceilings, wall flaws magnified by cove grazing light. BSH should adopt the aesthetic but publish honest TCO + a "is mainless right for you?" qualifier (which is itself a content marketing engine — Lemon8/Singlish translation of Chinese backlash).

C5 · Calculator-first funnel + render-vs-reality side-by-side is the proven trust loop

Reports that surfaced this: 03 (competitive · ERCO Track Configurator was rated closest pro-world UX to consumer needs), 04 (translations · Pattern #1 everyday-ladder anchor + "how many lights do I need?" calculator), 07 (India · Livspace + HomeLane SpaceCraft + DesignCafe — three Indian unicorns all use calculator → 3D render → quote funnel; render-vs-reality photo is the published trust signal across all of them), 08 (SG · Qanvast Renovation Calculator is the most-used existing tool but stops at price; no Singapore tool produces a render)

Synthesised claim: The consistent winning funnel — from India's $1B+ Livspace to Malaysia's Recommend.my to Indonesia's Dekoruma — is: (1) free calculator on landing page → (2) budget range + 3D render preview → (3) WhatsApp/email CTA → (4) live design session → (5) phased payment → (6) project delivery → (7) render-vs-reality testimonial post. Every step is reduced friction. The Asian buyer is fundamentally chat-led (WhatsApp/Line/KakaoTalk) and visual-led (render before reading). The Western funnel (long-form copy, signup, multi-step form) does not transfer.

Singapore Lens verdict: PASS — Singapore consumer behaviour mirrors the India/SEA pattern. Qanvast already trained Singaporeans to expect a renovation calculator. BSH's existing bshsg.com floor-plan estimator + EV consult tier (SGD 15) is the first two steps of this funnel. The missing pieces are: render output, render-vs-reality archive, WhatsApp Business floating button, phased-payment policy.

C6 · Engineering credibility wraps consumer-language outputs

Reports that surfaced this: 01 (DIALux · Documentation PDF rated Tier-A "as a signal, not content"; nobody reads the 60-page tender pack but the existence of it changes the price tier), 02 (portfolio · Habitat by Honestbee's single CRI97 line beat every technical chart), 03 (competitive · Soraa had the best LEDs in the world and died because they couldn't make buyers feel CRI 95+), 04 (translations · Pattern #10 the 5-second phone test — empower the buyer to verify the engineering claim themselves), 07 (India · Design Matrix uses "$120M+ specification value" headline number on homepage)

Synthesised claim: The buyer doesn't want to read engineering. They want to know engineering is behind the product, and they want to be able to verify it in 5 seconds with their phone. The pattern: lead with the photo and the mood word; footnote the SS 531 lux compliance, the Ra ≥ 90, the IES file download. Hard credibility numbers ("35 years of Singapore electrical contracting", "$X million in residential lighting installed", "EMA Class L1") beat soft "trusted by clients" copy.

Singapore Lens verdict: PASS — This is BSH's natural moat. As an EMA Class L1 licensee with 35 years of HDB experience, BSH can author the regulatory narrative no other Singapore retailer can. Million Lighting can claim 40-year heritage but cannot stamp an SS 531 compliance certificate. The DIALux output is the credibility layer; the photoreal render is the consumer layer. Both are mandatory.

C7 · The HDB-floor-plan dataset is the unbuildable moat

Reports that surfaced this: 03 (competitive · whitespace gap analysis — "Singapore sun path, HDB ceiling heights, Singapore-stock fixtures" — "100m moat once built"), 06 (KR · Archisketch holds 96% of Korean apartment blueprints — this is the entire moat that powers Ohou's 20M downloads), 07 (India · Foyr Neo is the technological proxy but ships generic floor plans), 08 (SG · HDB has <30 standard floor plans across all blocks built since 1990 — 4-room = 50% of supply, 5-room = 25%)

Synthesised claim: The Korean equivalent (Archisketch) took years to amass 96% of apartment blueprints and is now the entire moat behind Ohou. Singapore has the opposite problem: only ~30 standard plans across all post-1990 HDB stock, technically achievable by one CAD specialist in 6–8 weeks at under S$40K. Any HDB owner can search "Punggol 4-room standard" and pull their exact unit. No competitor (Hooga, Million, TLG, Threecubes, Delight) has built this. Coohom/Foyr could build it but would need ~12 months of Singapore-specific work to localise.

Singapore Lens verdict: PASS — This is the single highest-leverage idea in the entire 9-document study. Cost: not building it: if Coohom localises before BSH ships, the window closes.

Convergence summary table

Pattern	Independent sources	Lens verdict	Lens questions failed	Priority
C1 · Named scene grid UX	5 reports	PASS	0	P0
C2 · Buyer's own room rendering	5 reports	PASS	0	P0
C3 · SKU/service wedge	5 reports	PASS	0	P0
C4 · Hide the source, show the material	3 reports	ADAPT	Q10 (HDB mainless backlash)	P1
C5 · Calculator + render-vs-reality	4 reports	PASS	0	P0
C6 · Engineering wraps consumer-language	5 reports	PASS	0	P0
C7 · HDB-plan dataset as moat	4 reports	PASS	0	P0 — highest leverage

§4. Cross-report contradictions & tensions

Where reports disagree explicitly, or where unresolved trade-offs need a judgment call. Six tensions need Steven's decision before product strategy can finalise.

T1 · Build vs partner with Coohom/Foyr/KuJiale

Argue: PARTNER / EMBED

Report 5 (China) makes the case explicitly — Coohom has 30M furniture models, 8K render in <1 min, free tier, Asian aesthetic library that already serves Chinese Japandi. Report 7 (India/SEA) confirms Foyr Neo has a Singapore office and offers SaaS at ~$100/mo. BSH's competitive timing window is finite (~12 months). Build-from-scratch with Three.js + Blender Cycles + IES + custom UI is technically achievable but consumes 4–6 months of engineering time that could be spent on commercial-side build (LEW capacity, Climate Voucher registration, content production).

Argue: BUILD

Report 3 (competitive · Calculux cautionary tale) warns single-vendor tools die when the parent pivots. Coohom is Hong Kong-listed (Manycore 00068.HK) — if they decide to compete with BSH directly in SG (Coohom has a SG-localised product roadmap), BSH becomes a downstream reseller of someone else's funnel. Report 5 lesson: "Even when ecosystem is mature, don't build a proprietary one — but DO own the install layer". BSH's data (HDB floor plans + IES library + 1,686 CGD files) becomes platform leverage that compounds.

Recommendation (opinion, not option): Hybrid. Use Foyr Neo internally for the live 1-hour design session (Foyr SG office + $100/mo per designer + Singapore-time-zone support). Build bshsg.com's consumer-facing self-serve layer on Three.js + Blender Cycles batch-rendered Singapore floor plans (using existing 14-second render pipeline already proven on RTX 5070). The customer never sees Foyr; they see BSH-rendered images. Foyr is the internal productivity tool for the designer-led tier. This isolates BSH from Coohom/Foyr platform risk while accelerating Year 1 delivery. Decision deadline: 30 days. Cost differential: ~$30K (Foyr seats × 12 months) vs ~$200K (full custom build with Three.js + UI). The hybrid is ~$60K.

T2 · Cooperate with ID firms vs disintermediate them

Argue: COOPERATE

Report 8 (SG) is unambiguous: "Position against the ID firm — not in conflict with it. The ID is the gatekeeper. Frame the BSH DIALux output as the brief the homeowner brings to the ID, not as a replacement for the ID's role. The hostile message fails — the cooperative message wins." Lemon8 evidence shows homeowners save ~8% by walking into ID meetings with a plan. Existing ID firms (15+ profiled) are not the enemy — they are upstream of every BTO homeowner. Burning the channel kills volume.

Argue: DISINTERMEDIATE

Report 8 (SG) also shows: ID firms mark up fixtures 30–100% over retail. The ID's lighting offering is structurally a fixed-cost commodity ("10 free points + $40 per extra"). The ID has zero economic incentive to design lux-to-task. Report 5 (China) shows the Yeelight Pro / NVC model where the brand bypasses the ID with direct design service. Report 6 (KR) shows Ohou builds a full installer marketplace and is now the dominant funnel. Cooperation locks BSH at the bottom of the value chain.

Recommendation: Cooperate publicly, disintermediate operationally. The bshsg.com positioning is "the brief you bring to your ID" (cooperative). The actual product flow lets the homeowner book BSH's LEW install directly without the ID. The 50–80% of Singapore BTOs that hire a small ID firm or self-manage benefit from BSH's plan + install — they were never going to give the ID full markup anyway. The 20% who hire premium IDs (akiHAUS, Three-D, Black-N-White Haus) will use BSH's plan as a negotiation tool. BSH wins both segments. Critical rule: never publish anti-ID copy. Frame as additive.

T3 · Photometric accuracy (DIALux) vs photoreal beauty (Foyr/Coohom/Blender)

Argue: PHOTOMETRIC IS NON-NEGOTIABLE

Report 1 ranks raytrace render #1 of 13 DIALux outputs — but only if the photometry behind it is real. Soraa's failure (Report 3) was selling premium quality buyers couldn't verify. SS 531 compliance is a regulatory authority moat BSH alone can claim. Lemon8 regret evidence shows homeowners cite "too dim to chop ginger" — a calibratable lux failure. Without DIALux-grade numbers behind the render, BSH becomes Just Another Pretty Picture (Pepperfry, Coohom, Asian Paints visualiser).

Argue: PHOTOREAL BEAUTY IS THE SELLING POINT

Report 7 explicitly notes: "Foyr is photoreal-style, not photometric. The lighting is visually convincing but does not produce code-compliant lux values like DIALux evo." Foyr Neo has 50K+ users; DIALux has 750K but is engineering-only. Report 2's portfolio study: Behance freelancers who succeed publish furnished raytraced renders, never isolines. The buyer responds to feeling, not numbers. Coohom claims 99.8% photometric accuracy with KooLux while shipping consumer-friendly UI — proof both can co-exist.

Recommendation: Photoreal Blender Cycles render in the front-of-product. DIALux false-colour heatmap + SS 531 compliance certificate in a collapsible "engineering proof" panel below. Steven's existing 14-second Blender Cycles + IES + MCP pipeline already does this. The render is the seduction; the heatmap is the receipt. Best of both: the buyer sees Foyr-quality visuals AND DIALux-quality numbers because the same IES file drives both passes. Lead with photoreal. Footnote the photometry. Both layers exist on every PDP.

T4 · Asset-light (drop-ship) vs asset-heavy (showroom + warehouse)

Argue: ASSET-LIGHT

Report 7 (Indonesia · Fabelio cautionary tale) is the clearest data point: Fabelio raised $19M, filed bankruptcy 2022 after aggressive showroom expansion + heavy inventory + working-capital mismanagement. Modsy (Report 3) died on unit economics. BSH's existing CGD supplier relationship (Jojo @ sales32@szcgd.com per project_lightshop_cgd_supplier.md) is already drop-ship-ready. Singapore physical retail rent (Balestier) is expensive. Render-as-showroom + JIT shipping = lower fixed costs.

Argue: ASSET-HEAVY (some showroom)

Report 8 (SG) shows: Million Lighting, Light Craft, Sembawang Lighting House all maintain Balestier showrooms 30+ years and remain category-leading. Singapore HDB renovators trust touching the fixture. Report 5 (China) — Aqara has 400 physical experience centres. Without a touchable hero fixture, BSH is competing against the no-touch e-commerce tier (Shopee/Lazada) on price, not against the showroom tier on trust.

Recommendation: Asset-light core + one micro-showroom + roving install vans. Hold zero significant inventory; drop-ship from CGD Shenzhen + 2 backup Singapore wholesalers (Sim Lim Tower, Geylang). Maintain ONE small showroom — possibly the existing BSH office at low marginal cost — with 30 hero fixtures touchable and a 65" 4K display running the Blender render pipeline live. The 6–8 BSH vehicles already in service double as drop-ship + install ("roving showroom"). Beats Fabelio's failure mode; matches Singapore touch-and-feel expectation; preserves the LEW install moat that pure D2C can't replicate.

T5 · Free tier (RoomGPT/Coohom model) vs paid SaaS (SGD 49–99) for the plan

Argue: FREE

Report 3 evidence: RoomGPT did 6M+ generations free; Coohom free tier; Decor8 AI free tier. Free is now table stakes for AI-rendered room visualization. A SGD 49 paywall in front of the consumer self-serve flow may convert at 1–3% where free would convert at 15%+. The conversion is to the install — which is BSH's actual margin centre. Charging for the plan kills funnel volume.

Argue: PAID (SGD 49–99)

Report 8 (SG) makes this case explicitly: "SGD 15 is too cheap to feel professional; SGD 250 (ID 3D render per view) is too high for self-serve. SGD 49–99 is defensible — it sits below the SGD 100 per-point pain point." Free outputs are perceived as low-effort AI hallucination; paid outputs carry weight ("this is the plan I paid for, ID better respect it"). Soraa's lesson: when the buyer doesn't pay, they don't appreciate value. A SGD 49 charge is a quality signal.

Recommendation: Tiered: free preview, SGD 49 plan, SGD 99 with consultation, install bundled. (1) Free preview = thumbnail render of pre-loaded standard HDB plan with 1 swap. (2) SGD 49 plan = full Blender Cycles render of buyer's actual unit + 2-page PDF with BOM. (3) SGD 99 plan + 30-min Zoom with BSH engineer. (4) Install bundle deducts plan cost. The free tier captures 80% of curious traffic; the SGD 49 tier separates the renovating-in-90-days from the day-dreamers; the SGD 99 tier captures the highest-intent before install. Funnel logic: free → 49 → 99 → install (SGD 1,500–5,000 actual margin).

T6 · Korean K-drama aesthetic vs Japanese Akari/wabi-sabi heritage as the BSH brand voice

Argue: KOREAN (K-drama / "Crash Landing on You" / Ohou)

Report 6 (KR) explicitly: "Korea, by a noticeable margin." HDB ceiling 2.6m favours central pendant + cove + zone lamps (Korean model) over Japanese tatō bunsan (many small scattered lamps requiring more wiring points). Singapore HDB buyers cite K-dramas as inspiration on Renotalk, Lemon8, monoloft.sg. The Korean aesthetic photographs; the Japanese aesthetic requires. K-drama scenes ("Hometown Cha-Cha-Cha coastal warm", "It's Okay to Not Be Okay gallery mood") translate directly to bundle SKUs.

Argue: JAPANESE (Akari / Yamagiwa / Mayuhana / Tanizaki)

Report 6 (JP): premium pricing is unlocked by provenance and craft story. "Mayuhana" sells SGD 2K+ pendants because the story is Toyo Ito + Gifu lantern tradition + Tanizaki's In Praise of Shadows. The Korean look comes free with Pinterest; the Japanese story is what justifies the price tag. NHK's Haruka no Hikari proved every fixture deserves a 2-minute story. BSH cannot compete on price with Hooga or IKEA Trådfri — must compete on narrative.

Recommendation: "Sell the Korean look with the Japanese story." This is the synthesised position from Report 6's blockquote: stock fixtures that photograph like a K-drama set, but explain them with the Tanizaki / Akari / Mayuhana heritage. Concretely: the BSH bundle catalogue is named after K-dramas ("Crash Landing on You glass-house kit", "Hometown Cha-Cha-Cha coastal warm") at SGD 1,200–2,500 per bundle. Each fixture inside the bundle ships with a 2-minute story video on its Japanese/Singaporean design lineage. Cover photo = K-drama; product page = Akari narrative. The combo justifies premium positioning vs Hooga while capturing K-drama search intent on Lemon8/TikTok.

§5. Product framings for bshsg.com — ranked

Six framings evaluated. Each gets a one-line pitch, a text mockup of what the bshsg.com page would look like, regulatory compliance check, demographic fit, cost model, defensibility, and four star ratings.

★ RECOMMENDED PRIMARY ★

F1 · BSH Flat Plan — fixed-fee SGD 49–99 per HDB unit

"For SGD 49, walk into your ID meeting holding a photoreal render of your living room and a 2-page lighting plan we built for your exact HDB unit type."

Mockup sketch — what bshsg.com would look like

Hero (above fold): Single full-width before/after slider. Left = generic empty 4-room HDB living room with developer's bare batten. Right = the same room with BSH-spec'd Japandi kit lit photoreal in Blender Cycles. Single CTA button labelled "Get your flat's lighting plan — SGD 49." Below, a thin trust bar: "EMA Class L1 · 35 years · 12,000 HDB jobs · NEA Climate Voucher partner".

Second section: Three calculator cards: "HDB Lighting Plan (SGD 49)", "Condo Lighting Plan (SGD 79)", "Lighting Refresh (SGD 99)". Each card opens a 3-screen funnel — (1) pick your floor plan from a visual gallery of the 30 HDB standards, (2) pick a mood (Japandi / K-drama / Modern Asian / Heritage Chinese / Industrial Warm), (3) review SKU swaps with sliders for CCT + dim level. Each slider change retriggers a 14-second Blender render.

Third section: Scene grid for one selected room. 4 tiles: "Reunion dinner (warm dimmed)", "Saturday breakfast (3000K bright)", "Kids' study (4000K task)", "Late movie (cove only, 2700K)". Each tile is a render of the same room.

Footer: Engineering proof panel collapsed by default. Click expand: false-colour lux heatmap + SS 531 compliance table + IES file download + EMA LEW certificate reference. The buyer never sees this unless they want to.

Regulatory compliance check

SS 531 (lighting workplaces) — Plan output cites the borrowed-analogue values (lounge 200 lx, restaurant 300 lx, kitchen 500 lx) honestly per Report 9 §A1 caveat.
SS 638:2018+C1:2020+A1:2022 (electrical installations) — BOM filters to SS 638-compliant drivers with 30mA RCD compatibility flag.
BCA Green Mark 2021 — Plan auto-calculates LPD; flags GM-eligible (≤6 W/m² aggregate).
HDB Renovation Handbook — Auto-flags banned operations (structural wall hack, <2.4m clearance after false ceiling).
EMA Electricity Act — Recommended install workflow names BSH as LEW.
NEA MELS / Climate Voucher — Eligible SKUs marked with redemption badge.
SCDF Fire Code — Not applicable inside private flats; flagged for shophouse/condo common areas.

Demographic fit

BTO buyers age 28–34, HH income SGD 7–14k, polytechnic/degree, dual-income. They are mobile-first, WhatsApp-native, K-drama-watching, Lemon8/Xiaohongshu/Qanvast-browsing, and currently surrender lighting spec decisions to their ID firm with regret. ~28K new BTOs/year + the much larger resale renovation market. The 4-room flat type is ~50% of supply — primary target. 5-room ~25% secondary. Condo upgrader and ID firm partner are sub-flows.

Cost model

Line	Build cost	Per-unit margin (SGD 49 tier)	Per-unit margin (SGD 99 + install bundle)
Floor-plan digitisation (one-off)	$35K	—	—
Blender Cycles render farm (Mac Mini already paid)	$0	~$0.10/render compute	~$0.10
Foyr Neo SaaS for designer tier	$100/mo × 2 seats	—	—
Designer time per SGD 49 plan	—	$10 (15 min)	—
Designer time per SGD 99 + 30min consult	—	—	$40 (60 min total)
Plan revenue per unit	—	$49 - $10 = $39 gross	$99 - $40 = $59 gross
Install revenue per unit (avg)	—	—	$2,500 install × 30% margin = $750
Total margin per converted unit	—	$39	$809

Breakeven analysis: $35K one-off + $4K monthly opex (Foyr + 0.3 FTE) = $83K Year 1 fixed. Need ~2,000 plans/year at $39 avg = $78K (just under break-even at the plan tier alone). Or ~105 install-bundle conversions/year at $809 = $85K. Total Year 1 target: 2,000 plans / 100 installs = ~$163K margin → 100% over break-even. Achievable: 28K BTOs × 0.5% adoption = 140 install conversions; 28K × 7% plan adoption = 2,000 plans. Both within reach with sustained Lemon8/WhatsApp/Carousell content.

Defensibility

BSH's moat — what Foyr / KuJiale / Hue / a new VC-funded startup cannot replicate in 12 months:

EMA Class L1 contractor licence + 35-year history. No software company can sell hard-wired install. This is the conversion-to-margin step.
HDB floor-plan dataset. One-time cost, permanent moat until a competitor commissions the same.
Singapore Climate Voucher partnership. NEA-registered retailer status earns SGD 100–300 effective rebate at checkout — Coohom, Foyr can't offer this.
Local-language warranty + after-sale. BSH walks back to fix flicker, retail competitors don't.
35-year customer base. Repeat buyers + word-of-mouth in Singapore's small market.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

F2 · K-drama Kit Catalog — pre-curated 3-7 fixture bundles named after K-dramas

"Recreate the Crash Landing on You glass-house living room in your 4-room HDB — SGD 1,200, install included, delivered in 14 days."

Mockup sketch

Page structure: 12 bundle cards in a 3×4 grid. Each card has a hero photo (still from the drama or a render of the look reproduced in an HDB room), bundle name, price range, "shop the look" CTA. Bundles include: Crash Landing on You (warm glass-house), Hometown Cha-Cha-Cha (coastal pastel), It's Okay to Not Be Okay (gallery mood), Itaewon Class (industrial warm dim), Reply 1988 (heritage Asian), Start-Up (workspace), Penthouse (chandelier statement), Goblin (heritage royal), Vincenzo (modernist), Squid Game (red accent — for the bold). Each opens to a bundle page with the SKU list, the K-drama story + Tanizaki/Japanese narrative justification, photoreal render in a 4-room HDB, install timeline, total price.

Regulatory compliance check

All bundles filter to SS 638-compliant drivers + EMA-LEW-installable fixtures. K-drama IP is aesthetic reference only — bundle names use generic "K-drama" or drama titles in lowercase descriptive context ("the glass-house look popularised by Crash Landing on You"). Steven should run this past trademark counsel — using show names as bundle names is the line between cultural reference and IP infringement. Suggest using lowercase descriptive ("crash-landing look" / "hometown coastal kit") rather than capitalised brand names.

Demographic fit

BTO buyers 28–34, female-led aesthetic decisions, Lemon8 + Netflix Korean catalogue subscribers, willing to pay SGD 1,200–2,500 for a curated bundle that solves decision paralysis. ~30% of HDB BTO market self-identifies as "K-drama-inspired" per Renotalk / bornincolour.com / monoloft.sg surveys. 5-room + condo upgraders willing to spend SGD 2,000–4,000 on hero-bundle.

Cost model

Average bundle: SGD 1,500. Hardware COGS: ~$700 (CGD drop-ship). Install labour: $400. Gross margin per bundle: ~$400 (27%). Add SGD 49 plan tier in front = $449 funnel margin. Break-even at ~200 bundles/year = $80K. Year 1 stretch: 500 bundles × $400 = $200K margin. Marketing cost (TikTok / Lemon8 K-drama content): ~$20K/year.

Defensibility

Cultural curation is a brand-narrative moat that's hard to copy but not impossible. A competitor with 6 months and $50K could replicate. The defensibility layer is BSH's install bundle — competitors can sell the catalogue but can't promise "your kit installed in 14 days by a BSH LEW" unless they are themselves a SG contractor. Critical: register bundle names + photography copyright + maintain refresh cadence as new K-dramas land.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

F3 · ID Partnership Tier — sell measured plan to homeowners who walk into ID meeting holding it

"Get our SGD 49 lighting plan, hand it to your ID firm, save 8–22% on your electrical line item — no replacement of your ID needed."

Mockup sketch

Hero: A photograph of a Singapore couple sitting across a table from an interior designer, the BSH-printed PDF plan visible on the table between them. Headline: "Stop guessing. Walk in with the plan." Sub: "$49 buys you the brief 12,000 HDB owners wish they'd had." CTA: "Get my plan." Second section: testimonial carousel from real ID firms (Renozone, Carpenters, LE Interi) endorsing the BSH brief. Third section: side-by-side Lemon8 quote shrinkage stats ("Reduced our 4-Room BTO Electrical Quote from $7K to $6.4K" — sourced from real Lemon8 post per Report 8). Fourth section: list of "ID-firm partners" — IDs who give a 5% loyalty discount to BSH-plan buyers (a co-marketing flywheel).

Regulatory compliance check

Same as F1 — plan complies with SS 531 (borrowed analogues honest), SS 638, HDB, EMA LEW. Plus: the explicit ID-handoff framing requires zero direct involvement in the ID's permit / hacking / install workflow — BSH stays in plan-and-supply role. Lower regulatory risk than F1's bundled install model.

Demographic fit

The same BTO 28–34 cohort, but specifically the 70% who hire an ID rather than self-managing. Higher AOV at the ID end (since the ID does the install). BSH captures the plan margin + supplies fixtures wholesale to the ID at agreed discount.

Cost model

Plan revenue: $49 × 2,000 = $98K. Fixture wholesale supply to ID firms: ~30 fixtures per project × $40 avg margin × 200 projects = $240K. Net: ~$338K Year 1 if execution lands. Install revenue zero — but ID firms are happy to source from BSH directly (cuts their distributor markup).

Defensibility

The flywheel locks in: IDs love BSH plans because they save the ID design time (35-year contractor as free consultant); homeowners love the plan because it shrinks their electrical quote; BSH wins on fixture wholesale + plan revenue. The replication cost for a competitor is the trust relationship with 15+ Singapore ID firms — that's a 12–18 month build. Critical risk: if one large ID firm decides BSH plans are commoditising their margin, they could blacklist BSH SKUs. Mitigation: tier the wholesale discount so the IDs win materially.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

F4 · Smart-Home + Install Bundle — Chinese fixtures wrapped in BSH local install warranty

"Yeelight / Aqara / Tuya smart lighting installed by a 35-year EMA Class L1 licensed contractor — with a 2-year Singapore warranty no Taobao reseller can match."

Mockup sketch

Category page: "Smart lighting installed by BSH". Hero: video of a BSH technician installing an Aqara scene panel + Yeelight downlight in a 4-room HDB, walking the customer through the Aqara Home app. Below: 3 bundle tiers — Starter (1 master switch + 4 downlights, $399 installed), Pro (Aqara scene panel + 8 downlights + dimmer + cove strip, $999), Whole-Home (full 4-room kit + Home Assistant gateway, $2,400). Critical badge on every page: "Works without a neutral wire — BSH-curated." Other badges: "Climate Voucher eligible · saves SGD 100–300 at checkout"; "Matter compatible"; "2-year BSH warranty + 1-year extended dealer warranty."

Regulatory compliance check

This is the riskiest framing on Lens Q6 (no-neutral) and Q7 (smart protocol). Mandatory mitigations: (a) every smart switch SKU pre-tested for SS 638 RCD compatibility, (b) "needs neutral / no-neutral compatible" badge mandatory on every PDP, (c) BSH LEW installs the switch (because 80% of pre-2015 HDBs lack neutral, rewiring is a permit-requiring operation), (d) Aqara/Yeelight fire-safety certifications surfaced. Aqara is HDB fire-rated per Interlock.com.sg. Tuya is not — exclude Tuya from BSH curated catalogue.

Demographic fit

Tech-aware HDB owners 28–40, often male-led decision, willing to spend SGD 400–2,400 on smart layer. Currently sourcing from Lazada / Carousell / Interlock.com.sg. Pain points: setup complexity, no-neutral incompatibility, warranty void on imported direct goods. BSH's wedge is local-language setup support + warranty.

Cost model

Average bundle: $999. Hardware COGS: $500 (CGD or Aqara reseller). Install labour: $200. Gross margin: $300 (30%). Volume estimate: 500/year × $300 = $150K. Climate Voucher rebate is pass-through (no margin impact).

Defensibility

Aqara, Yeelight, Tuya products themselves are commodity (anyone can resell). BSH's moat: (a) bundled LEW install (no other Singapore retailer can install + warranty smart switches in one truck-roll), (b) no-neutral filtering done by 35-year electricians (Interlock can resell but won't install), (c) Climate Voucher pre-checkout integration. Threat: if Aqara / Xiaomi opens a Singapore direct-install service. Window: ~24 months.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

F5 · Lemon8 Content Engine — Singlish translation of mainless lighting backlash + reno regrets

"The honest-broker brand. Build trust through public technical literacy: 'is mainless lighting an IQ tax? Here's what 12,000 HDB jobs taught us.'"

Mockup sketch

Editorial pillar pages on bshsg.com/learn: "12 lighting regrets every HDB owner makes (and how to avoid them)" / "Is mainless 无主灯 lighting an IQ tax? Singlish translation of the Chinese backlash" / "Why 80% of HDB switches can't take a smart switch (the no-neutral problem)" / "Climate Voucher 101 — claim your SGD 400 before Dec 2027" / "Cove lighting hits 80% adoption in HDB — should you skip it?" Each piece: 1,500 words, embedded 60-second Lemon8 / TikTok video, downloadable cheat sheet PDF, three concrete BSH-spec'd alternatives at the bottom. Cross-link to F1 plan funnel.

Regulatory compliance check

No regulatory implications — content only. Honesty is the asset. Should explicitly cite SS 531 / SS 638 / HDB Renovation Handbook clauses so credibility is verifiable. Includes negative content ("here's what BSH doesn't recommend") — counter-intuitive trust signal.

Demographic fit

Top-of-funnel for every other framing. Captures Pinterest/Lemon8/Xiaohongshu/Renotalk lurking BTO buyers ~6–12 months before key collection. Drives SEO + organic share. Singlish + Mandarin + English versions for full demographic coverage.

Cost model

One-off content production: 20 pillar pieces × $1,500 = $30K. Ongoing: 2 pieces/month × $800 = $19K/year. Conversion to F1 plan funnel: target 2% → 40 plans/$1,500 piece = 800 plans/year × $39 = $31K direct attribution. Bigger value: SEO authority + Lemon8 follower base — multi-year compounding.

Defensibility

Content takes years to compound. First-mover SEO advantage on "HDB lighting Singapore" + "mainless lighting Singapore" + "Climate Voucher LED" is permanently valuable. The Lemon8 algorithm rewards consistent posting cadence. A competitor copying the same pieces 18 months later will still rank below BSH's earlier publish dates.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

★ DO NOT BUILD ★

F6 · Coohom / Foyr Embed — license international 3D engine, sell hosted on bshsg.com

"License Coohom's free-tier or Foyr Neo's SaaS, white-label it on bshsg.com, route SG users into it."

Why this is a SKIP framing

It fails 4/10 Lens questions out of the box (HDB ceiling defaults, CCT defaults, neutral-wire awareness, permit pathway awareness) — fixable with adaptation work but ~4–6 weeks engineering. The core problem is strategic, not technical: embedding makes BSH a thin wrapper on someone else's platform, which Calculux's death (Report 3) proves dies when the parent pivots. If Coohom decides to localise SG directly (Manycore's 200-country expansion roadmap suggests they will), BSH's customers churn to Coohom-direct overnight.

Better: use Foyr Neo internally as a productivity tool (see T1 recommendation), but never expose it to the consumer. Front-end remains bshsg.com-rendered + Blender Cycles-rendered + BSH-branded.

Feasibility★★★★★

Revenue potential★★★★★

Strategic fit★★★★★

Risk (1=safest)★★★★★

Framings ranked summary

Framing	Feasibility	Revenue	Fit	Risk (1=safe)	Composite	Verdict
F1 · BSH Flat Plan (SGD 49–99)	★★★★★	★★★★☆	★★★★★	★★☆☆☆	17	SHIP
F3 · ID Partnership Tier	★★★★☆	★★★★★	★★★★★	★★☆☆☆	17	SHIP
F5 · Lemon8 Content Engine	★★★★★	★★★☆☆	★★★★★	★★★★★	18	SHIP (parallel)
F2 · K-drama Kit Catalog	★★★★☆	★★★★☆	★★★★☆	★★★☆☆	16	SHIP after F1
F4 · Smart-Home Install Bundle	★★★★★	★★★☆☆	★★★★☆	★★★★☆	16	SHIP after F1
F6 · Coohom/Foyr Embed	★★★★☆	★★☆☆☆	★☆☆☆☆	★☆☆☆☆	8	SKIP

Unhedged synthesis

Ship F1 + F3 + F5 as a single integrated product launch. All three are mutually reinforcing: F5's content engine drives traffic into F1's calculator funnel; F1's outputs are the artifact F3 routes to ID firms; F3's ID firm partnerships generate testimonials for F5's content. The three together form the wedge BSH owns alone in Singapore. F2 (K-drama bundles) is the natural upsell once F1 has 1,000 plan buyers. F4 (smart bundles) is the natural sidesell to F1 install conversions. F6 is a strategic trap — use Foyr internally only, never expose.

§6. Recommended first 2-week sprint

The concrete, buildable thing that proves the strongest framing (F1 · BSH Flat Plan) is real. Not a fully launched product — a working prototype that converts internally and externally to demonstrate the wedge before Steven commits Year 1 budget.

Sprint goal

By end of Day 14: a working bshsg.com page where a homeowner can (a) select their HDB unit type from a gallery of 5 floor plans, (b) pick a mood (Japandi / K-drama / Modern Asian / Heritage Chinese / Industrial Warm), (c) see a Blender Cycles photoreal render of the chosen mood applied to their unit's living room — within 60 seconds. The output includes a PDF cover page they can save, a SKU list, a SGD 49 paywall to unlock the full plan. Success metric: 100 plan starts in week 3 (first week after launch), with ≥10% converting to paid plans (≥10 paid plans = SGD 490 first revenue + signal that the wedge is real).

File-level changes to bshsg.com

DAY 1–3Digitise 5 HDB floor plans in Blender

4-room BTO standard (90m², 2.6m ceiling, ~50% of supply)
5-room BTO (110m², 2.6m ceiling, ~25% of supply)
4-room resale 1990s (90m², 2.55m ceiling)
5-room resale 1990s (110m², 2.55m ceiling)
2-room Flexi (45m², 2.6m ceiling)

Each plan: locked geometry, locked furniture (Japandi-default sofa + dining table + bed + storage), 3 light positions per room locked. Output: 5 × .blend files in C:\bsh\dev\bsh-website\public\hdb-plans\. Owner: Steven + 1 contracted CAD artist if needed (per Report 8 / Archisketch playbook — quote: 6–8 weeks for 30 plans, so 5 plans in 3 days is achievable for a CAD-fluent operator).

DAY 4–6Pre-render 25 scene-mood combinations

5 floor plans × 5 moods × 1 hero render each = 25 PNGs at 1920×1080
Each render uses CGD .ies files from the existing 1,686-file library + 140K library files
Mood definitions locked: Japandi (3000K downlights + 2700K cove + warm dim pendant), K-drama (3000K downlights + 2700K cove + matte black + brushed brass pendant), Modern Asian (3000K downlights + magnetic track + recessed cove), Heritage Chinese (3000K downlights + brass chandelier + altar light), Industrial Warm (Edison filament + brass spot + exposed bulb)

14-second render × 25 = ~6 minutes total compute on RTX 5070 (already proven). Output: 25 PNGs in C:\bsh\dev\bsh-website\public\renders\. Cost: ~$0 (existing hardware + skill).

DAY 7–9Build the calculator page in Next.js

Route: bshsg.com/lighting-plan
Step 1: visual gallery of 5 floor plans (cards with thumbnail + dimensions + popularity)
Step 2: visual gallery of 5 moods (mood-board card with descriptive text + 1 example render)
Step 3: lookup the matching pre-rendered PNG from the 25-image library, display full-screen with overlay UI
Step 4: Stripe paywall at SGD 49 — unlocks full 2-page PDF download with SKU list + install slot booking

Tech: existing Next.js codebase + Stripe SG account (already set up per BSH context). New components: FloorPlanPicker.tsx, MoodPicker.tsx, RenderViewer.tsx, PaywallGate.tsx. Estimate: 3 days for one full-stack developer.

DAY 10–11Generate the SGD 49 PDF deliverable

2-page PDF template: page 1 = the photoreal render + scene summary + SKU list + total price; page 2 = false-colour lux heatmap (engineering proof) + SS 531 compliance note + EMA LEW certificate ref + install slot booking link
Auto-generated from the chosen plan+mood combination using existing IES + Blender outputs
Send via email + WhatsApp delivery to buyer

Tech: Cloud Function (Firebase) using HTML-to-PDF (Puppeteer). Estimated 2 days.

DAY 12–13Wire WhatsApp Business floating CTA + email capture

WhatsApp Business floating button on every page (per Report 7 cross-market pattern — Indonesia, Malaysia, India all confirmed)
Email capture for non-paying funnel exit ("Send me my preview render for free")
Lemon8/TikTok pixel for retargeting

1 day implementation.

DAY 14Soft-launch + 5 Lemon8 posts

Publish 5 Lemon8 posts in 60-second video format: (1) "BSH Flat Plan launch — SGD 49 vs SGD 250 ID render", (2) "Same 4-room BTO, 5 moods, instant render", (3) "Climate Voucher SGD 100–300 off our plans", (4) "Watch Blender Cycles render your room in 14 seconds", (5) "What HDB owners regret about lighting (Singlish translation of Lemon8 retrospectives)"
Test paid traffic to bshsg.com/lighting-plan from existing Lemon8 follower base

Marketing budget: $200 for first-week Lemon8 boosted posts. Owner: 1 part-time content creator (existing).

Success metric — how Steven knows it worked

Hard metric (binary): ≥10 paid SGD 49 plans in the first 14 days post-launch = SGD 490 first revenue + 10 customers who have voluntarily handed over money for a plan they value.

Soft metric (qualitative): ≥3 of those 10 customers respond to a follow-up survey within 7 days saying "I'm bringing this plan to my ID meeting" — confirming the F3 ID Partnership Tier hypothesis is real.

Failure metric: <3 paid plans in the first 14 days = the pricing or the perceived value is wrong, iterate before scaling. ~25 plan starts but zero paid = the funnel is broken (paywall placement, PDF preview, scarcity messaging) — fixable.

What Steven learns regardless of outcome

If 10+ paid plans: wedge confirmed. Move to Year 1 plan — digitise remaining 25 HDB floor plans, hire 1 designer, launch F3 ID partnership pilot with 3 ID firms.
If 1–9 paid plans: demand is real but conversion mechanics need iteration. Test pricing (SGD 29 vs SGD 49 vs SGD 79), test paywall placement, test PDF preview snippets. Three weeks of iteration before committing more capital.
If 0 paid plans + ≥100 starts: demand is real but value delivery is wrong. The render isn't matching expectations, or the PDF isn't worth the SGD 49. Iterate on render quality + PDF depth.
If <50 starts: demand-side problem — Lemon8/SEO isn't pulling. F5 content engine is the prerequisite. Pause F1 build, invest in 90 days of content first.

Every outcome teaches something. The cost of running the sprint is ~2 weeks of Steven's time + 1 contractor week + $500 cash. The cost of not running it is committing Year 1 plans on theory.

§7. Open decisions Steven must make

Eight judgment calls where the research is ambiguous and Steven has to choose. Each has my recommendation flagged as opinion — but the call is his.

D1 · Pricing tier — SGD 29 / 49 / 79 / 99 / 149 for the plan

Research shows: Brite.sg charges nothing for guidance; Elite Boss charges SGD 250/3D render view; Delight 2-room BTO package SGD 408. The SGD 49–99 band sits in unfilled whitespace.

Option A: SGD 29 plan tier — maximises volume (~2.5x conversion vs $99) but signals low quality.

Option B: SGD 49 plan tier (recommended in Report 8) — matches Lemon8 negotiation pain-point ($100/concealed point shift = plan saves more than it costs).

Option C: SGD 99 plan tier — closer to ID-firm 3D-view price; signals premium.

Option D: Free preview + SGD 149 full design + install-credit refund.

My recommendation (opinion): Option B — SGD 49 tier as primary, SGD 99 tier with 30-min Zoom consult as upsell. The SGD 29 tier feels like a Lemon8 voucher; SGD 99+ requires explicit value delivery; SGD 49 hits the "I'd pay this to save $100 on a shifted downlight" mental model precisely. Run the 2-week sprint at SGD 49 to test.

D2 · Build engineering team in-house vs contract Foyr SG + Singapore freelance designers

Report 7 confirms Foyr Neo has a Singapore office; freelance Singapore designers via Carousell/Telegram at SGD 800–2000 per render available. In-house FTE: ~SGD 6K/month each.

Option A: Hire 1 full-time designer + 1 part-time CAD operator from Day 1.

Option B: Use Foyr Neo SaaS (~$100/mo × 2 seats) + 1 freelancer at SGD 1500/mo on retainer for first 6 months.

Option C: Steven personally drives Blender Cycles renders for first 100 plans + adds team after revenue proven.

My recommendation: Option B for months 1–6 (lower fixed cost, faster iteration), Option A from month 7 if volume crosses ~1000 plans/year. Avoid Option C — Steven's time is the highest-leverage strategy asset, not the highest-leverage production asset.

D3 · Foyr partnership terms — embedded vs internal-only

Foyr Singapore office makes negotiation easy. Embedded = Foyr branding visible, white-label = no Foyr brand visible.

Option A: Embedded (Foyr widget shown on bshsg.com, "Powered by Foyr" footer).

Option B: Internal-only (designers use Foyr, customers never see it; outputs are exported to BSH-branded PDFs).

Option C: No Foyr — build custom Three.js + Blender Cycles only.

My recommendation: Option B. Foyr is a productivity tool for designers — not a customer-facing brand. BSH brand integrity matters more than time-to-market. Per T1 recommendation, hybrid model: Foyr internal + Blender Cycles customer-facing.

D4 · Climate Voucher retailer registration — apply now vs after launch

NEA Climate Voucher SGD 400/HDB household valid through Dec 2027. Registration is finite operational work (per Report 9). Each redeemed voucher = SGD 100–300 effective off BSH SKUs at checkout.

Option A: Apply for registration Day 1 of sprint (parallel to product build).

Option B: Wait until F1 plan is shipped and proven.

My recommendation: Option A. NEA registration lead time can be 30–60 days. The voucher is one of the strongest pull mechanisms in the entire 9-report study and applies retroactively. Filing the paperwork during product sprint is parallel work.

D5 · ID-firm partnership rollout — sequential vs parallel

15+ ID firms profiled in Report 8. Approaching all 15 simultaneously dilutes; sequential is slow.

Option A: Pilot with 3 IDs (Renozone + LE Interi + 9creation) first, expand after 3 months.

Option B: Open-application "BSH ID Partner Programme" with self-serve sign-up.

Option C: Skip ID partnership tier for first 6 months — go consumer-direct only.

My recommendation: Option A. The 3 pilots produce 3 detailed case studies in 90 days — enough to publicly market F3 in month 4. Approaching premium IDs (akiHAUS, Three-D, Black-N-White Haus) before the pilots are de-risked invites rejection.

D6 · Render quality threshold — Blender Cycles default vs higher-fidelity

Existing 14-second render is Cycles + IES at 1920×1080 = good. Higher fidelity (Twinmotion path tracer at 4K, 90+ sec) = excellent. Cost: 6× compute per render.

Option A: Stay at 1920×1080 Blender Cycles 14-second — current proven pipeline.

Option B: Upgrade hero renders only to 4K + Twinmotion for hero images; keep 1080p Cycles for self-serve.

Option C: Full 4K + 60-second renders on all SKUs.

My recommendation: Option B. Hero shots on homepage and proposal covers at 4K. Self-serve gallery at 1080p (mobile-first; nobody needs 4K on a phone). Avoid Option C — diminishing return per dollar.

D7 · K-drama bundle naming — risk trademark vs use generic descriptors

Drama show titles ("Crash Landing on You") are trademarked. Using as bundle names risks IP claims from Studio Dragon / tvN.

Option A: Use show names verbatim in lowercase ("the crash landing on you look") — fair use likely.

Option B: Use generic descriptors ("Korean glass-house warm", "Coastal pastel kit", "Gallery mood kit") with show name in subhead.

Option C: Avoid K-drama framing entirely.

My recommendation: Option B. The descriptor names are searchable; the show name in subhead captures Lemon8/Google search intent without inviting cease-and-desist. Steven should run final naming past Singapore IP counsel before launch (~$500, 1 week).

D8 · Geographic ambition — Singapore-only vs SG+SEA from Day 1

Report 7 evidence: Qanvast operates SG+MY natively. Coohom serves 200+ countries. Singapore is 5.9M; ASEAN-5 + India is ~2B.

Option A: Singapore-only Year 1 (28K BTOs + resale market = enough TAM at the right margin).

Option B: Singapore + Malaysia Year 1 (Qanvast model — same regulatory framework, ~95% overlap on HDB-like apartment renovation pattern).

Option C: Full ASEAN-5 expansion Year 1.

My recommendation: Option A. Singapore is BSH's regulatory authority moat — every framing leverages SS 531, BCA Green Mark, EMA LEW, HDB Permit. Expansion to MY+ID dilutes the moat to "another lighting brand". The right model is Year 1 Singapore-only + revenue ≥SGD 500K, then Year 2 evaluate MY entry (different regulator: Suruhanjaya Tenaga, not EMA).

§8. Top-10 next steps — ordered by leverage / effort ratio

Digitise 5 HDB floor plans in Blender 3 days

The single highest-leverage atomic action. Unlocks F1, F2, F3, F4 simultaneously. Without it nothing else ships.

Pre-render 25 scene-mood combinations 3 days

5 plans × 5 moods = 25 hero PNGs. Already-proven Blender Cycles + RTX 5070 pipeline. Forms the gallery.

Apply for NEA Climate Voucher retailer registration 2 days work, 30–60 days lead time

Parallel to product build. Worth SGD 100–300 effective per-customer rebate, valid through Dec 2027.

Ship the SGD 49 calculator page on bshsg.com/lighting-plan 7 days

Existing Next.js stack + Stripe SG + WhatsApp Business floating button. First conversion engine.

Publish first 5 Lemon8 posts in Singlish/Mandarin/English 5 days

F5 content engine kickstart. Pillar pieces: regret list, mainless backlash, no-neutral, Climate Voucher, cove decision tree.

Sign Foyr Neo SaaS (2 seats, Singapore office) 3 days incl. procurement

Internal designer productivity tool for the SGD 99 + consult tier. ~$200/mo opex. Per T1 / D3 recommendations.

Initiate 3 ID-firm partnership pilots (Renozone + LE Interi + 9creation) 2 weeks outreach + 30 days pilot

F3 wedge. Wholesale supply terms + plan-handoff workflow + co-branded case studies.

Build the no-neutral switch compatibility lookup tool 1 engineer-week

Per Report 6 (Daiko Japan model). Free SEO + lead-gen for F4 Smart-Home tier. First Singapore competitor to publish this.

Commission 10 short-form K-drama "recreate the look" videos for TikTok / Lemon8 3 weeks production

F2 K-drama bundle marketing. ~SGD 8K production cost. Per Report 6 KR cross-pollination.

Digitise the remaining 25 HDB floor plans + commission 125 more scene renders 6–8 weeks, ~$35K

Complete the Archisketch-style blueprint moat (Report 6 + 8). Permanent defensibility against Coohom / Foyr SG localisation. The biggest single capex of Year 1.

§9. Bibliography meta-index

Aggregate of all 9 source reports' bibliographies (~700 URLs total). Grouped by domain authority. Full URLs in source reports — this is a meta-index of where evidence comes from, not a re-listing.

Tier 1 — Singapore Government primary sources (.gov.sg, .sso.gov.sg, .bca.gov.sg)

~25 documents across Reports 8 + 9. Including:

Enterprise Singapore Standards eShop — SS 531-1:2006(2019), SS 531-2:2008, SS 531-3:2019, SS 638:2018+C1:2020+A1:2022, SS 563:2010(2017)
BCA — Green Mark 2021 scheme + technical guide, CONQUAS Private Residential Manual, BCA Energy Efficiency Technical Guide
HDB — Renovation Guidelines for Building Works, Electrical Works Renovation Guidelines, HDB List of Approved Materials, OCS for 2-Room Flexi Annex A, MyNiceHome HDB Renovation Guide, Directory of Renovation Contractors (DRC)
EMA — Engaging Licensed Electrical Workers, Worker Licences, ELISE Licensed Worker Search
SCDF — Fire Code 2023 Chapter 8 Clauses 8.1, Fire Code 2023 PDF
NEA — MELS & MEPS, Climate Friendly Households Programme (Climate Voucher), Enhanced CFHP press release
SSO (Singapore Statutes Online) — Workplace Safety and Health Act 2006, Electricity Act 2001, Electricity (Electrical Installations) Regulations
URA — DC 13-13 Night Lighting Guidelines, balcony incentive / façade controls
SP Group — SS 638 Code of Practice summary
SingStat / Department of Statistics — Income from Work and Household Income survey 2022/2024, HDB Key Statistics 2024/2025

Tier 2 — Singapore industry trade press & portals

Qanvast (qanvast.com/sg) — 20+ articles cited: renovation calculator, HDB ceiling heights, ceiling lighting types, BTO flat type guide, fengshui taboos, Reddit hot takes, cost benchmarks
Hometrust (hometrust.sg) — lighting design ideas, track lights, OCS article
Renotalk forum (renotalk.com) — HDB BTO lighting threads, cove lighting, electrical loading
HardwareZone forums — BTO lighting plan, concealed ceiling wiring, recessed LED downlight
Brite Singapore (brite.sg) — DIALux explainer, "how to calculate lights needed", SS 563 emergency lighting compliance
The Lighting Gallery (TLG) (tlgsg.com) — cove lighting HDB guide, 2700K vs 3000K, owns "cove lighting Singapore" SEO
HomeGenie SG (homegenie.com.sg) — HDB Lighting Design Room-by-Room, Ceiling Fan Size Guide, electrical installation costs
FixFirst SG (fixfirst.sg) — illegal HDB renovations, where to buy lights, home office lighting
9Creation, Sky Creation — BTO Renovation Timeline, regret guides, Japandi for HDB
Sol Luminaire (solluminaire.com.sg) — Beyond the Bulb, 3000K-consistency recommendation
Home & Decor SG, EdgeProp, 99.co, PropertyGuru — editorial features, lighting mistakes, HDB renovation guides
Lemon8 Singapore — first-person reno regret posts, electrical planning, K-drama HDB aesthetics
Renopedia SG, RCS Renovation Contractors — false ceiling trends, BTO 2026 guides, renovation cost guides

Tier 3 — International industry references (DIALux, BIMobject, manufacturer docs)

DIAL (dial.de + dialux.com) — DIALux evo 14 product page, knowledge base (raytracer, false colours, isolux, isolines, UGR, daylight, emergency lighting, luminous intensity, documentation, results overview, working plane, luminance, colour rendition, street lighting), DIAL community project showroom (~30 posts)
DIAL Photometry Service, DIALux Academy training
Behance — DIALux Evo / DIALux interior design searches; specific designer galleries (Artem Shevchuk Kyiv, Anna Denisova St Petersburg, Gurpreet Singh, Yermek Iskakov, Beshoy Nagy KSA, Sumaiya Eliyaz)
LBNL Radiance + pyradiance + frads — open-source photometric simulation
Lighting manufacturer KBs — Erco Track Configurator + Spotlight reference, Megaman, BEGA, KLM Lighting, ZGSM, Acuity Brands Visual Lighting, Hubbell ChalmLite, Philips Calculux, Lighting Analysts AGi32 + ElumTools

Tier 4 — Competitive software platforms

Coohom + KuJiale + Manycore (HKEX 00068.HK) — KooLux, virtual showroom, AIGC Lab, SpatialVerse, SpatialTwin, IPO PRs, Fortune / SCMP / KrAsia coverage
Foyr Neo (foyr.com) — product, 3D visualization, Singapore office
D5 Render, Twinmotion, Lumion, Unreal Engine 5, Blender, V-Ray — competitor stacks
Casambi, Philips Hue + SpatialAware, Lutron Caséta, Nanoleaf, IKEA Home Smart / TRÅDFRI / DIRIGERA, Yeelight, Cync, LIFX, Noon Home (acq.) — smart lighting apps
RoomGPT, Magnific AI Relight, Adobe Firefly, ChatGPT image, Decoratly, ReRoom AI / DecorMatters, PromeAI, Marble 1.1 / Luma Genie, Modsy (dead), Soraa (acq.) — AI image / 3D world generators
Houzz Pro, Wayfair View in Room 3D, IKEA Place AR, West Elm Design Crew, Planner 5D / Coohom / HomeByMe, Sweet Home 3D / Live Home 3D, Havenly — furniture + room planner

Tier 5 — China ecosystem (Yeelight, Aqara, Tuya, Mi Home, KuJiale, Tmall, Xiaohongshu)

Yeelight (cn.yeelight.com / en.yeelight.com) — Pro smart lighting, Pitaya magicians, whole-house guide
Aqara (aqara.com) — global, solutions, HomeKit scene port, smart lighting setup
Xiaomi Mi Home (home.mi.com) — 3D view feature (Red Dot 2025), Xiaomi For All write-up
Tuya Smart (tuya.com) — SmartLife docs, Razer game lighting
Opple Lighting, NVC, Honyar, FSL, TCL Lighting, Panasonic China, Philips China — traditional brand pages
Tmall + Taobao + JD + Pinduoduo — AR open dev platform, 3D virtual showroom, Apple Vision Pro Taobao, AR/VR alliance
Tongyi Wanxiang, Doubao, Wenxin Yige, Meitu, Coohom AIGC Lab — Chinese AI image generators
Xiaohongshu, Douyin, Zhihu, Bilibili — home content analysis
Vanke, Country Garden, Neri&Hu Design — showflats with starchitects, gooood.cn coverage
Zhihu 无主灯 mainless lighting backlash threads, CSDN, ThePaper, Shanghai Zhuang Wang regret coverage

Tier 6 — Japan + Korea ecosystems

Koizumi, Panasonic, Daiko, Endo, Yamagiwa (Mayuhana, Akari), Iris Ohyama, Toshiba Lighting, Maxray (Ushio) — Japanese pro brands with consumer simulators
LPA Lighting Planners Associates (Kaoru Mende), Nipek, Lightbasic, Sirius, Reiko Chikada, Ichiro Iwasaki — Japanese design studios with Singapore portfolio overlap (Gardens by the Bay, Jewel Changi, Marina Bay)
Akari Light Sculptures by Isamu Noguchi (Ozeki Gifu) — Noguchi Museum shop, Dezeen, ArchDaily Mayuhana feature
RoomClip (Japan interior SNS, 6M+ photos, NTT DOCOMO Ventures), Houzz Japan, Rakuten AR, Karimoku 3D, LOWYA OkuROOM, Asahi Optics CRI
NHK Asaichi 照明術, Casa BRUTUS 居心地のいい照明術, Haruka no Hikari NHK drama 2021, FUJIOH Japandi lighting — JP consumer education
Ohou / Bucketplace — Korean super-app (20M downloads, $182M Series D), Vertex Growth / TechCrunch coverage
Archisketch — 3D engine + 96% of Korean blueprint moat, Ohou partnership
Samsung SmartThings, LG Innotek, AGO Lighting (Euljiro Seoul), Lumir, KILS — KR smart home + lighting designers
Naver HyperCLOVA X, Kakao Karlo 2.1 — Korean AI text-to-image
Coupang, 11번가, Gmarket, Vittz — Korean e-commerce
Crash Landing on You / The Penthouse / Itaewon Class K-drama analyses, RM (Kim Namjoon) apartment tour, BTOnomics Taobao guides, Lemon8 K-drama HDB recreations

Tier 7 — India + SEA

India consumer brands — Wipro Consumer Lighting, Havells, Bajaj Electricals (AWE Plus), Crompton (MyCrompton), Syska, Orient (Orismart), Eveready, Anchor (Panasonic), Surya Roshni (STIC), Goldmedal (i-Touch, iWorld), Philips Hue India + SpatialAware (Jan 2026 release)
India interior platforms — Livspace (IKEA-backed, 200K projects, IKEA SG/MY/AU), HomeLane (SpaceCraft), DesignCafe, Bonito (Gauri Khan), Pepperfry (AR/3D), Urban Ladder, Asian Paints Beautiful Home
India 3D viz SaaS — Foyr Neo (Hyderabad + Singapore office), Coohom India localised, ArchiPlanner AI, ARki Building+Room Planner
India lighting consultancies — Design Matrix, Klove Studio, Lirio Lopez, Lux Light, Space Lumens, Veda, Studio Lotus
Indonesia — Dekoruma (Assemblr AR case study), Fabelio (defunct 2022 — The Runway cautionary tale), Hannochs, Visicom, Schneider Wiser Indonesia, Tokopedia + Shopee Indonesia lighting categories
Vietnam — Rang Dong, Dien Quang, Expert Market Research Vietnam LED, Asia Designers Directory trends 2026–2030
Thailand — Lamptan, Baan Lae Suan / room.baanlaesuan.com, Decor8 AI
Malaysia — Qanvast Malaysia, Recommend.my, Megaman Malaysia, Studio Bikin KL
Philippines — Akari Lighting, Firefly Electric (FELCO), Omni Electrical, BluPrint magazine, Real Living, Lazada Philippines lighting

Tier 8 — Social proof & cross-market consumer references

Lemon8 Singapore — @miracullious, @beansand.beams, @chiating.home, @soul_pavilion, @furfilise, @puipui_home, @a1001sqfthomee, @theveggiehome, @dumbo_geor, @chengyiinteriordesign (Taobao lighting guides + reno regrets + electrical planning)
BTOnomics — Taobao lighting buying guide series
Renopedia SG, monoloft.sg, bornincolour.com, prestige-affairs.com — Singapore K-drama HDB aesthetic articles
Pinterest, Xiaohongshu — visual moodboard discovery
Apartment Therapy, Tom's Guide, Wirecutter, ArchDaily — Western mass-market editorial
Tatler Asia, Indesign Live, The Peak Magazine, Wallpaper, Surface magazine, Designboom, Dezeen — premium design editorial

Tier 9 — Defunct / archive (cautionary tales)

Modsy — TechCrunch / Fortune / Business of Home / Surface magazine shutdown coverage (July 2022, $72.7M raised, capital-intensive 3D rendering pipeline failure)
LIFX — Feit Electric acquisition Aug 2022 (cash burn April 2022)
Noon Home — Racepoint Energy / Savant acquisition Jan 2020 (~$50M raised, premium switch hardware)
Finally Bulb / Lucidity Lights — defunct ~2019 ($23M raised, "acandescent" induction tech)
Soraa — Ecosense acquisition 2020 ($175M raised, GaN-on-GaN LED, "great tech, no narrative")
Calculux — deprecated Philips legacy
Fabelio — Indonesia bankruptcy 2022 ($19M raised, asset-heavy showroom model)

Methodology notes

This master document does not re-list individual URLs (~700 across the 9 reports). Each Tier above refers to the source reports where the URL is documented. Steven can drill into any Tier by opening the relevant report (01–09).
Singapore-domain (.gov.sg, .sg, brand-direct .com.sg) is treated as Tier 1 authority for any Singapore-specific claim.
Government documents not directly web-scrape-able (binary PDFs) were triangulated via at least two independent industry sources citing the same clause + number, with primary URL retained for verification.
Pricing figures are 2024–2026 SGD nominal. Re-validate quarterly against SP-tariff and FX (SGD/USD/RMB).
Convergence patterns (§3) are flagged only where 3+ independent reports landed on the same claim without coordinating. Contradictions (§4) are flagged only where two reports actively disagree on a recommendation.
This master document was synthesised from ~10,300 lines of source HTML across 9 reports between 2026-05-16 09:00 SGT and end-of-day, by Claude Opus 4.7 (1M context) under direction from Steven Choo.