Building a DTC Brand Commerce Site: Membership Tiers, Promotion Engine, Inventory Sync, and Three Budget Levels

1. State of the industry

Taiwan's retail sector posted NT$4.8448 trillion in 2025, down 0.2% year over year — only the third annual contraction in 25 years. Online retail, however, reached NT$671.6 billion, up 2.8%, accounting for 13.9% of all retail and 14.8% in Q4, per the Ministry of Economic Affairs (raw tables in the online retail sales survey; breakdown in this Business Next report). Online is still growing, but slowly — incremental revenue no longer falls out of opening one more marketplace storefront.

Brand owners report the same three wounds: commissions eating margin, customer data locked inside someone else's platform, and online and offline inventory counted separately. A brand site is not there to beat the marketplaces. It moves the high-repeat, remarketable slice of your customers onto ground you own.

Three anonymized competitor sketches

• Brand A, skincare: SaaS storefront plus LINE binding. Homepage leads with subscription refills, four membership tiers, automated birthday coupons, and product pages carrying ingredient notes and a sensitive-skin Q&A — its SEO wins come from ingredient long-tail terms, not brand terms.
• Brand B, apparel: headless build. Size guide recommends by height and weight, product pages show live store stock with a collect-in-store option, and the return policy sits on the product page rather than in the footer.
• Brand C, food gifts: four seasonal campaigns a year drive revenue. Its promotion engine must run buy-three-get-one, free shipping over NT$1,200, and a corporate bulk-order form at once, with pre-orders capping production risk.

2. What these customers search for

Vertical long-tail terms convert best. Give each a dedicated landing page instead of cramming them into one services page.

3. Non-negotiable features

• Product model with variants: SKU, colour, size, volume as structured attributes, not text baked into images. Variants share reviews, hold separate stock.
• Single source of truth for inventory: one sellable quantity shared by site, POS, and marketplaces, with reservation on order and timed release.
• Payment mix: cards with installments, ATM transfer, convenience-store codes, store pickup with cash on delivery. ECPay's published rate for standard merchants is 2.75% on domestic cards (pre-tax, plus NT$1 per order), 1% on ATM, NT$31 per store code; contracted merchants negotiate 1.85%–2.75%.
• Logistics and e-invoicing: C2C and B2C store shipping, home delivery, chilled options, automatic invoice issuance and credit notes.
• Membership: tiers, points, birthday and upgrade triggers, LINE Login, phone number as primary key.
• Promotion engine: threshold discounts, buy-N-get-M, bundles, time windows, per-customer limits, stacking priority — expressed as data, not hard-coded conditionals.
• Checkout optimisation: guest checkout, address autofill, live shipping calculation, abandoned-cart capture.
• SEO and structured data: Product markup, ProductGroup for variants, MerchantReturnPolicy for returns.
• Analytics wiring: GA4 ecommerce events, server-side Meta CAPI, Google Merchant Center free listings.
• Performance floor: product page LCP under 2.5 seconds, responsive images in next-gen formats.

4. Worth adding vs waste of money

5. Three budget levels

• Entry, NT$80,000–180,000: deep customisation on a SaaS platform (annual fee roughly NT$20,000–60,000), brand visuals, payment, logistics and e-invoice setup, GA4 and Merchant Center, basic membership and single-layer coupons. Fits sub-NT$5M first-year revenue and under 200 SKUs.
• Standard, NT$250,000–600,000: semi-custom or headless front end, membership tiers and points, promotion engine, LINE OA binding and segmentation, two-way ERP/POS inventory sync, abandoned-cart and repeat-purchase automation. Fits NT$10M–100M revenue with recurring campaign cycles.
• Advanced, NT$800,000–2,500,000: custom or headless commerce core, multi-warehouse inventory, click and collect with clienteling, cross-border currencies and languages, CDP and marketing automation. Fits multi-channel brands with owned warehouses or overseas ambitions.

Budget separately for NT$12,000–60,000 monthly maintenance and the gateway contract (ECPay lists NT$13,000 per year plus NT$5,000 setup for contracted merchants). Marketplace commission is the invisible line item: base transaction fees commonly start around 5.5%, and once payment fees, campaign subsidies, and on-platform ads stack up, exceeding 15% of revenue is routine. That gap is the business case for a brand site.

6. Stack recommendations, with reasons

• Next.js App Router front end: render product pages statically with ISR and invalidate on demand, so a price or stock change does not trigger a full rebuild — the pragmatic answer to many pages updated often.
• Commerce core, pick one: few SKUs and simple campaigns, consume the SaaS platform API as a headless backend and inherit its payment and compliance work; complex rules or multi-warehouse, build on Laravel or Medusa so promotion and inventory logic stays yours.
• PostgreSQL: stock reservation needs transactional locking; row-level locks plus JSONB for product attributes save real effort.
• Local payment and logistics providers: ECPay or NewebPay handle store pickup and e-invoicing; layer Stripe on top only for cross-border. No overseas gateway touches Taiwan's store-pickup rails.
• Meilisearch or Algolia for on-site search: Chinese tokenisation and typo tolerance are not worth hand-rolling.
• Phone number as the member primary key, LINE Login and email secondary, because stores, support, and SMS all key on the phone.

7. Traps and how to avoid them

• Trap 1: marketplaces only, customer data owned by others. Fix: the brand site need not outsell the platforms in year one. Run it as a member pool and content asset, pulling people back with first-purchase coupons and subscription refills.
• Trap 2: overselling across channels. Fix: one sellable quantity, reservation at checkout, timed release, safety buffer on hot items. Reject any sync slower than five minutes.
• Trap 3: promotions hard-coded. Fix: model them as condition-action-priority rows so marketing schedules campaigns without an engineering ticket.
• Trap 4: coupons drained, free-shipping thresholds miscalculated. Fix: recompute every discount and shipping fee server-side; the front end only displays. Enforce per-customer limits, validity windows, minimum spend.
• Trap 5: no product structured data, so Google cannot read price or availability. Fix: add Product, ProductGroup, and MerchantReturnPolicy markup and mirror it into Merchant Center.
• Trap 6: 3MB hero images pushing mobile LCP past five seconds. Fix: transcode on upload, serve device-appropriate sizes over a CDN, prioritise the first image and lazy-load the rest.
• Trap 7: member data scattered across POS, site, LINE, and email tool. Fix: unify the member ID before buying a CDP. Reverse that order and you have purchased an expensive bin.

8. The 30 / 60 / 90 day roadmap

• Days 1–30: reconcile payment and logistics ledgers, verify e-invoices, validate GA4 ecommerce events, publish to Merchant Center, audit top on-site search queries, launch first-purchase coupons and abandoned-cart emails.
• Days 31–60: switch on membership tiers and points, segment LINE broadcasts, complete product page content (ingredients, sizing, care), A/B test checkout, rewrite pages for the highest-return SKUs.
• Days 61–90: build a campaign model with margin floors, open pre-orders and back-in-stock alerts, pilot click and collect, and measure the median days from first to second purchase so automation fires at the right moment.

9. How ScriptWalker fits this sector

We build retail commerce in three shapes: deep customisation on a SaaS platform, a headless front end over an existing commerce backend, and a custom commerce core with multi-warehouse inventory and click and collect. Packages and starting prices:

• Brand site starter — from NT$120,000: SaaS customisation, payment, logistics and invoice setup, GA4 and Merchant Center, SEO structured data.
• Commerce standard — from NT$380,000: membership tiers and points, promotion engine, ERP/POS inventory sync, LINE OA binding and automation.
• OMO / headless advanced — from NT$900,000: multi-warehouse inventory, click and collect, cross-border currencies, CDP integration.
• Monthly operations — from NT$12,000: performance monitoring, campaign support, version upgrades.

10. FAQ

We already sell on marketplaces. Do we still need our own site?

Yes, but not to overtake them immediately. Marketplaces bring new customers and volume; your own site protects margin and repeat purchase. Between transaction fees, payment processing, campaign subsidies, and on-platform advertising, marketplaces routinely absorb more than 15% of revenue, while a brand site pays roughly the 2.75% gateway rate. More importantly, phone numbers, purchase cycles, and browsing behaviour never leave the platform.

SaaS platform or custom build?

Two thresholds decide it: more than 1,000 SKUs, and promotion rules that must stack across channels. If neither applies, a SaaS platform is enough — spend the money on content and ads. If either applies, look at headless or custom, because the platform's rule engine and inventory model will start to constrain you. Building custom below roughly NT$30M annual revenue usually means paying for problems that have not arrived.

How tight does inventory sync need to be?

A single source of sellable quantity, reservation at the moment of order, and sync latency under five minutes, with a safety buffer on hot items — say withholding 5% of stock. The real disaster is not underselling; it is cancelling oversold orders mid-campaign, which costs margin and reputation at once.

How long until SEO pays off?

Brand terms in 2–4 weeks, product terms in 2–3 months, vertical long-tail and content terms in 4–6 months. The accelerant is finishing structured data and product page content first — ingredients, materials, sizing, use cases. Those pages are simultaneously an SEO asset and a return-rate reducer.

11. Decision checklist and next step

• ☐ Can you state marketplace commission plus ad spend as a percentage of revenue?
• ☐ How many remarketable phone numbers or LINE IDs do you actually own?
• ☐ How many SKUs and variants, and how fast will that grow in twelve months?
• ☐ Do promotion rules need to stack across channels (coupon + points + free shipping)?
• ☐ Do you have physical stores needing click and collect or live stock display?
• ☐ Who owns inventory today — ERP, POS, or a spreadsheet?
• ☐ Do you know the median days from first to second purchase?
• ☐ Do product pages carry Product structured data and Merchant Center listings?
• ☐ Is mobile product page LCP under 2.5 seconds?
• ☐ Is cross-border on the roadmap within twelve months?

Fewer than six checked means the gap is measurement, not a website. More than six and you are ready for a spec conversation: send us your SKU count, channel mix, and campaign calendar, and we will return a feature list with a price band.

• Email: [email protected]
• Phone: 0916-224-047
• LINE: @ufv9089p