How to Build an E-commerce Store: Products/Inventory, Payments, Logistics and Returns — the Complete Plan (with Costs and a 90-Day Roadmap)

Opening: A Common Pain Scenario

A handmade-skincare brand hired a cheap vendor to build a templated store, live at NT$45,000. Three months in, the problems surfaced: inventory synced by hand; payments took credit cards only (no COD, no convenience-store pickup); returns handled manually over LINE, burning 20 store-manager hours a month on reconciliation and refunds alone. They thought they'd bought "a site that sells"; what they got was "a catalog with a checkout button." The real cost of e-commerce was never putting products online — it's everything you handle daily after launch.

Applicable vs. Not-Applicable Scenarios

When custom/semi-custom e-commerce fits:

• You have a brand identity and long-term plans, and don't want to be locked into platform commissions
• Products have variants (size, color) and need linked inventory
• You need Taiwanese local logistics (convenience-store pickup) and diverse payments
• You plan membership, repeat purchase, or subscriptions

When a ready-made platform is the better first step:

• You sell only 1–5 items and are still validating: test with Shopee or IG shopping first
• Budget under NT$30,000 but full features expected: mismatch — start on Shopify's monthly plan
• No one to process orders and support daily: no system survives an unstaffed store

Alternatives Matrix

Full Process Breakdown (Phases, Time, Tools)

• Phase 1: Requirements & information architecture (1–2 weeks) — map product structure, variants, pay/logistics needs; deliverable: feature list + wireframes (tools: Figma, Notion).
• Phase 2: Payments/logistics planning (3–5 days) — pick payments (ECPay, TapPay, Stripe) and logistics (convenience-store pickup, home delivery); confirm fees and payout cycles.
• Phase 3: Development & integration (3–5 weeks) — product/inventory admin, cart, checkout, payment callbacks, logistics integration; deliverable: testable environment.
• Phase 4: Testing & soft launch (1–2 weeks) — real small-amount payment tests, refund-flow rehearsal, load testing; deliverable: test report.
• Phase 5: Launch & handover (3–5 days) — switch to live payments, admin training, ops docs (tool: UptimeRobot monitoring).

Full Cost Breakdown

Beyond the build fee, budget these "hidden" costs up front:

• Payment processing fees: credit cards roughly 2.75%–3.4% (plan-dependent) — deducted on every order; confirm rates on ECPay or Stripe's official pricing.
• Convenience-store logistics fees: tens of NT$ per parcel, usually with a deposit.
• SSL certificate: free via Let's Encrypt, but EV certificates cost extra.
• Hosting: e-commerce carries more traffic and data — VPS/cloud from NT$1,000–5,000+/month.
• Ongoing maintenance: package updates, payment-policy changes, peak-season scaling — budget a monthly retainer.
• Pay/logistics integration hours: routinely underestimated; a full convenience-store pickup integration (including returns) can be several days of work.

Implementation Truth vs. Client Imagination

• Imagined: "Just hook up payments." Truth: the hard part is callbacks and reconciliation — paid but no notification, double charges, refunds; these edge cases are the work.
• Imagined: "Inventory just stays accurate." Truth: selling one stock pool across channels (site + Shopee + physical) makes sync its own engineering problem; skip it and you oversell.
• Imagined: "Launch is the finish line." Truth: launch is the start — peak traffic, payment-policy updates, return disputes are ongoing costs.

Common Traps and How to Avoid Them

• Trap 1: Only one payment method. → Before launch, confirm your audience's habits and cover at least credit card + convenience store to cut cart abandonment.
• Trap 2: Going live without a real small-amount payment test. → Run a real low-value charge including a refund to confirm callbacks work.
• Trap 3: Overselling from unsynced inventory. → If you sell across channels, plan a single source of truth for inventory from day one.
• Trap 4: Returns handled entirely by hand. → Build an RMA (return authorization) flow into the admin; don't let support track it on LINE.
• Trap 5: No monitoring, so you learn the site is down from customers. → Add UptimeRobot-style monitoring plus payment-failure alerts.
• Trap 6: Ignoring the mobile checkout experience. → Most Taiwanese orders come from phones — test checkout on mobile first.

Success Metrics + 90-Day Post-Launch Roadmap

• First 30 days: watch add-to-cart rate, checkout completion, abandonment; fix the biggest hole in the checkout funnel first.
• 60 days: watch repeat-purchase rate and average order value, roll out email/retargeting; analyze low-converting product pages.
• 90 days: watch overall ROAS and return rate, decide whether to add membership/subscription, and plan peak-season scaling.

Decision Checklist

• ☐ Do my products have variants that need linked inventory?
• ☐ Do I need Taiwanese local logistics like convenience-store pickup?
• ☐ Which payment methods does my audience prefer (card/store/COD)?
• ☐ Will I sell across channels (site + Shopee + physical)?
• ☐ Do I have staff to process orders, support and returns daily?
• ☐ Do I need membership, repeat purchase or subscriptions?
• ☐ Does my budget cover payment fees and ongoing maintenance?
• ☐ Do I need to integrate with an existing ERP/CRM?

If you check 4 or more, custom or semi-custom e-commerce will cost less long term than pure templating.

FAQ

How much does an e-commerce site actually cost?

It depends on complexity. Pure templating/store platforms start at NT$1,000/month; self-hosted WooCommerce a few tens of thousands to build; deep custom (with ERP integration, cross-channel inventory) from NT$150,000. The real difference is the depth of pay/logistics integration and inventory sync, not how pretty the pages are.

Store platform or custom — how do I choose?

Still validating, few items: start on a store platform. When commissions, customization limits or data lock-in constrain you, or you need deep internal-system integration, move to custom. It's a "cost structure" choice, not a "which is better technically" one.

How high are payment fees?

Taiwanese card processing commonly runs ~2.75%–3.4%, varying by plan and volume, with different payout cycles. Always confirm on official pricing pages like ECPay and Stripe, and factor it into per-order margin.

Will I keep spending after launch?

Yes. Payment-policy changes, package updates, peak-season scaling and returns are ongoing costs — budget a monthly retainer. Treat e-commerce as "running a store," not "building a site," and your budget won't drift.

Call to Action

Want to know whether your store should use a platform or go custom? We offer a free 30-minute consult to work out payments, logistics, inventory and costs in one pass. ScriptWalker e-commerce projects start at NT$150,000.

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