Short answer: The heart of an inventory system is a record of every stock movement, not a sheet of current quantities. For an SMB with 3 warehouses and about 2,000 SKUs, a custom build takes about 9 weeks and NT$200,000–450,000; with fewer than 300 SKUs and one sales channel, inventory SaaS is enough.
A hardware parts wholesaler runs 3 warehouses and 1,800 SKUs on one shared Excel file. Sales staff deduct stock separately for the website and Shopee, and warehouse staff update the sheet at month-end. Last quarter's count gap was 6%, and "in stock online, empty on the shelf" cancelled 37 orders. The staff are not careless; Excel simply stores how much there is now, and nobody knows how the number changed.
When It Fits and When It Does Not
Bottom line: companies with multiple warehouses, multiple channels or batch and expiry tracking benefit most; small catalogs on one channel can wait.
| Good fit | Hold off |
|---|---|
| 2+ warehouses or stores transferring stock | Under 300 SKUs in one warehouse |
| 2+ channels sharing stock (website, Shopee, momo) | Selling on one marketplace whose back office is enough |
| Batch and expiry tracking for food or cosmetics | Make-to-order with almost no stock |
| Monthly count gaps above 2% | Owners have not settled warehouse and SKU rules |
Alternatives Matrix
Bottom line: the right choice depends on channel count and rule complexity, not company size.
| Option | Pros | Cons | Cost range |
|---|---|---|---|
| Excel / Google Sheets | Free, familiar | No movement history, concurrent edits break | NT$0 |
| Inventory SaaS | Live in a week, built-in reports | Limited multi-warehouse and channel rules | About NT$1,000–6,000 per month |
| ERP module | Integrated with accounting | High consulting fees, clunky | From NT$300,000 |
| Custom inventory system | Fits your warehouses, batches and channels | Takes 2+ months | NT$200,000–450,000 |
Step-by-Step Process
Bottom line: 3 of the 9 weeks go to data cleanup, and that is what decides success.
- Weeks 1–2, SKU audit: clean SKUs, unit conversions and warehouses in Google Sheets; the deliverable is a clean item master. Register barcodes under GS1 Taiwan rules to avoid conflicts with channels later.
- Week 3, process design: draw receiving, shipping, transfer and count screens in Figma; the deliverable is a clickable prototype.
- Weeks 4–7, development: a Laravel back office with a stock movement ledger and safety stock alerts, using queues for marketplace sync via APIs such as the Shopee Open Platform.
- Week 8, opening count: scan every warehouse with phones and replay last month's documents.
- Week 9, go-live: run in parallel for 2 weeks, with LINE alerts for shortages and transfers.
Real Cost Breakdown
Bottom line: beyond development, expect about NT$30,000–80,000 in hidden first-year costs.
- Development: NT$200,000–450,000, driven by warehouse and channel count.
- Barcodes and labels: GS1 membership and a label printer, about NT$10,000–30,000.
- Hosting and backups: about NT$12,000–30,000 a year.
- Opening count labor: 2–3 working days for 3 warehouses, with overtime.
- Maintenance: from NT$3,000 a month to handle marketplace API changes.
Reality vs Expectations
Bottom line: a system makes errors visible; it does not make numbers accurate on its own.
- Expected: accurate from day one. Reality: gaps grow in month one as hidden errors surface.
- Expected: staff will adapt. Reality: shipping without scanning takes 4–6 weeks to unlearn.
- Expected: real-time marketplace sync. Reality: API rate limits usually mean updates every few minutes.
Common Traps and How to Avoid Them
Bottom line: most failures come from unclear rules, not technology.
- One product with 3 SKUs → merge the item master before launch.
- Recording balances, not movements → every movement needs a document and an operator.
- All channels share one number → set channel reserves to prevent overselling.
- Guessing safety stock → use 90-day average sales times replenishment days.
- Counting once a year → cycle count high-value or fast-moving items weekly.
Success Metrics and a 90-Day Roadmap
Bottom line: getting count gaps under 1% within 90 days is success.
- Day 30: 95% of shipments scanned; fix item master errors.
- Day 60: oversell cancellations under 5 a month; tune safety stock.
- Day 90: cycle count gaps under 1%; add replenishment suggestions.
Decision Checklist
Bottom line: if more than 6 of these 11 are yes, plan an inventory system.
- ☐ 2+ warehouses or stores
- ☐ Selling on 2+ channels
- ☐ Cancelled orders due to stockouts last quarter
- ☐ Count gaps above 2%
- ☐ Batch or expiry requirements
- ☐ Warehouse staff spend 1+ hour a day updating Excel
- ☐ Transfers are arranged verbally on LINE
- ☐ Reorder timing relies on gut feel
- ☐ More than 500 SKUs
- ☐ Willing to spend 2 weeks cleaning SKUs before launch
- ☐ A warehouse lead or manager can drive adoption
Frequently Asked Questions
How long does an inventory system take?
About 6–8 weeks for one warehouse under a thousand SKUs; 9–12 weeks with multiple warehouses and marketplace sync. Most time goes into data cleanup, not code.
Do we need dedicated barcode scanners?
Not at first. Phone cameras can scan barcodes; consider dedicated scanners once you ship over 100 orders a day.
What is the difference between inventory SaaS and custom?
SaaS is cheaper and faster but you follow its rules; custom costs more but fits your warehouses, batches and channel rules.
Do we have to stop shipping at go-live?
No. Do the opening count over a weekend and run old and new systems in parallel for 2 weeks before switching.
Next Step
ScriptWalker (a Taiwanese Laravel / Flutter custom development studio) offers an inventory management system package starting at NT$200,000, including SKU cleanup, phone-based scanning counts and marketplace stock sync. Send us your Excel file and warehouse notes for a free 30-minute review:
- Email: [email protected]
- Phone: 0916-224-047
- LINE: @ufv9089p