"Contract signed, NT$450,000 deposit paid. Three months into development, the vendor's PM stopped replying, the Slack channel went silent for a whole week, and when we finally phoned we learned the five-person studio's lead engineer had quit along with two colleagues." That was a late-night message from an e-commerce operations director. He was left holding an admin panel that was only 60% done, a database nobody could read, and an August launch date. This is not a rare script — some version of it plays out for SMBs every single year.
The real question is not "are they technically good?" but "will this company survive until my project ships?" Skill shows up in an interview; financial and delivery stability hides deep. This article gives you a way to read that risk before you sign, plus a survival process to cap the damage if it happens anyway.
Myths: what you think is right, but isn't
- Myth 1: Older company = safer. Truth — a ten-year-old shop that is now just one owner taking gigs is riskier than a healthy two-year-old team. Judge current delivery capacity, not tenure.
- Myth 2: A cheap quote is a win. Truth — a bid 30% below market usually survives by squeezing headcount and stalling payments, the exact structure that snaps mid-project. Cheap often means "vanishes halfway."
- Myth 3: A contract protects me. Truth — a contract lets you sue later, but what you recover rarely covers launch delays and re-tendering. Contracts guard against disputes, not disappearance.
- Myth 4: Big firms never fail. Truth — big firms hand small jobs to their juniors and cut unprofitable projects first. Your priority in their queue is what matters.
Core framework: the vendor-stability scoring formula
Turn "will they vanish?" into a takeaway tool. Stability Index = (Financial health × 0.35) + (Delivery resilience × 0.35) + (Dependency concentration × 0.20) + (Transparency × 0.10). Score each 1–5, weight, and convert to a percentage.
- 70+: safe for a long contract and a higher deposit share.
- 50–69: proceed, but on milestone payments with source code pushed to your Git in real time.
- Below 50: accept only short, splittable, easily-handed-over tasks.
Dependency concentration matters most: if the whole project rides on one engineer, that is a single point of failure — the moment they leave, stability drops to zero. Taiwan has over 1.63 million SMBs, more than 90% of all enterprises (SME and Startup Administration, MOEA); churn and exit are normal, so betting on one person is the biggest risk of all.
Three contrasting scenarios
| Company type | Common mistake | Stability focus |
|---|---|---|
| 20-person legacy manufacturer (first digitization) | Hiring the cheapest, smoothest-talking freelancer | Insist on a team not an individual; confirm handover docs and code ownership |
| 50-person growth-stage e-commerce | Trusting a big system vendor's brand | Ask who actually builds your project and where it sits in their priority queue |
| 5-person startup (fundraising) | Trading equity or deferred payment for build work | Confirm the vendor's cash flow survives your payment cycle so you don't both run out of air |
Hidden costs: the real bill when a vendor vanishes mid-project
- Sunk paid amounts: deposit plus interim payment of NT$300k–800k, usually unrecoverable.
- Re-tender premium: taking over someone's half-finished work runs 20–40% more than a fresh build, because the new team must first decode the old code.
- Code-archaeology hours: undocumented handover costs 80–160 hours just to understand the architecture — at NT$1,800/hr, roughly NT$140k–290k.
- Lost revenue from delay: in peak season, each week slipped often costs a mid-size store NT$200k–500k.
- Internal management drag: your team chasing status, holding coordination calls, handling legal — an underestimated hidden labor cost.
Add it up and one "vanished halfway" event costs an SMB roughly NT$800k–2,000k — far more than was saved by choosing the cheap vendor in the first place.
The KPI scorecard: 8 financial & stability signals
| # | Signal | How to check | Points |
|---|---|---|---|
| 1 | Company registration & solvency status | MOEA Business Registration lookup — tax ID, capital, dissolution | 15 |
| 2 | Payment-structure transparency | Accepts milestone payments; deposit reasonable (≤40%) | 15 |
| 3 | Team size & role coverage | At least 2 people can cover the same module — no single point | 15 |
| 4 | Attrition / turnover | Key staff stable over the past year; verify via LinkedIn | 10 |
| 5 | Real-time source delivery | Code pushed weekly to your Git, not handed over only at closeout | 15 |
| 6 | Documentation & handover norms | README, environment setup, deployment docs exist | 10 |
| 7 | Past-project survival rate | Reachable former clients; are their projects still running? | 10 |
| 8 | Litigation & reputation record | Judicial Yuan judgment search for delivery disputes | 10 |
Out of 100. A vendor below 60 is not automatically disqualified, but you must bind the risk with milestone payments plus real-time source delivery.
How ScriptWalker maps to this (and where we don't fit)
We serve SMBs with Laravel and Flutter. The two models we most often use to answer the "vendor might vanish" fear:
- Milestone-delivery model: payment only after each milestone is accepted, with code pushed weekly to the client's own Git — you always hold the latest version, so even a terminated engagement hands over painlessly.
- Long-term maintenance-partner model: monthly fee, transparent hours, single point of contact, replacing "one person carries everything" with "team backup plus documentation" to remove single points of failure.
Honestly, we are not the right fit when:
- Your budget is 30% below market and you also demand unlimited free maintenance forever.
- You are an early startup with no cash-flow plan, wanting to pay only in equity or indefinitely deferred payment.
- Requirements change daily and you refuse any written milestones or acceptance criteria.
- You see outsourcing as cheap labor and won't share Git or documentation.
Survival playbook: when a vendor is starting to fail
- Week 1 (stop the bleeding): immediately request, in writing, that the vendor deliver the latest source, database, design files, and account access to your own Git and cloud. Freeze any unpaid amounts and document all communication.
- Month 2–3 (stabilize): bring in a second team for a code health check to assess how much is salvageable; keep the original vendor's handover contact for at least two weeks and reconcile the feature list and data tables item by item.
- Day 90 (rebuild): complete the handover or decide to rewrite, switch the new contract to milestone payments with real-time source delivery, and add "key-personnel change notice" and "documentation delivery" clauses to avoid a repeat.
Decision checklist
- ☐ I verified the vendor's tax ID and solvency in the business registration system
- ☐ They accept milestone payments; deposit is under 40%
- ☐ At least two people can take over my project — no single point
- ☐ Code is pushed weekly to my own Git
- ☐ The contract assigns source code and IP to me
- ☐ They provide README, deployment, and environment docs
- ☐ I contacted at least one verifiable former client
- ☐ I checked for delivery-dispute litigation records
- ☐ The contract has a key-personnel change-notice clause
- ☐ I know the cost and time to hand over if they vanish
- ☐ I kept all communication and delivery records
- ☐ Payment is tied to delivery progress, not to elapsed time
FAQ
Is refusing to deliver source code weekly a red flag?
A clear one. A healthy team has no reason to withhold code you paid to build; withholding usually means they want to lock you in by keeping the code hostage, or their progress is secretly behind.
Is a five-person studio always riskier?
Not necessarily. Small teams can be more focused and faster to respond. What matters is single-point-of-failure risk and cash-flow health, not headcount. Score them; don't judge by size.
I already signed and paid before noticing they're failing — now what?
Execute the Week 1 "stop the bleeding" steps immediately: rescue the source and access, freeze unpaid amounts, and document everything. The sooner you act, the more assets you recover.
Get in touch
If you are evaluating a vendor, or already holding a half-finished project that needs rescuing, take the scorecard above and use it. Need a third party to run a risk health check? We can help. ScriptWalker uses milestone delivery and transparent hours so you hold the wheel at every point in time.
- Email: [email protected]
- Phone: 0916-224-047
- LINE: @ufv9089p