What to Outsource, What to Keep In-House? Splitting One Engineer's Job Into 6 Categories

An 80-person traditional-industry owner asks: "I hired an engineer at NT$70k/month, but he alone handles the website, fixes the ERP, and takes on AI projects — he's always stuck. Should I hire another, or outsource part of it?" This isn't an either/or of "outsource vs. in-house"; it's "an engineer's job can be split into a few categories — which suit outsourcing, which must stay in-house." This piece gives you a repeatable responsibility-split framework so you stop deciding with the crude "all or nothing."

Industry myths debunked

Myth: hiring an in-house engineer is cheaper than outsourcing. Truth: salary is only 60% of the cost; add labor/health insurance, equipment, management, recruiting amortization, and the risk of "the project breaking when he quits," and the true annual cost is often 1.5–1.7× the salary.
Myth: core systems must be built by your own people to be safe. Truth: safety comes from "documentation, permissions, and handover," not "whether the person sits in the office." Undocumented in-house staff are as dangerous as undocumented outsourcing.
Myth: outsourcing means throwing the whole project out. Truth: the mature approach is "division of labor" — outsource the standardizable, verifiable parts; keep the parts tightly bound to trade secrets and real-time decisions.
Myth: finding one all-around engineer solves it. Truth: all-around engineers are scarce and expensive, and "one person doing everything" is itself the biggest single point of failure.

Core framework: 6 job categories × the outsourcing safety line

Split an engineer's work into 6 categories, asking three questions each — "Standardizable? Needs real-time presence? Bound to trade secrets?" — to decide whether it lands in "suits outsourcing / hybrid / keep in-house":

Job category	Standardizable	Needs real-time presence	Recommendation
Website / marketing-site dev & maintenance	High	Low	Suits outsourcing
App / new product development	High	Low	Suits outsourcing (with maintenance contract)
Data analysis / reporting	Medium	Low	Hybrid (outsource build, internal use)
Internal ERP / core process customization	Medium	Medium	Hybrid (outsource dev, internal owns requirements)
Real-time ops / emergency firefighting	Low	High	Keep in-house or sign an SLA
Business decisions / data governance / permissions	Low	High	Definitely in-house

Three typical scenarios contrasted

20-person startup: no full-time engineer. Conclusion: outsource website and App entirely, keep a "tech-savvy point of contact" (the founder can double up) to own requirements and acceptance.
80-person traditional firm: 1 engineer near burnout. Conclusion: outsource the website and new App to free him for the ERP and internal real-time needs; sign a maintenance SLA to cover firefighting.
200-person company: a 3-person team. Conclusion: keep core systems in-house, outsource "peak-period overflow projects" and "non-core marketing sites" to avoid permanently expanding headcount for short-term peaks.

Full hidden-cost list

The true annual cost of an in-house engineer at NT$70k/month: salary NT$840k, labor/health insurance and pension contributions ~NT$120k–150k, equipment and software licenses ~NT$50k, recruiting amortization (agency fees/interview hours) ~NT$50k–80k, management cost (manager's time) ~NT$100k, and the priciest of all — the "single point of failure": the 30-day handover gap when he quits can delay a project with hard-to-estimate losses. The total true annual cost often reaches NT$1.2M–1.4M, far above the nominal NT$840k.

KPI scorecard for evaluating an outsourcing partner

☐ Do they provide a written scope doc and acceptance criteria?
☐ Do they hand over source code and docs (no lock-in)?
☐ Is there a clear maintenance SLA and response time?
☐ Are they willing to do knowledge transfer/training?
☐ Is the quote itemized and transparent (no "lump sum" black box)?
☐ Do they have cases in the same industry or type?
☐ Is the point of contact single and stable?
☐ Do they proactively raise risks and "not recommended" items?
☐ Are IP and data-ownership clauses clear?
☐ Is the exit mechanism (handover at contract end) spelled out?

ScriptWalker's corresponding model + when it doesn't fit

ScriptWalker offers four collaboration models: one-off projects (website/App build), monthly maintenance (Retainer, from NT$5,000/mo), advisory (architecture/tech-selection consulting), and "overflow dev" (catching your internal team's peak spillover). We don't fit these clients: factory production-line systems needing 24-hour on-site firefighting; those treating outsourcing as "cheaper long-term labor than internal"; and teams unwilling to do any documentation or permission management — these engagements usually leave both sides unhappy.

Transition playbook

Month 1: map the current engineer's work into the 6 categories, mark the "should outsource" items; concurrently build basic docs and a permission list.
Months 2–3: hand the first "suits outsourcing" item (usually the website or an App) to the partner as a trial run, shift the internal engineer to core work; set up a weekly sync.
Day 90: review the outsourced delivery quality and whether the internal engineer's load improved, and decide to expand outsourcing or adjust the split.

Decision Checklist

☐ Is my engineer "one person doing everything"?
☐ Which jobs are standardizable and verifiable?
☐ Which jobs are bound to trade secrets or need real-time presence?
☐ Do I have basic system docs and a permission list?
☐ Have I calculated the "true annual cost," not just salary?
☐ If the engineer quit tomorrow, could the project survive?
☐ Do I need "permanent headcount" or "peak coverage"?
☐ Can I accept outsourcing non-core items like the website/App?
☐ Have I prepared an SLA for emergency firefighting?
☐ Have I written down my criteria for evaluating a partner?

FAQ

Won't outsourcing core-system-related work create security/confidentiality risk?

Risk comes from "no permission control and no docs," not "outsourcing" itself. The right approach: give outsourcing only the necessary access, sign confidentiality and IP clauses, and route all changes through version control and logging. A disciplined outsourcing partner is often safer than an undocumented in-house person who could quit anytime.

I already have one engineer — will outsourcing make him feel replaceable?

The framework's point is "division of labor, not replacement." Outsourcing the standardizable website/App frees him for more core, presence-requiring work (ERP, real-time decision support). In communication, frame outsourcing as "support that offloads his chores"; most engineers welcome it.

How do I judge whether a job "can be outsourced"?

Ask three questions: Standardizable (clear spec and acceptance)? Needs real-time presence (on-call)? Bound to trade secrets (touches core data/decisions)? Jobs leaning "no" on all three almost always suit outsourcing; jobs leaning "yes" on all three stay in-house.

How is an outsourced maintenance SLA usually priced?

Commonly a monthly retainer, tiered by "response time + hours included per month." For example, a basic plan at NT$5,000/mo includes a few maintenance hours and business-hours response; an emergency plan includes shorter response times and holiday support. The key is spelling out "response time," not buying a vague "maintenance."

Call to action

Not sure what to outsource and what to keep? ScriptWalker offers a 30-minute free responsibility-split consultation to map your current tech work onto that 6-category table and give you a clear "outsourcing safety line." Contact: