Comparison
Employ the engineers yourself, or contract the capability. It turns on whether the need is permanent, whether you can hire and keep people, and which knowledge must stay inside.
Build in-house when software is the product, or the capability will be central to the business for years, and you can hire and keep the people. If customers buy the software, it will be iterated on indefinitely and its knowledge is part of your competitive position, that capability belongs with employees. Contracting it means renting something you need permanently.
Outsource when the need is real but temporary or partial: a programme with an end date, a specialism needed continuously but part-time, extra capacity on a funded plan, or cover while recruitment runs. In those cases a permanent hire is a decision you later have to unwind, and contracting matches the shape of the need.
Many organisations run both on purpose. Permanent staff hold the systems that carry domain knowledge and product direction, and external capacity carries work that is bounded, specialist or temporary. It works when the split is decided deliberately, and fails when the core of the product goes outside and nobody internally can explain how it works.
In-house and outsourced development both work, and both fail in predictable ways. Good and bad engineers exist in each model. The difference is what you commit to and where knowledge accumulates. An in-house team is a permanent capability: a fixed cost, slow to assemble and slow to change, which builds understanding of your business every month it exists.
Outsourced development is contracted capacity: faster to start and adjustable in both directions, with knowledge kept only as far as the arrangement is designed to keep it. The deciding questions are whether the capability must be permanent, whether you can hire and retain the people in your market, and how much business knowledge has to stay inside the organisation.
Open what matters to you.
Salaries, rates and employer costs vary by market and specialism, so compare structures. An in-house engineer is a fixed cost: salary plus contributions, benefits, recruitment, equipment, licences and management time. Vacancies cost the delivery they would have produced, usually silently, as a plan that keeps slipping. Compare a contracted rate against salary plus true employer cost, the expected time to fill the role and the chance it stays unfilled.
A contracted engineer costs more per hour because the rate carries recruitment, time between engagements, cover, tooling and overhead. The difference buys the removal of the fixed commitment and the hiring cycle. Fixed and variable costs behave differently under uncertainty. If funding is committed for a year, demand is seasonal or the programme could be cancelled, flexibility is worth more. If the work is certain and permanent, employment is usually cheaper.
An engineer who has worked on your system for a year is far more effective on it than an equally skilled newcomer, mostly through context: why an odd decision was right, which integration misreports, what the business means by its own terms. Employment is the arrangement most likely to keep that context. High attrition loses it as thoroughly as a rotating supplier, and many teams rely on one person to hold it.
Outsourcing loses knowledge by default and keeps it by design. Insist on named individuals, an interview and paid overlap before any change of person, work in your repositories and review process, and documentation written as the work happens. Then apply one test in either model: if the person who knows a system best stopped tomorrow, delivery should continue. Ending an engagement should leave a running system your own people can operate.
Approving headcount does not create capability. Writing the role, sourcing, interviews, an offer, a notice period that can run to months, and onboarding come first, and one declined offer restarts much of it. Contracted capacity removes the search and the notice period. Access is then the usual bottleneck: accounts, permissions, security reviews and devices prepared before the start date take days, and discovered in week one they can absorb a fortnight.
Employment gives simpler controls: your vetting, devices and policies. It does not bring quality or security practice by itself. Least-privilege access, review before merge, automated checks and audit logging work in either model. External engineers should be named users in your identity provider, removed through your normal leaver process, with UK GDPR roles documented where personal data is involved. Intellectual property should be assigned to you on creation.
Divide work by permanence and by proximity to the domain. Permanent staff hold the systems where business logic lives and decisions are frequent and consequential. External capacity carries migrations, mobile clients, infrastructure layers, data pipelines and extra throughput on funded programmes. Do not contract out the part of the system you most need to understand, and give every significant system a named internal owner, even when external engineers do the work.
A transitional hybrid also works, where external capacity is there to be replaced. External engineers keep delivery moving while recruitment proceeds, in the same repositories and review process, and permanent joiners pair with them on real work. Write the handover down with a date, an owner and a budget. An intention to insource without those is how a temporary arrangement becomes a permanent dependency.
Moving work in-house is an insourcing programme and needs resourcing as one. Hire first and overlap deliberately: new engineers pair on real work and take ownership of one system at a time while the previous owners are still available, and only then does external capacity reduce. Ending the contract on the day the new hires start turns every question into archaeology. Poorly documented, untested systems take longest.
Moving work out is mostly arrangement design: which systems the external team owns, who internally stays accountable, and how work stays visible every week. Transfer knowledge before departures, while the people who hold it are still there. In both directions, keep repositories, cloud accounts, CI, the issue tracker and documentation in your control, so a switch changes who works and nothing has to move.
Not reliably. A contracted engineer’s rate is usually higher than a salary looks, because it carries recruitment, management, tooling, cover and overhead. Outsourcing changes the shape of the commitment: you stop paying for unused capacity. Compare total cost with total cost, including vacancies. Permanent hiring often wins for steady long-term work, and contracting for bounded or uncertain work.
Dependence: software that works but that nobody you employ can explain, change or operate without the supplier. It comes from arm’s-length delivery, no internal owners, visibility limited to status reports and documentation promised for the end. Work in your repositories and accounts, name an internal owner for every significant system, and make sure you could give notice today and still run the system.
Months, in most markets. Each role needs sourcing, interviews, an offer, a notice period and onboarding, and a declined offer restarts much of it. First hires are usually slowest, because there is no team yet for candidates to join. Ramp-up then depends on the codebase: a tested system with a working pipeline is quick to join, and an undocumented one is slow for anyone.
Usually only part. Keep the systems where business logic lives with permanent staff, and contract work that is bounded, specialist, temporary or peripheral. Outsourcing an entire product makes sense when you have no engineering organisation yet, product direction owned internally, and an explicit plan to build the permanent team alongside it. Understanding of an outsourced core does not arrive later by itself.
Run both, with the transition written down. Keep recruiting if the capability is permanent, and use contracted capacity so the plan stops bending around people who do not exist yet. External engineers work in your repositories and process, new joiners pair with them on real work, and external capacity reduces with notice as the internal team arrives. Give the handover a date, an owner and a budget.
Start a project
Tell us what you’re building and what it has to work with. We’ll tell you which option fits and why.
Or email [email protected]