Game development outsourcing: what it costs and how to manage it

What gets outsourced, and what it costs

Working ranges for a small to mid-size commercial project. Treat them as sanity checks rather than quotes: rates vary considerably with region, seniority and how well the work is specified.

Outsourced versus in-house, honestly

Comparing a contractor's hourly rate to an employee's salary is the most common budgeting mistake in this area, because only one of those two numbers is the whole cost.

The practical test: is there twelve months of this work, and is somebody able to direct it? Two yeses point to hiring. Anything else usually points to outsourcing, and the freelance-versus-studio side of that decision is covered in freelance, studio, or asset store.

Pricing models, and when each one fails

For most first engagements, fixed price on a small, well-defined first milestone is the right structure. It gives both sides a low-risk way to find out how the other works before committing to the whole scope.

The costs nobody quotes for

These are the reason an outsourced workstream that looked cheaper on the quote comes in level or higher. None of them are anybody's fault; they are simply not in the fee. The wider budget these sit inside, including store fees, ratings, certification and post-launch runway, is in what a game actually costs to make.

1. Your management time. Somebody on your side briefs, reviews, answers questions and accepts deliveries. Budget 10 to 20 percent of the contract value in your own hours, and more for the first month.
2. Review and feedback cycles. Every milestone needs someone to actually look at it properly. Teams that rush acceptance pay for it two milestones later.
3. Integration. Delivered assets have to be brought into the build, named, hooked up and tested. This is nearly always more work than expected, and it is yours.
4. Rework from vague briefs. The most expensive line item in outsourcing, and the most preventable. A brief that leaves interpretation open produces work that has to be redone at your cost or theirs, and either way it costs time.
5. Timezone drag. A question that takes two minutes to answer can take a day to resolve across a twelve-hour gap. Over a project that adds up to real schedule.
6. Knowledge that leaves. When the engagement ends, the understanding goes with it. Documentation and source files are how you keep it, and they have to be asked for.
7. Onboarding, every time. Each new vendor has to learn your project. Continuity has real value, which is why a slightly more expensive partner you have already worked with is often the cheaper choice.

How to scope so the quote is accurate

The accuracy of any quote you receive is capped by the quality of what you sent. This is the highest-leverage hour in the whole process.

• Make it countable. Screens, assets, clips, words, hours. A number, even a wrong one, beats an adjective.
• Include the states and the variants. The gap between "14 screens" and "14 screens with empty, error, loading and disabled states" is most of the price.
• Give real content. Longest strings, biggest numbers, fullest lists. Work scoped against tidy sample data breaks on real data, late.
• Name the technical constraints. Engine, target resolutions, inputs, platform requirements, performance budgets.
• Say what done means. File formats, naming conventions, what has to be delivered for a milestone to be accepted.
• Say who reviews it and how fast. Vendors schedule around your response time whether you specify it or not.

Managing an outsourced workstream

1. 01

   Start with a small paid pilot

   One milestone, a few days of work, real money. You learn more about how someone works from one delivered milestone than from any number of calls, and it is a cheap way to be wrong.

2. 02

   Write acceptance criteria before the work starts

   What has to be true for a milestone to be accepted. Without it, "done" is a negotiation every time, and both sides end up frustrated.

3. 03

   Name one point of contact on each side

   Feedback from four people in three channels is the most reliable way to destroy an outsourced workstream. One voice each way.

4. 04

   Batch feedback, and be fast

   One consolidated list per milestone, delivered within an agreed window. Scattered notes over several days cost far more than the same notes sent together.

5. 05

   Take source files at every milestone

   Not at the end. This single habit converts most worst cases from a loss into a delay.

6. 06

   Review against the criteria, not against taste

   If something is wrong that was not in the brief, that is a change, not a fix. Recognising the difference honestly keeps the relationship, and the budget, intact.

When it goes wrong

The failure modes are predictable and mostly preventable by structure rather than by choosing better vendors.

What to keep in-house, always

Outsourcing works best on well-bounded production. A few things should not leave the building, regardless of budget.

• Creative direction. You can buy execution. Deciding what the game is, and what it looks and feels like, is the thing you cannot delegate and stay in control.
• The core loop. The part players actually spend their time in should be tuned by whoever cares most about the game.
• Final acceptance. Somebody internal must be able to say no, and must actually look.
• Institutional knowledge. Whatever you learn about your players, your build and your pipeline needs to stay with you when a contract ends.

Interface work sits in an interesting middle: the execution outsources well, but the decisions about what the player sees and in what order are close enough to design direction that they need someone internal engaged. The best outcomes I have had are the ones where a client stayed involved in the flow decisions and let me own the execution.