Freelancer vs agency vs dev shop

You decided to build. Now you pick who builds it, and the three options fail in different places.

You have settled build against buy. The next call is freelancer vs agency vs dev shop, and it is not a budget question: picking the break you can live with is most of the decision.

So who should build it?

A freelancer, if the work is one person wide — a first MVP usually is. An agency, if several specialists must work on it at once and somebody has to keep them in step. A dev shop, if you want a small senior team that does all of it.

• Freelancer. Cheapest, quickest to start, and it stops when they stop.
• Agency. Capacity, process and cover, paid for partly in layers you never meet.
• Dev shop. Senior people doing the work rather than selling it, with no bench behind them.

If build against buy is still open, settle that first — our note on build vs buy for small teams is one level up.

When is a freelance developer the right choice?

When the work is one person wide and stays that way. That is the whole test if you are deciding whether to hire a freelance developer or an agency: one developer who can hold the thing in their head beats a team, because adding people adds meetings, not speed.

• The job has edges. A site, an integration, one feature on a system that already exists.
• You know what you want and can say it plainly, or you are technical enough to steer it week by week.
• There is already somewhere for it to live and someone to run it afterwards.

A first MVP is usually this shape, so if you are asking who should build your MVP, a freelancer is often the honest answer. Ask what happens if it works, though: a prototype nobody else can pick up is a rebuild you have not been quoted for.

Continuity is what you take on. One person gets ill or gets a better offer, and the project stops where it stood. Nobody reviews the code. Nobody argues with the first idea.

What are you paying an agency for?

Coordination, mostly. The rate covers the developer, the designer, the project manager who keeps them in the same week, and the account manager who tells you how it is going.

That is worth paying for when the work really does run in parallel. A brand, a campaign, an app and the system behind it, all landing on a date that cannot move, is more than a small team can hold. Agencies are built for that, and they survive somebody leaving halfway.

It stops being worth it when the layers outnumber the work. The tell: the person who understood your problem is not the person building it, and a small question waits in a queue for days. On a modest custom system, that gap is most of the invoice.

What is a boutique dev shop, and why does nobody explain it?

It is a small team that takes the whole build itself, senior people writing the code instead of selling it. Nobody markets it clearly because "small agency" sounds like a worse agency and "team of freelancers" sounds like a worse freelancer.

Compared with an agency, a boutique dev shop trades layers for continuity: smaller, and everyone on the call also writes the code.

"Software development company" describes the paperwork, not the size, so comparing a software development company with a freelancer tells you nothing until you know how many of their people are on your project.

What you are buying is that nothing gets handed over: the person who understood the problem is still there when the decisions are made. That is 99six — two founders, strategy and engineering, nobody in between. The about page says who does which half.

The trade is capacity. A small shop has no bench, so above a certain size the honest answer is an agency. A good one tells you where its ceiling sits before you ask.

Freelancer vs agency vs dev shop: how many people at once?

One person at a time, a freelancer. A few who need to talk daily, a dev shop. More than fit around one table against a fixed date, an agency. One question decides it: how many people need to work on this at the same time. Not the budget, and not the deadline.

• One person at a time. A freelancer, and spend some of what you saved on a review of the code.
• A few people who need to talk every day. A dev shop, where coordination is a conversation, not a process.
• More people than fit around one table, against a fixed date. An agency, whose margin buys the process that keeps them pointed the same way.

Most small business software is the middle case sold as the third one. That is where the money goes.

What you are buying is how few people have to agree before something ships.

What changes when the process is the business?

Who you hire changes completely. If how you work is the reason customers pick you, the build is somebody sitting with your operation until they understand why it is shaped that way.

Give a freelancer a spec and you get the spec. Give an agency a discovery phase and you get a document, handed to a developer who was in none of those conversations. Both are fine when the software is generic, and neither is when the value sits in the parts nobody wrote down.

So ask who will still be there when your process meets the database schema.

How would a good builder tell you they are the wrong fit?

Early, without being asked, and with a reason you can check. Anyone worth hiring can describe the shape of project they are not good at.

We turn down the projects that do not fit, either one clear piece where a team is only overhead or work that is wider than we are, and where we can we say who would be better suited. That is what the book page promises before you ever get on the call: you describe the problem, we tell you whether we can help, and if we cannot, who can. Saying so on the first call costs us a project and saves you a bad one.

Ask all three the same questions. The answers separate them faster than any portfolio:

• Who writes the code, and will I ever speak to them?
• What happens if you are unavailable for a month?
• What kind of project do you say no to?
• Who owns the code, the repository and the hosting at the end?
• What does month two cost, when it is live and something breaks on a Friday?

None of the three is the safe choice. The safety is in whether the answer costs the person giving it something. That is what the book page is for: thirty minutes with both of us, and you leave with a yes, a no, or a name.

If that is your week too, thirty minutes is enough to find out.

Book 30 minutes →