Three companies quote the same project. One calls itself an agency, one a dev shop, one a studio. The numbers land close enough that the choice looks like a coin flip. It is not.
The words, and what they hide
A product studio owns a product from research through launch and past it, with the same people doing the thinking and the building. An agency sells deliverables against a brief. A dev shop sells engineering capacity against a specification. The difference is not headcount or price. It is who carries the answer to whether the thing works.
Agency is a word borrowed from advertising. The model is a brief in, deliverables out, with a client services layer sitting between you and whoever does the work. That model is good at campaigns and brand, where the deliverable is the product.
Dev shop is the engineering half on its own. You bring a specification, they build against it. Fast and cheap when the specification is right. Expensive when it is wrong, because nobody in the arrangement is paid to tell you so.
Studio is the older word, out of design and film. It means the people who make the thing are the people who decided what to make. There is no handoff, because there is nobody to hand off to.
Where the models actually diverge
The gap shows up in three places, and none of them appear on the quote.
| Agency | Dev shop | Product studio | |
|---|---|---|---|
| Starts with | A brief | A specification | A question |
| Sells | Deliverables | Hours | Outcomes |
| Who you talk to | Account manager | Project manager | The people building it |
| Owns whether it works | You | You | Them |
| After launch | A new brief | A new contract | The same people |
The last row is the one that costs money. A brief-driven engagement ends the moment the deliverables land, which is also the moment you find out what you actually needed.
The question that separates them
Ask all three what they would do if discovery proved the project should not be built.
An agency reshapes the brief. A dev shop builds it anyway, because the specification is the contract. A studio tells you, and that conversation is the thing you were paying for.
When a studio is the wrong answer
If you have a finished specification, product judgment in-house, and steady work for the same skills every month, a studio is overhead. Buy engineering capacity instead. The model earns its price when the problem is still a question, not when it is already an instruction.
Plenty of good work needs nothing this deep. A landing page needs a designer. A well-specified integration needs an engineer. Reaching for the heaviest model on a light problem is its own kind of waste.
What to check before signing
- Ask who will be in the room in month three. Names, not roles.
- Ask what happens to the price when scope changes, and get that answer in writing before it does.
- Ask for work done under your constraint rather than in your industry. A regulated build in a different field beats an unregulated one in yours.
- Ask what they have talked a client out of. A shop that never has is a shop without the standing to.
Market data on this distinction is thin, because every label here is self-assigned. What the data does show is scale: Clutch puts the average software development project near $132,000 and around thirteen months.
Thirteen months is long enough to regret the cheapest quote.