Cost is the wrong first question
Most founders open with the day rate comparison, and it is a genuinely misleading number. A designer's salary is not the cost of an in-house designer: recruiting time, ramp-up, equipment, tooling, management attention and the risk of a bad hire all sit on that side of the ledger. An agency's rate is not the cost of an agency either, because it usually excludes the time your team spends briefing and reviewing.
More usefully: both options cost roughly what good design costs. What differs is what you are buying, continuity or range.
In-house buys continuity
An in-house designer accumulates something an agency structurally cannot: memory. Why that flow is odd, which customer complained, what was tried in 2024 and failed. After a year, that context is worth more than raw craft.
In-house is the right answer when the product is your whole business, when design decisions come up daily rather than in projects, and when you have someone who can give a designer direction. That last one is the constraint founders underestimate. A designer with no product direction produces beautiful work that solves the wrong problem, and it takes about six months to notice.
An agency buys range and a start date
Agencies are worth it in three situations. You need several skills at once and none of them full-time: research, interface, front-end, back-end. You need to start now, because hiring well takes three months and you do not have three months. Or the work is a bounded project with an end: a redesign, an MVP, a design system.
The structural advantage is that a studio has seen the same problem in other contexts. We have shipped across fifty-seven documented projects in eight sectors, so a checkout problem in one is informed by three others. An in-house designer sees your product deeply and one product only. Both kinds of knowledge are real; they are just different.
Five questions that decide it
- Is design a project or a permanent function here? Project → agency. Function → in-house, eventually.
- Can someone give a designer direction on Monday morning? If not, in-house will underperform regardless of who you hire.
- Do you need more than one skill? One skill → hire it. Four → you are describing a team.
- What happens if this slips two months? If the answer is serious, hiring is too slow.
- Would you rather own the knowledge or rent the range? Answer honestly, and the choice usually makes itself.
The answer is often both, in sequence
The pattern we see most is a studio building v1 and a first designer being hired around the time the product has real users and a real roadmap. That sequencing works: the agency absorbs the risky, wide-ranging early phase, and the in-house hire inherits something coherent rather than a blank file.
Which is why we hand over the source, the Figma and the repository from day one. If your first designer joins in a year and cannot pick up where we left off, we did the job badly.