Agency, studio, or freelancer: how to pick one for your website rebuild.

The three vendor models for a website rebuild are agency, studio, and freelancer. They differ in headcount, cost structure, and how the buyer interacts with the people doing the work. Most marketing copy blurs these intentionally, because the markup at one end of the spectrum disappears when buyers can see the structural differences clearly.

An agency is typically 10 to 200 staff, with dedicated account management, sales, design, development, project management, and quality assurance. The buyer signs with a sales team and runs day-to-day work through an account manager who relays to the team. The agency owns the relationship; the named designer or developer pitched in the proposal is rarely the named designer or developer who actually delivers.

A studio is typically one to ten people, often principal-led. The buyer signs with the principal and works with the principal directly. There is no account manager between the buyer and the work. The studio owns the relationship, and the named principal is the same person who delivers.

A freelancer is one person, no team. The buyer hires on hourly or fixed-bid terms for a defined scope. The freelancer is the same person across pitch, delivery, and follow-up. Hourly rates are the lowest of the three; project complexity is capped by what one person can hold in their head.

The U.S. Bureau of Labor Statistics tracks Web Developers (SOC 15-1254) at strong projected employment growth through 2032 (BLS, 2024), and the Stack Overflow Developer Survey 2024 sampled 49,390 professional developers across the categories that staff all three models (Stack Overflow, 2024). The talent pool is the same; the packaging is different.

The honest cost question is not “what is the hourly rate” but “how much of every dollar you spend goes into building the thing.” Each model has a different answer.

At an agency, a meaningful share of your spend funds overhead the buyer never directly sees. Sales commission, account management, office lease, project management software, recruiting and onboarding, internal QA, and partner-channel margin all draw against the project budget before any code is written. Industry surveys at Clutch and similar B2B services directories show typical agency billing rates ranging from roughly $100 to $300 per hour, with the higher end concentrated in coastal and brand-led shops (Clutch, 2024). The portion that actually pays the people building your site is often 50 to 70 percent of the invoice. Buyers paying $200 per hour to an agency are paying $100 to $140 of it to people building the thing and $60 to $100 of it to people coordinating it.

At a studio, overhead is lean. Solo or small-studio principals carry the same coordination work the agency account manager would carry, but as a part of their own week, not as a separate cost center. Studio hourly rates of $125 to $225 are common, with the difference between a studio rate and an agency rate accounted for almost entirely by the absence of the agency coordination layer. The portion that funds the actual build is typically 85 to 95 percent.

At a freelancer, the rate is lowest because coordination is not a cost line. The Stack Overflow Developer Survey 2024 reported a U.S. median compensation of $130,000 for full-stack developers (Stack Overflow, 2024), which translates to roughly $62 per hour at full utilization. Independent freelancer hourly rates quoted on hiring marketplaces (Upwork, 2024; GoodFirms, 2024) cluster across a wide range from $30 to $150 per hour, with the variance driven by experience, specialization, and geography. The portion that funds the build is essentially the entire rate, but you are also paying for one person’s judgment, not a team’s.

The longer reference on what each tier should actually cost in 2026, with ranges, drivers of cost, and what is worth paying for versus what is overpriced, lives at What a Website Should Cost in 2026.

Every vendor model trades something off. The honest framing is what you lose, not what you gain.

With an agency, you give up direct access to the person doing the work. The senior people pitched in the proposal are often pulled to bigger accounts; the junior team that delivers your project is rarely named in the contract. You also give up speed. Every change request routes through an account manager, gets logged, gets scoped, and comes back as a change order. Communication half-life is measured in business days, not hours. In my own field experience, agency-client churn tracks communication friction more than it tracks delivery quality. Clients leave agencies they cannot reach, even when the work itself is good.

With a studio, you give up redundancy. If the principal is sick, on vacation, or in a deep block on another project, you wait. Studios scale work by sequencing, not by adding headcount, which means the calendar is the constraint. You also give up breadth. A studio that is excellent at custom-built service business sites may be the wrong shop for a 50,000-product e-commerce replatform. The principal’s judgment is the asset; pushing it outside its zone of fit is how studio engagements fail.

With a freelancer, you give up architectural review. One person making every call without a peer to push back is how scope drifts, performance regressions go unnoticed, and post-launch maintenance gaps appear. You also give up senior strategic framing. The best freelancers are excellent at a defined scope you bring them. They are not, in most cases, the right shop for a strategic rebuild that requires a thesis on what the site should be in the first place.

Numbers are ranges, gathered from public industry surveys (Clutch, 2024; GoodFirms, 2024; Upwork, 2024) and the Stack Overflow Developer Survey 2024. Geographic and specialization variance is meaningful; the table reports the central two-thirds of typical observed values.

The right vendor depends on the project shape, the budget, and how the buyer wants to spend their attention. Three rules cover most cases.

The middle option is where most service businesses with a real dependence on their website actually fit. Service businesses below $1M in revenue rarely justify an agency budget. Service businesses above $5M in revenue with a working website rarely justify the freelancer model when the website goes wrong. The studio range covers most of the buyers in between.

Each vendor model has a characteristic way it goes wrong. Recognizing the failure mode in advance makes it easier to interview against.

Agency bait-and-switch. The senior team named in the proposal is the team you meet at pitch. The team that delivers is junior. The fix is to demand named delivery staff in the contract, with weekly time-on-project reporting. Any agency unwilling to commit to named individuals is signaling that the named individuals will not be on your account.

Agency scope-creep billing.Change requests routed through account management become change orders, and change orders become invoices. The fix is to define done in the contract before signing. Any agency that resists writing “done” into the scope is signaling that “done” is the revenue model.

Studio principal bottleneck.The principal architect’s calendar is the gating constraint. If a project needs three months of focused build, the principal is unavailable for a third project for that period. The fix is to confirm calendar before signing and to prefer studios who are honest about availability rather than studios who promise everyone everything.

Freelancer scope discipline failure. The freelancer starts on a defined scope, hits an architectural decision the original scope did not contemplate, and either makes the call alone (often wrong) or stalls waiting for direction. The fix is to either keep the scope tightly bounded or accept that a freelance engagement may need a senior reviewer brought in for architectural calls, which often erodes the cost advantage that drove the freelance choice.

If your business runs on the website, hire a studio. If the website is a digital pamphlet that occasionally generates an inbound, a good freelancer is the right tool. If you are already at agency budget and have a real reason to be there (multiple properties, multiple stakeholders, procurement requirements), hire the agency, but contract for named senior people on your account day-to-day.

The trap is paying agency rates for studio work, which is the most common mistake I see. Buyers in the $40,000 to $150,000 range often default to agency procurement because that is who they have heard of, and they end up paying 30 to 50 percent of their budget for coordination layers that do not improve the outcome. A studio at the same total price spends 30 to 50 percent more on the actual build, which shows up as a better site.

The other trap is paying freelancer rates for studio work. Buyers with a defined budget often optimize for hourly rate alone, which rewards the cheapest available freelancer, which produces a site that launches but does not stand up under real traffic, real edge cases, and real maintenance. The lowest hourly rate is not the lowest total cost when the lifetime of the site is two years instead of seven.

For the technical specifics of what I actually build at studio rates, the Next.js development page covers when the stack is the right call, the engagement model, and what the deliverables look like.