Before you compare quotes, you make a bigger decision: which model should the work run through at all? Freelancer, agency and in-house team differ less in price than in three things — who carries the risk, where the knowledge stays, and how quickly decisions get made.
This article is an in-depth expert contribution from our content cluster. Discover the complete overview on our main page:All services at a glance →
- Comparing prices deceives: an hourly rate says nothing about total cost. What decides it is coordination effort, knowledge loss, and who fixes mistakes.
- Every model has a hard limit: the freelancer's is availability, the agency's is distance from the actual craft, the in-house team's is utilisation.
- The honest question is not "who is cheaper" but: is this task permanent, and does it belong to the core business? The answer follows almost by itself.
- Why comparing prices misleads
- The four axes of the decision
- The three models in profile
- The hard limits — including mine
- The hybrid that often gets overlooked
- What to demand regardless of model
- Decision guide by starting position
- The first meeting as a selection tool
- How external and internal work together
1. Why comparing prices misleads
The usual approach is to lay quotes side by side and compare the numbers. That almost inevitably leads to a wrong decision, because the three models put entirely different costs into entirely different pots.
An employee does not cost their salary but salary plus overheads, workplace, tools, training and the time during which they are not yet productive. An agency costs not only its fee but also the coordination effort across several people. A freelancer often looks cheapest and carries a risk that only becomes visible when they are unavailable.
That is why this article deliberately avoids hourly and daily rates. They are the least suitable yardstick that exists for this question — and for the same reason I work at fixed prices myself: what counts is the price for a result, not the price for an hour.
2. The four axes of the decision
1. Who carries the risk?
With a fixed price the provider does: if it turns out harder than expected, that is their problem. With time-and-materials billing you carry it. With your own employees you carry it permanently, including in periods with no project.
2. Where does the knowledge stay?
Often the decisive point. Knowledge existing only in an external person's head leaves the building with them. Knowledge in your own team stays — as long as the person stays. Documentation is therefore not a formality but the actual object of delivery.
3. How quickly are decisions made?
Every layer between you and the person actually building costs time and precision. A direct line is not a comfort but a quality feature: questions get resolved in conversation instead of travelling via two intermediaries.
4. How many disciplines do you need at once?
A website needing copy, imagery, engineering and campaigns simultaneously is a different thing from a scoped technical project. The more specialisms run in parallel permanently, the more a structure with several heads carries.
3. The three models in profile
Freelancer / specialist
Short paths, deep expertise in their own field, no surcharge for structure. You talk to the person who also builds. Limit: availability and breadth.
Agency
Several disciplines under one roof, cover during absence, established processes. Limit: distance between sales and delivery, and who actually ends up working.
In-house team
Full control, knowledge stays in the building, available at all times. Limit: fixed costs regardless of utilisation, and the difficulty of staying current.
The usual case: a mixture
In practice almost nobody decides purely. The question is not which model, but which part of the work belongs in which model.
4. The hard limits — including mine
A comparison in which the author's own model happens to always win is not a comparison. So here are the limits openly, mine included.
Freelancer: the bus risk
One person can fall ill, be overbooked, or stop. Anyone hiring an individual has to settle beforehand what happens then — and whether the result is documented well enough for someone else to continue. With me the rule is therefore: what I build is handover-ready, and I state up front what I do not cover.
Agency: who actually works?
Selling is often done by experienced people, delivery by junior ones. That is not inherently bad — it only has to be transparent. The decisive question in a first meeting is: who exactly works on our project, and who will we be talking to?
In-house: the utilisation trap
Creating a role for a task that only arises periodically produces fixed costs and idle time. And a single specialist without peers falls behind professionally, because there is nobody to exchange ideas with.
All three: dependency
Whether agency, freelancer or a single internal person holding special knowledge — vendor lock-in arises wherever access, accounts and knowledge do not belong to you. That is a contractual question, not a question of model.
5. The hybrid that often gets overlooked
Between "one person alone" and "a whole agency" sits a model that frequently fits mid-sized companies best: one fixed point of contact who carries responsibility, plus a network of specialists for what they do not do themselves.
That is exactly how I work. You speak to the same person throughout, there is no ticket queue and no junior being trained on your project. For topics beyond my core — specialised security work, for instance — I refer partners deliberately rather than pretending one person can do everything. The difference from an agency is not breadth of competence but proximity to delivery: whoever sells also builds.
The question "who actually works on my project?" tells you more about the quality to expect than any comparison of reference lists.
6. What to demand regardless of model
-
A fixed price for a defined scope
Not because time-and-materials is disreputable, but because a fixed price forces the provider to understand the scope beforehand. Anyone unable to name a fixed price has not yet grasped the task — or wants to leave the risk with you.
-
A commitment on who concretely works
By name, not by role. Plus the arrangement for what happens if that person becomes unavailable.
-
All access in your name
Domain, hosting, analytics, code repository, model accounts. It is the only effective insurance against dependency — and the point at which separations get expensive when it is missing.
-
Documentation as part of the delivery
Not as an appendix but as a named component. The test question: could someone else continue with it without asking the original developer?
-
An honest list of exclusions
What is not included? A provider who does not supply that list has either not thought it through or is deliberately withholding it.
7. Decision guide by starting position
Freelancer or specialist at a fixed price. A relaunch, an automation, a prototype — tasks with a defined end are the ideal case for this model.
Agency or a settled network. When copy, design, engineering and campaigns run simultaneously and permanently, a structure with several heads carries better than one person.
In-house team. When the work never runs out and creates the difference against competitors, it belongs in the building — with external support for peaks and specialist topics.
None of the three — first a small, paid analysis. Awarding a project whose scope nobody knows is the most expensive variant of all, regardless of model.
8. The first meeting as a selection tool
Reference lists and portfolios are poor selection criteria. They show what a provider delivered at their best, under conditions you do not know, with a team that may now be a different one. The most informative tool is the first meeting — provided you run it as an examination rather than a presentation.
Five questions whose answers reveal more than any reference
1. "What would you advise us against?"
Anyone without an answer either did not listen or sells everything on principle. A provider talking you out of something has understood your situation and is risking their own contract for it — the strongest signal available in such a conversation.
2. "Who is at this table, and who works later?"
The decisive question with agencies. With an individual it answers itself — but there the question of cover arises instead. Both are legitimate, both must be answered.
3. "What is not included?"
The answer separates thought-through quotes from hopeful ones. Anyone who genuinely understood the scope can name its edges. Anyone who evades will have the discussion later — during delivery and under time pressure.
4. "What was your last project that went wrong?"
No career exists without such projects. What matters is not the mistake but whether someone can name it and what came of it. A flawless record usually means selection in the storytelling, not in the work.
5. "What does the end of the collaboration look like?"
Handover, documentation, access, knowledge transfer. Anyone answering that confidently is not planning around your dependency — and that is what you want to know before signing, not after.
A warning sign that often gets missed
If a provider asks not a single question about your business in the first meeting and instead presents a solution immediately, that is not a sign of experience but of standardisation. It can fit — if your case really is standard. It does not fit if the very reason you are seeking outside help is that it is not.
Quick check before awarding
9. How external and internal work together
Because the answer is almost always a mixture in practice, it is worth looking at how that mixture works without friction. Three splits have proven themselves, and they differ in where the boundary runs.
By time period: external builds, internal operates. The provider sets a system up and hands it to someone at your company who maintains it afterwards. That only works when the handover is part of the assignment from the start rather than appended at the end — otherwise your team inherits something it does not understand.
By subject area: what concerns your business stays internal, what requires specialist knowledge goes external. Your people know your customers and processes better than any provider; a provider knows the technical environment better than your people. This split uses both strengths instead of setting them against each other.
By utilisation: internal runs the base operation, external absorbs peaks and special projects. The big advantage is that no role is created for work that only arises periodically — and the big disadvantage is that external capacity is not available on demand. Anyone working this way should announce demand in advance rather than calling in an emergency.
The most common source of friction across all three variants is not technology but an unclear boundary: two parties each believing the other is taking care of it. A single sentence in the contract or project description naming responsibility for ongoing operation prevents most of those cases.
Conclusion
Freelancer, agency and in-house team are not quality tiers but answers to different questions. An individual is strong on scoped projects with short paths and weak on availability. An agency is strong on breadth and resilience and weak on the proximity between sales and delivery. An in-house team is strong on control and weak on fixed costs.
Anyone answering two questions honestly — is the task permanent, and does it belong to the core business? — has effectively already made the decision. Everything else is checking the five points from chapter 6, and those apply equally to all three models.
Have a vision?
Let's check together how we can make your idea take flight.
Book your free strategy call nowExtended Specialized Glossary
Vendor Lock-in
Dependency on a provider that arises when switching becomes technically or economically disproportionate.
MVP (Minimum Viable Product)
The smallest version of a product that already creates real value and can therefore be tested in the market.
