Agency, Freelancer or In-House? The Decision Compared

The short answer depends on your ad budget, and it comes in these tiers: below 3,000 euros of media budget per month you run it yourself. 3,000 to 10,000 euros: a freelancer. 10,000 to 150,000 euros: an agency. Above 150,000 to 240,000 euros: your own team. Between 10,000 and 15,000 euros the number of channels decides: one channel argues for a freelancer, two or more for an agency. The upper figure is calculated, not observed: based on a 10 per cent fee, three roles and 5,000 to 8,000 euros of fully loaded cost per person. Different assumptions move it considerably.

The second answer is the one most people would rather not hear. All three options fail for the same reason when they fail. Not because of the person or the agency, but because of messy measurement, an unclear target metric, and the fact that nobody in the company owns contribution margin. You have to solve that yourself, whoever you hire.

I am Nils Fehr, founder of AdSynergy in Berlin. We have worked with over 50 brands and managed more than 25 million euros in advertising budget. My vested interest in the answer "hire an agency" is obvious, which is why this piece also sets out when I advise against it.

What do the three options cost and deliver?

The ranges below are our own observations from proposals, salary conversations and accounts that have been put in front of us, as of October 2026. They are not a market study and not a price list, and they shift with seniority, channel mix and region.

As of October 2026. Terms and notice periods are a matter of negotiation and say nothing about our own conditions, which are set out in the proposal.

When is an agency the right choice?

You do not buy an agency for the hours, you buy it for breadth and comparison: several specialist roles under one contract, and people who have many accounts as a benchmark. Cover when someone is out comes on top of that, provided it is written into the contract.

How do you know an agency fits?

• You run more than one channel, or you want to within the next few months.
• Creative volume is your bottleneck, not account maintenance.
• You need media buying, creative and tracking, but you do not want to fill three positions.

When should you save yourself the agency?

• Your media budget is below roughly 5,000 euros a month.
• You are looking for someone who only executes. A freelancer does that more cheaply.
• You have no time for questions and approvals. An agency without a counterpart on your side is money burned.

What is the most common mistake brands make with agencies?

Brands hand over the responsibility along with the work. The contract is signed, so internally the topic counts as closed. After that the reports arrive, nobody asks questions, and six months later the dashboard and the books do not match.

An agency can only optimise what it can see. If you do not share cost prices, shipping costs and return rates, you get optimisation against ROAS, because that is the only number inside the ad account. When you are choosing, ask to be shown how decisions are made: ask which campaign was switched off most recently, and which number that was based on.

When is a freelancer the right choice?

A freelancer is often the cheapest way to buy experienced hands, and the option with the highest single point of failure: seniority without the overhead, but also without cover.

How do you know a freelancer fits?

• Your media budget sits between roughly 3,000 and 10,000 euros a month.
• You have exactly one channel that matters.
• You need a clearly bounded piece of work: a tracking setup, building an account, a series of tests, a second opinion.
• You run things yourself and need craft, not strategy.

When does a freelancer not fit?

• You need creative in series. That is a production job, not a single role.
• More than two channels are meant to run at the same time, or two weeks of absence would be expensive.
• Nobody internally can judge the work on its merits. Then the corrective is missing.
• Disguised employment (Scheinselbstständigkeit). Anyone who works only for you on an ongoing basis, keeps fixed hours and works with your tools and processes can be treated as falsely self-employed. Back payments of social security contributions land on you as the client. Several clients, their own equipment and a clearly bounded scope of work keep the risk small; if in doubt, the status determination procedure (Statusfeststellungsverfahren) run by the German state pension insurance, Deutsche Rentenversicherung, settles the question.

What is the most common mistake brands make with freelancers?

Many brands buy hours and expect accountability. A retainer of ten hours a month covers account maintenance, not strategy, not creative, and not repairing broken tracking. So write down in advance what is included, which number the work is judged on, and what happens when the person is unavailable. Ask for a short test log as part of the deliverable, because the knowledge sits with one person and leaves with them.

When is in-house the right choice?

In-house is usually expensive to build and can become cheaper to run, once the budget is large enough. You build knowledge that stays.

How do you know in-house fits?

• Your media budget is permanently above roughly 150,000 euros a month, and you can fill more than one role.
• Advertising is not a function in your business but part of the product, with subscription models for instance.
• Product and media decisions are linked so closely that external coordination costs more time than the work itself.
• You plan in years and can absorb three to six months of onboarding.

When does in-house not fit?

• You can hire one person but you need three skill sets.
• Your business is strongly seasonal and the budget swings by a multiple. A position does not swing with it.
• You cannot judge, on the merits, who you are hiring. A bad hire easily costs half a year.
• You want to open a channel that nobody internally has built before, Google Ads with Shopping and Performance Max for instance.

What is the most common mistake brands make with in-house?

Many hire one person and expect a team: the job ad looks for someone to cover Meta, Google, TikTok, creative, email and reporting. That profile costs more than the advertised range and rarely stays long, because it is structurally overloaded.

At what ad budget does each option add up?

The rule of thumb: fees plus tools should stay permanently below roughly 20 per cent of your media budget. 20 per cent is the ceiling for small budgets, not the normal case; 10 to 15 per cent is usual.

What fits below 3,000 euros of media budget a month?

At this level no externally managed setup covers its own cost. A fee of 2,000 euros against a media budget of 3,000 euros is 67 per cent of the budget; out of 5,000 euros of total spend, 40 per cent then goes into management rather than reach. Realistically you run it yourself, perhaps with four to eight freelancer hours.

What fits at 3,000 to 10,000 euros of media budget?

A freelancer for the main channel plus external creative production. On a media budget of 5,000 euros, an agency fee of 2,000 euros costs 40 per cent of the budget; out of 7,000 euros of total spend, 29 per cent then goes into management. An agency pays off here above all when tracking and the account structure still have to be built. We ourselves only work from 5,000 euros of monthly ad budget, because below that the maths does not add up for you.

What fits at 10,000 to 150,000 euros of media budget?

In this range an agency is, in the cases we see, usually the most workable arrangement, and there are no reliable market figures on it. At 30,000 euros of media budget and a 10 per cent fee you pay 3,000 euros a month and get media buying, creative and tracking. A single in-house position costs roughly 5,300 to 9,500 euros once you count fully loaded cost and tools, and it does not cover those three areas.

Where is the calculated tipping point towards in-house?

Calculated for one person: a fee of 10 per cent exceeds the fully loaded cost of one position from roughly 53,000 to 95,000 euros of monthly media budget. That is only half the calculation, because one person does not cover Meta, Google, creative and tracking at the same time.

For a multi-channel mix you need three roles: media buying, creative production, and data and tracking. That is 15,000 to 24,000 euros of fully loaded cost plus tools, which moves the tipping point to roughly 150,000 to 240,000 euros of media budget. Below that, in-house is a decision for control, not for saving money, and you carry the cover and recruiting risk yourself.

Which calculation counts in the end?

Everything above is cost accounting. The question that counts is a different one: does the management of your advertising generate more additional contribution margin than it costs?

You can only check that backwards, and only approximately. Set the contribution margin of the three months before you hired against the average of months four to six afterwards, at a comparable media budget. An example: 12,000 euros before, 16,200 euros after, a fee of 3,000 euros, leaving 1,200 euros. Only geo or holdout tests, in which one region is deliberately left without advertising, show clean incrementality. Everything else counts seasonality and product changes along with it.

Which hybrid models work in practice?

The either-or question is usually the wrong question to begin with. We see three combinations regularly.

In-house management plus external creative production

The model we most often see running stably. Account maintenance is learnable and repetitive, while creative volume needs people who are filming, editing and writing continuously. A team asked to do both delivers too little on creative, because the account work is urgent every day and creative is only important. Why creative remains the biggest lever is covered in the piece on Meta Ads for DTC brands.

Agency to build, in-house to run

Sensible when a new channel is being added. The agency builds structure, tracking, naming conventions and the first series of tests, after which someone internally takes over running it. That needs two conditions: the handover is written into the engagement from the start, and there is a written test log. Without those, the build phase becomes the permanent state.

In-house lead plus a freelancer per channel

One internal person owns the target metric, budget allocation and measurement, and one freelancer works on each channel. Cheap and flexible, but it stands or falls with that person: if they are out, three channels optimise against each other.

What works in no hybrid model at all: two parties on the same channel with the same budget. Then nothing gets optimised, things get negotiated. And if freelancers are permanently embedded, have fixed hour allocations and are managed internally, settle the disguised employment question with a tax adviser or a lawyer. That is not something we can assess.

How do you know it is time to switch?

• From freelancer to agency: the bottleneck is no longer the account, it is the number of creatives. A second channel has been waiting for months, and one holiday recently brought everything to a stop.
• From agency to in-house: the fee is above the fully loaded cost of the two or three roles you need. The work is mostly routine, you ask the big questions yourself, and there is someone internally who would take it on.
• From in-house back to external: the one person who knew everything has resigned. Nobody from outside has looked into the account for a year. The explanations for falling efficiency sound the same every quarter.

How do you prepare the switch without losing data and your learning curve?

The points below come from practice and are not legal advice. Have contracts, notice periods and ownership questions reviewed by a lawyer before you sign. The expensive part of a switch is never the fee, it is the knowledge that disappears. You settle these points before you give notice, not afterwards.

• Ownership of the accounts. Ad accounts, Business Manager and the Google Ads account run on accounts you own yourself. External parties get access, not ownership.
• Measurement infrastructure. Pixel, Conversions API, analytics property, tag manager container and server container belong to you. Anyone sitting on someone else's containers loses the history.
• Data processing agreement. Personal data is processed through ad accounts, pixels and analytics. The agreement under Article 28 GDPR (Auftragsverarbeitungsvertrag) belongs in the engagement, together with the rule that access is revoked on the day it ends.
• Raw creative material. Not only the exported ads, but project files, raw footage and copy variants. Without source files the next party starts from scratch.
• Account structure and test log in writing. Naming conventions, exclusion lists, negative keywords, audience logic, and alongside that: what was tested, what the result was, what was discarded. Without it the next party tests the same things again, at your expense.
• Overlap instead of a hard cut. Two to four weeks of parallel operation cost a one-off fee and save a rebuild.
• No rebuilding in the first month. Learning data needs quiet. Understand first, then change.

What stays the same across all three options and is still usually missing?

In the accounts I see in the audit, it is usually the same three gaps. None of them depends on who has been operating the account.

What does clean measurement mean in practice?

If only the browser pixel is measuring, conversions are lost depending on the browser and the consent situation. How many depends on the setup, and only a reconciliation between the ad account, the shop and analytics will show it. Alongside that you need a consistent UTM convention, deduplication between browser and server, and one source that counts when there is a dispute. If all three systems show different revenue, every discussion about channel budgets is a waste of time. More on this in the piece on conversion rate optimisation.

Why do you need one target metric instead of seven?

Many brands measure ROAS, CPA, click-through rate, hook rate, frequency and add-to-cart rate all at once, and still have no number that a decision hangs on. Settle on one leading metric, usually contribution margin per order or the ratio of total revenue to ad spend, and derive the target value from your margin: break-even ROAS is 1 divided by your contribution margin.

At a contribution margin of 40 per cent, a ROAS of 2.5 covers the advertising cost exactly. Your target value sits above that, by as much as you want to keep as margin. Every other number is diagnosis, not a target.

Who owns contribution margin?

One person in the company has to own that number, and it has to be someone who knows the ad spend as well as the cost prices, shipping, returns and discounts. This role cannot be outsourced. An agency owns its work, a freelancer owns their hours, a team owns its channel. The number at the end can only be owned by someone who sees all the parts.

With those three things in place, each of the three options has a realistic chance. Without them, every one of them gets difficult.

How do you decide in a week instead of a quarter?

Six questions, answered in writing, spread over six days.

• Day 1: set the media budget and the number of channels for twelve months, realistically.
• Day 2: calculate contribution margin per order after goods, shipping, returns and discounts.
• Day 3: set the creative requirement per month and name who produces it.
• Day 4: name the person who owns contribution margin.
• Day 5: have two conversations, one per option.
• Day 6: decide, put the target metric in writing, settle what happens when someone is out.

If day 2 or day 4 stays open, the choice between the three options is not your first problem. If you are above 5,000 euros of media budget and day 3 stays open, creative is your bottleneck and not account maintenance. How we approach channels, measurement and creative is set out in performance marketing from Berlin, and in shop-specific terms in the piece on Shopify implementation.

My conclusion

The budget decides the general direction. Channel breadth and creative requirement decide the model. And the quality of your measurement decides whether any of it works at all. The best decision is rarely the purest one, it is the one that clears your bottleneck, and often that is a hybrid.

If you are not sure where your bottleneck sits, an outside look into the accounts helps before you start thinking about contracts. It costs you 30 minutes: a call, then your audit within 24 hours (business days), free and without obligation. The terms are on the page about the free ads audit, and questions go through contact.

Frequently asked questions