Freelancer, Agency or In-House: How to Choose

Most articles comparing these three are published by an agency, and they conclude that you should hire an agency. The freelancer versions conclude the opposite. Neither is lying exactly; they are answering a different question from the one you asked, which is not "which is better" but "which is better for a business like mine, right now".

So here is the version that starts from your situation.

What you are actually choosing between

Four options, not three, because the fourth one is real and often correct.

• A freelancer or independent consultant. One person, usually deep in one or two channels, working directly with you. The person you brief is the person doing the work.
• An agency. A team, usually with an account manager between you and the people executing, covering more channels and carrying more process.
• An in-house hire. An employee whose whole attention is your business, and whose knowledge accumulates inside your company instead of leaving with an invoice.
• Nobody yet. You keep doing it, badly and cheaply, until there is enough revenue or enough clarity to make one of the above worth it. For a business under a certain size this is not failure, it is sequencing.

That fourth option deserves more respect than it gets. If you do not yet know which channel works for you, paying anyone a monthly retainer to find out is an expensive way to buy information you could get from a few hundred dollars of your own testing.

The variable that actually decides it

Not budget. Not company size. It is how many channels you need run, multiplied by how much of the work is repeatable.

One or two channels that need judgment every week is a freelancer's shape. Six channels that each need somebody's hands on them is an agency's shape, because no single person is genuinely good at six. Anything where the work is constant, high volume and specific to your product is an in-house shape, because you are paying repeatedly for context that an employee would simply have.

Here is the same idea as a table, on the dimensions that actually differ. I have left cost out of it deliberately, because the honest answer is that all three can be made to cost roughly the same and the differences below are what you are really buying.

Read the continuity row honestly if you are considering a freelancer. It is the genuine weakness of the model and no amount of good work removes it. If one person going quiet for two weeks would materially damage your revenue, that is an argument for a team, and I would tell a client so.

When an agency is the right answer

These are real, and I lose work to every one of them.

• You need four or more channels run at once. Paid search, paid social, email, SEO, creative production, maybe retail media. One person covering all of that is either lying or spread too thin to be good at any of it.
• You need volume production, not just strategy. If you need forty creative variants a month, that is a studio, not a consultant. Some agencies are genuinely excellent at this and a solo operator cannot match the throughput.
• Procurement requires it. Larger organisations often cannot contract a sole trader: insurance minimums, vendor onboarding, security review, invoicing through a purchase order system. That is not a marketing decision and you will not win it.
• You need cover, guaranteed. Campaigns that cannot go unattended during a launch week or a holiday period need more than one pair of hands, contractually.
• Nobody internally can judge the work. An agency's process, reporting cadence and account management exist partly to make the work legible to people who cannot evaluate it directly. That has real value if your team has no marketing expertise at all.

When hiring in-house wins

• The work is constant and product-specific. If your advertising depends on knowing this week's inventory, margins, seasonality and customer complaints, an employee who sits in those conversations will beat an outsider who has to be told.
• You are past the point where the knowledge should keep leaving. Every month you pay an outside party, the learning about your customers accumulates on their side. At some scale that becomes the expensive part.
• You need somebody accountable in the room. Not reachable. In the room, in the meeting, arguing about the roadmap.
• Compliance or sensitivity makes external access awkward. Regulated categories, patient data, financial data. Sometimes the simplest answer is that the person handling it works for you.

The usual failure mode here is hiring one generalist and expecting them to be excellent at paid social, search, email, SEO and creative. That person is rare and is not available at the salary the role was budgeted at. A common and better structure is one in-house owner who holds the strategy and the context, plus specialists brought in per channel. That is often what I am, and I would rather describe the arrangement accurately than pretend a freelancer replaces the hire.

When a freelancer wins

• One or two channels carry your growth. Concentration is the whole advantage. A specialist who has run the same channel for years across dozens of accounts has seen your problem before.
• You want the operator, not an account manager. With one person there is no translation layer between the brief and the build, which is where a lot of intent gets lost.
• The account needs judgment more than volume. Diagnosing why a cost per lead is bad is not a throughput problem. In one account it took 35 campaigns of iteration to move a cost per lead from $86 to $3.91, and the useful part was the decisions, not the labour.
• You need to start now. No onboarding sequence, no statement of work cycle.
• Budgets are modest. Below a certain spend, agency retainers consume the money that should be buying media.

Own your accounts, whoever you hire

This is the section I actually want you to keep. The single most expensive mistake I see is not choosing wrong between these three options. It is letting whoever you hire own your advertising and analytics assets, and then discovering at the end of the relationship that your history, your audiences and your data belong to them.

It usually happens innocently. An agency spins up a new ad account inside its own business portfolio because that is faster than waiting for your access. Two years later you leave, and the campaign history, the pixel and the custom audiences stay behind. Rebuilding is not just administrative: you lose the accumulated optimisation data, and new ad sets go back into the learning phase from zero.

The correct setup is the same in every case and takes about twenty minutes.

1. You create the business portfolio and the ad account. Not them. Then add the freelancer or agency as a partner with access to your assets, which is exactly what Meta's partner access flow is designed for [1]. Access can be withdrawn in one click and everything stays yours.
2. The pixel or dataset belongs to your portfolio. If server-side tracking is being set up through the Conversions API, confirm it points at your dataset, not one you cannot see [2].
3. You are the Administrator on the Google Analytics property. Give the contractor Editor or Analyst, whichever they need, and keep Administrator yourself. Analytics permissions are role based and designed for exactly this [3].
4. You are the verified owner in Search Console. This one matters more than people realise: only a property owner can grant permissions to other users, so if the agency verified the property, they control who else sees it [4]. Verify it yourself through DNS, then add them as a full user.
5. Domain, hosting and DNS stay in your name, on an email address that is not the contractor's.
6. Ad spend is billed to your card, not theirs. Paying media through a third party's account means you cannot audit what was actually spent. I have always had clients pay Meta directly for this reason, and I do not charge a percentage of ad spend, because it rewards spending more rather than spending well.

Any competent professional will agree to all six without hesitation. Reluctance is the most useful signal you will get in the entire hiring process, and it is worth deliberately testing for.

What to ask, and what a good answer sounds like

Ask fewer, harder questions. These five are the ones that separate people quickly.

The failure question is the one I would not skip. Anyone who has run accounts for a few years has a story about a tracking break that cost real money, or a campaign that never worked. Someone who cannot produce one has either not done much or is not being straight with you. It is the reason every case study on this site includes what did not work, and why the caterer case study says out loud that the authority score went down while traffic more than quadrupled.

Red flags, in order of how much they cost you

1. Guaranteed rankings or guaranteed lead volume. Nobody controls the auction or the algorithm. A guarantee means either a meaningless metric or a coming argument.
2. Resistance to you owning the accounts. See above. Treat it as disqualifying.
3. A price quoted before anyone looked at your account. That is a package, not a plan. I quote after an audit because a number given before looking is a guess.
4. Reporting that only contains numbers that went up. Impressions and reach rise easily and mean little. Ask for cost per lead, cost per acquisition, or revenue.
5. No named person with a public track record. You should be able to find the individual, not just the brand.
6. Charging a percentage of ad spend with no floor. It aligns their incentive with your budget growing, not with your cost per lead falling.

The decision, compressed

What this does not mean

• I have never run an agency. Everything above about how agencies work internally is observation and client reports, not experience. Some agencies do not resemble the description at all.
• Category beats individual, but only weakly. A good agency beats a mediocre freelancer and a good freelancer beats a mediocre agency. The framework above narrows your shortlist; it does not pick the person.
• My sample is skewed. Businesses that hire an independent specialist after an agency are more likely to have had a poor agency experience, because the happy ones never came to me. I have tried to correct for that here and probably have not corrected fully.
• No pricing figures in this article are implied. I have deliberately not quoted rates for any of the three, because the ranges are wide enough that any number I gave would mislead more than it helped.

If you take one thing: decide based on how many channels need running and how much internal context the work requires, then, whoever you pick, make sure every account is in your name before the first campaign goes live.

Related reading

• The 12 checks I run when auditing a Facebook Ads account
• Facebook Ads vs Google Ads: which one do you need?
• Ranking a medical clinic on Page 1 in three months
• How my own engagements are structured