Most marketing teams that have brought things in-house over the past few years did it for the same reasons: cut costs, move faster, and get more control.
In year one, it usually feels like a win. Agency spend comes down, things move quicker, and for the first time it feels like you’re actually in control.
Then year two rolls around and it gets a bit messier. The savings stop improving, the team ends up bigger than you planned, and everyone’s busy all the time. But the results, the ones that were meant to justify the whole shift, don’t really move.
This happens more often than anyone likes to admit. It is not because bringing things in-house was the wrong call. Most teams just start with the wrong things.
You moved the visible work. The harder stuff stayed.
When teams bring marketing in-house, they usually start with execution. Paid search, paid social, programmatic, campaign management. These roles are easier to define and hire for, and they feel like the most immediate way to take control.
What gets underestimated is how much sits underneath strong execution. Running paid media in a way that drives meaningful business results depends on more than platform knowledge. It requires testing infrastructure built over time, measurement systems that go beyond surface-level metrics, and a depth of experience across different market conditions and account structures.
Creative is often one of the largest drivers of performance, but it only works when it is part of a structured testing approach. Iteration cycles, clear hypotheses, and consistent feedback loops between creative and media are what allow performance to improve over time. Without that system, creative becomes subjective, and results plateau quickly.
Measurement has the same issue. Optimizing to platform metrics or short-term efficiency targets can make performance look stable while limiting growth. What matters is how activity connects back to revenue. That requires a measurement approach that reflects incrementality, contribution to actual sales, and where future growth is coming from, not just what has already converted.
At the same time, the role of the agency often narrows instead of evolving. What’s left is a partner expected to contribute strategically, but without the full context, authority, or integration needed to do that well.
The result is a split that works against itself. Execution moves in-house without the supporting systems and experience that make it effective. The agency is still involved, but in a reduced capacity that limits its impact. Costs increase, complexity increases, and performance becomes harder to diagnose.
What that costs you in practice
When the split lands this way, accountability fragments. Nobody truly owns performance outcomes end to end. The in-house team owns the execution but not the decisions that shape it. The agency owns some of the strategy but not the authority to act on it quickly.
The tell is in how measurement works. Teams in this situation use measurement to report on what happened rather than direct what happens next. When something underperforms mid-campaign, acting on it requires a conversation, a meeting, sign-off from someone not close enough to the data to move fast. Platform reps fill the vacuum. Decisions that should be made internally get deferred by default. You’ve added headcount and complexity without adding control, and more people in a broken structure just makes it harder to see what’s broken.
What a good agency relationship actually does
Getting the most out of a hybrid model means using the agency for what it’s actually built to do: cross-brand pattern recognition, testing infrastructure that compounds over time, platform access that doesn’t transfer with a job offer. Not just the work that didn’t make it onto the hiring plan.
Platform access is the part that’s hardest to see until you’re on the outside of it. Agencies maintain relationships with Google, Meta, TikTok, and others at a level most individual brands don’t reach independently. That means earlier access to products still in development, before they’re available to everyone else. Our partnership with TikTok, for example, gave us alpha access to Market Scope and Consideration Ads. The beauty brand we ran it for saw ad recall of 27.8% against a 6.57% benchmark, favorability up 15.1% from flat, and over 5,000% growth in their consideration audience. Our strongest brand lift results to date. A brand building that capability internally would have been starting from zero, without the relationship that made the access possible in the first place.
What the right split looks like
The question worth asking isn’t how much of your marketing you’re running in-house. It’s whether the split you have is intentional or just what accumulated over time.
Getting it right means mapping it explicitly: what does the in-house team genuinely own, what does the agency own, and where are the handoffs that slow decisions down? The framework we use with clients maps capabilities from functions where internal proximity is a genuine advantage, things like media governance, budget ownership, and brand decision-making, to functions where external infrastructure and cross-client expertise will outperform a single in-house team almost every time. The goal is to make the split deliberate rather than inherited.
If you’ve in-housed some of your marketing and performance still isn’t where it should be, the question isn’t what else to bring in-house. It’s who actually owns performance outcomes in your current setup, and is your measurement telling you what to do next or just recording what already happened? If the answer to either isn’t immediate, that’s where to start.
