Should we bring programmatic media in-house?

Every few years the in-housing debate comes back. The answer for most mid-market brands is not yes or no. It is: take control of the parts that matter, and stop paying for the parts that don't.

The in-housing conversation almost always starts with a fee. Someone compares what the agency charges for programmatic against the cost of two or three hires and concludes the brand could do it more cheaply itself. Sometimes that's right. More often the math is incomplete, and the real question underneath it is getting missed entirely.

What you're actually buying from the agency

Programmatic delivery through an agency bundles several things together into one line: platform access and seat costs, trading expertise, data partnerships, planning and strategy, reporting, and the agency's margin layered across each of those. When brands talk about in-housing, they usually mean taking over the trading specifically. But trading is the part of the bundle most likely to already be automated by the platform itself, and least likely to be where the actual value sits.

The value sits in control: control of the platform contract, control of the data, control of the supply path, control of measurement. A brand can have all of that without hiring a single trader.

The hybrid most brands should actually run

Own the platform contracts outright. Whatever demand-side platform the agency is trading through, the seat should be in the brand's name, with the brand's data sitting inside it. If the agency relationship ends, the history and the learnings stay with the brand instead of walking out the door.

Own the data. Log-level data, first-party data, and the measurement stack belong to the brand as a matter of course, not as a negotiated favor. This needs to be written into the contract explicitly, because it rarely happens by default.

Own the supply path decisions. Curated deals, private marketplaces, and direct publisher relationships should be set up on the brand's terms, not inherited wholesale from whatever the agency's trading desk already has in place. In audits we run regularly, it's common to find the same inventory being routed through far more supply-side platforms than anyone at the brand actually chose or would choose if asked. Nobody signed off on most of it. It accumulated.

Then, separately, decide who actually trades. For most mid-market brands, an agency or a specialist trading partner operating inside the brand's own seat, on a transparent fee structure, is more efficient than standing up a full in-house team. Senior programmatic talent is expensive to hire and genuinely hard to retain, and a small in-house desk carries real key-person risk that a brand rarely accounts for when it first runs the build-versus-buy math.

When full in-housing actually makes sense

When programmatic trading is a core competency of the business itself, not a support function. When spend is large enough to justify a real team rather than one or two generalists wearing multiple hats. And when the brand already has the operating discipline to run measurement and supply path management on its own, without leaning on an outside partner for that judgment. That combination describes some retailers, some direct-to-consumer brands, and some platform businesses. It doesn't describe most mid-market companies, and there's no shame in that. Not every capability needs to live in-house to be well controlled.

The mistake to avoid

Bringing trading in-house while leaving the agency's platform contracts, data, and supply path decisions exactly where they were. That combination hands you the headcount cost of an in-house team without any of the control that was supposed to be the point of going in-house in the first place. It's the most expensive version of this decision, and it's also the most common mistake we see brands make.

List what you actually want to control. List what you want to stop paying for as a bundled margin you can't see. Then look line by line at what the agency is genuinely charging for each piece of the bundle, rather than treating the whole fee as a single number to negotiate down. The right structure for a given brand tends to be obvious once the bundle is actually unpacked. The mistake almost every brand makes is trying to answer the in-housing question before doing that unpacking at all.

The decision is also not permanent. Ownership of contracts and data can be renegotiated into the brand's name well before any staffing question is settled, which means the higher-stakes build-versus-buy call can wait until the control question is already resolved and no longer clouding the math.