In-house or agency media buying: the decision

The in-house versus agency question resurfaces at every studio the moment monthly spend crosses the point where an agency's retainer starts to look larger than a junior buyer's salary. The maths alone rarely settles it. The two options trade different things away, and the spreadsheet only prices one of them.

Fee models shift incentives, not just cost

Retainer structures, percentage-of-spend structures and hybrid performance fees each put a different kind of pressure on what an agency recommends. Percentage-of-spend deserves the most scrutiny. Under that model the agency's revenue grows with the budget whether or not efficiency follows, so a quarter where CPIs drift upward still pays the agency more, which is worth raising directly in any renewal conversation. A flat retainer removes that particular distortion; it can also leave less reason to chase incremental gains once the fee is locked in.

Transparency and speed favour in-house, with caveats

In-house buying gives direct account-level visibility into bids and audience settings, plus the platform-level test results behind them. Agency arrangements vary widely in how much of that raw data gets shared versus summarised into a monthly report, and the gap matters most exactly when it's least convenient, while somebody is trying to diagnose a sudden CPI shift in real time. In-house teams also tend to iterate creative and bidding changes faster, since there's no brief-and-approve cycle running through an external partner. That advantage narrows, though, for a studio testing only one or two markets rather than running programmes across a dozen.

Agencies still win on channel-level pattern recognition

An agency managing several client accounts across a channel accumulates pattern recognition faster than a single in-house team working with one studio's data ever can. That's a genuine argument for agency partnerships on newer channels a studio hasn't built expertise in yet (connected TV, say, or a new regional programmatic partner). It isn't a reason to default to agencies everywhere.

Where the crossover tends to sit

Informal industry practice places the crossover somewhere around the monthly spend level where one or two dedicated buyer salaries would cost less than the agency fee. The exact point moves with regional buyer cost and channel complexity, so model it per studio rather than borrowing a fixed rule of thumb from another team's experience.

The switching cost nobody puts in the comparison

Neither option is static once chosen. Moving from agency to in-house, or the reverse, carries a transition cost that rarely appears in the initial comparison yet shapes how the decision actually plays out. Bringing buying in-house means rebuilding platform-level learning that an agency accumulated over months or years, because account history doesn't always transfer cleanly even with a cooperative handover, and neither do audience exclusions or creative learnings. Moving the other way carries the opposite risk: an agency starting fresh on an account loses the context an in-house buyer built up about what the studio already tried and what failed.

A defined overlap period is the practical answer. Run the outgoing arrangement alongside the incoming one for four to eight weeks rather than a hard cutover on a single date, even though it costs more during that window. The overlap gives the incoming buyer, whether that's a hire or an agency, time to absorb account history before taking full ownership, and it materially reduces the efficiency dip a cold handover typically produces in the first month.

This switching cost is also the reason to revisit the decision on a planning cycle rather than reactively. A decision made under pressure, after a bad quarter with an existing agency for instance, tends to skip the overlap period entirely in the rush to change something. That's usually when the transition costs the most.

A scorecard, run by channel

Weight transparency separately from speed, and cost separately from channel expertise, for each channel on its own rather than deciding in-house-versus-agency once for the whole budget; that produces better outcomes than a single studio-wide call. A studio can reasonably run paid social in-house because volume and internal expertise justify the investment, while keeping a specialist agency for a channel like connected TV or a market it's entering for the first time.

The hybrid pattern that actually works

The pattern that holds up best across studio sizes keeps the core, highest-spend channels in-house. It retains an agency specifically for new-channel testing and market entry. And it writes a data-sharing clause into every agency contract guaranteeing access to raw platform data rather than only summarised reporting. Revisit that split at each planning cycle. Deciding once and leaving it unreviewed for years is how the arrangement drifts away from where the studio actually is.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• China's outbound publishers: the UA playbook
• Retail media networks and mobile games