How do you choose a paid media agency?

Paid media is the most crowded agency category in growth marketing. Roughly half of the ~150 agencies in the Growth Partner Index touch Meta, Google, or TikTok in some capacity, which means selection is not about finding a competent buyer. Competent buyers are easy to find. Selection is about finding a team that will still be sharp two years in, after the senior operators from the pitch deck have rotated to new accounts and the iOS/Android/Google privacy goalposts have moved again.

The defining shift since 2022 is that platform-reported ROAS has become structurally unreliable. Apple's App Tracking Transparency, third-party cookie deprecation in Chrome, and Meta's modeled-conversions push mean two agencies running identical campaigns on the same account can report ROAS numbers 30–50% apart, both honestly, just by toggling attribution windows and modeled-conversion settings. The good shops know this and proactively bring incrementality testing, geo holdouts, MMM, or clean-room work. The mediocre shops continue to dashboard platform ROAS and hope you don't ask hard questions.

Pitches almost always feature senior operators. Strategists with 10+ years of Meta or Google experience, often with a recognizable brand on their resume. The single most important question to ask is whether those people will actually be on your account, and for how many hours per week. Holdco-owned agencies and many fast-growing independents have a structural mismatch: senior operators sell, juniors deliver. This is not always disastrous. A well-run pod with one senior strategist overseeing three juniors can outperform a single overstretched veteran. But it should be explicit, not hidden.

When a case study is presented, ignore the headline ROAS and ask three follow-ups. First, what was the baseline before the agency took over? Second, what is the attribution window and conversion source? Third, what was the spend trajectory. Did ROAS improve because the account was scaled efficiently, or because spend was pulled back? A 4x ROAS on $50k/month after a 3x at $200k/month is not a win, it's a contraction. Strong agencies volunteer this context. Weaker ones cherry-pick the favorable month.

Paid media agencies have natural operating bands. Below roughly $30k/month, boutique shops and consultancies dominate because the economics don't support a full pod. Between $30k and $250k/month, mid-market independents thrive. They can dedicate a senior operator plus one or two analysts and still make margin. Above $500k/month, the holdco and tier-one independent network kicks in because incrementality testing, brand-safety controls, and multi-market coordination become full-time jobs.

Hiring outside your tier is the most common. And most expensive. Mistake. A boutique shop at $1M/month spend will be overwhelmed within a quarter; a holdco team at $25k/month will quietly assign your account to a junior pod and lose interest. Ask agencies what their median client spend is and the range. If your spend is below the 25th percentile or above the 75th percentile of their book, the fit is questionable.

Across Meta, TikTok, and increasingly Google, creative is now the dominant performance lever. Algorithmic delivery has commoditized audience targeting, which means the agencies that win are the ones that produce or commission more concepts, test them more rigorously, and refresh winners before fatigue sets in. Ask any paid agency how many creative variants they shipped against your spend tier last quarter, who produced them, and how decisions to scale or kill were made.

Agencies that 'don't touch creative' are increasingly a poor fit unless you already have a strong in-house performance creative team or a separate creative partner. The handoff cost between media and creative is real, and the agencies with the highest creative-media integration scores in the Confidence Score consistently outperform on extended-window measurement.