Performance marketing agencies sell accountability. The pitch is that you pay for outcomes rather than activity, that every dollar is traceable, and that the relationship is governed by a dashboard instead of a quarterly opinion. The pitch is attractive and partly true. What complicates it is that measurement has become harder rather than easier: privacy changes, cookie restrictions and multi device journeys mean the tidy attribution that made this model possible now requires judgement. This page explains what these agencies actually do, how each fee model changes what they do, and the small number of questions that show whether a candidate measures honestly.
What the discipline covers, and what it quietly excludes
In practice the label covers paid media across search, social, shopping, display and increasingly retail media, plus the conversion work directly attached to it: landing pages, offers, feeds and tracking. That is a real and demanding craft. What the label often excludes, without saying so, is everything that makes paid media cheaper over time. Brand familiarity lowers your cost per acquisition because people click and convert more readily on a name they recognise, and organic visibility carries demand that would otherwise be bought. An agency paid on last click outcomes has little incentive to invest in either, which is how companies end up with efficient campaigns and a business that stops growing the moment spend pauses. The practical fix is not to distrust the model but to name the boundary. Decide who owns the parts of demand generation the performance agency will not touch, and do not let the dashboard's silence on those parts be read as evidence they do not matter. FTC guidance on online advertising and marketing is a useful grounding for the claims and disclosures side of the work, which is easy to neglect when the whole focus is the auction.
Fee models change behaviour more than briefs do
A percentage of media spend is the oldest model and its flaw is obvious: the agency earns more when you spend more, and the moment to cut spend is precisely when its revenue depends on holding it. A flat retainer removes that pressure and replaces it with a different one, since a scaling account earns the agency nothing extra and may end up under attended. A cost per acquisition or revenue share model looks ideal, but it makes attribution the contract, and that is where disputes start: the agency counts a conversion its platform reports, your finance team counts a paid invoice, and the two numbers never match. Hybrid models, a modest base plus a bonus tied to one agreed and independently visible number, are the compromise most experienced buyers settle on. Whichever model you pick, ask two questions before signing. What happens to your fee if the right recommendation is to reduce spend for a quarter. And which system of record decides the outcome number, yours or the platform's. An agency that has thought about incentives will answer both immediately, because it has had the argument before.
Testing an attribution claim in ten minutes
Ask three questions and listen for hedging. First: when your dashboard reports a conversion, where does that number come from, the ad platform's own reporting or our system of record. Platform reported conversions include modelled estimates and each platform claims credit generously, so summing them across channels routinely produces more conversions than the business actually had. Second: how would you tell whether the campaign caused the sale rather than captured someone who would have bought anyway. Good answers involve holdouts, geographic tests or incrementality experiments and an admission that these are imperfect. Bad answers involve a longer explanation of the attribution model. Third: what will you tell me when the platform's number and my finance team's number disagree, because they will. The candidates worth hiring answer that they reconcile to your numbers and treat the platform as directional. When you are comparing the best performance marketing agencies for a specific account, that reconciliation question predicts the quality of the relationship better than any case study, because it is the moment where accountability either exists or evaporates.
Questions people ask about performance marketing agencies
Is paying on a cost per acquisition basis safer for me?
It moves risk to the agency and moves the argument to attribution. Before agreeing, settle which system counts a conversion, what happens with refunds and cancellations, and how long the attribution window is. Where those cannot be agreed cleanly, a base fee plus a bonus on one trusted number is usually the more workable structure.
How much should management fees be relative to spend?
Judge the fee against the work rather than as a ratio. Ask how many hours a month the account gets and from whom, then compare that with what the same hours would cost elsewhere. A percentage that looks small on a large account can buy very little attention, and one that looks large on a small account may be the only way it gets any.
Can an agency guarantee a return on ad spend?
Only by defining the terms narrowly enough to make the guarantee safe. Read what the guarantee measures, over what period and with what exclusions. A written plan with a named accountable metric and an honest reconciliation process is worth more than a guarantee you would never enforce.
Who should own the ad accounts?
You should, in every case, with the agency granted access. Account ownership carries your historical data and learning, and starting from a fresh account after a switch costs real efficiency. Agree ownership in writing at the start, when it is free to arrange.