January is when most of my clients finalize the number for the year, and almost none of them ask the one question that actually determines what that number buys: how much of it is working media.
Working media is the share of a budget that buys placements, the actual impressions, clicks, and airtime. Everything else, agency fees, ad tech fees, data licensing, verification, production, platform margin, is non-working. On a straightforward paid search or paid social buy, working media usually lands between 85 and 92 percent. On programmatic display and video, once you stack a DSP fee, a data management fee, ad verification, and a supply path that runs through two or three exchanges before it reaches a publisher, working media can drop to 60 percent or lower. Nobody puts that number on the cover slide.
I've reviewed enough media plans to see the same pattern every January. A client signs off on two million dollars. They assume two million dollars of media runs. What actually runs is closer to one point four million, and the other six hundred thousand was always going to fees before a single impression served. That's not fraud. Most of those fees are disclosed somewhere in a rate card or an SOW. It's just never converted into the one number that matters, which is: for every dollar I approve, how many cents actually buy something. The disclosure structure is doing exactly what disclosure structures do: each fee is legible on its own page and the sum is legible nowhere, because no party in the chain benefits from adding it up. The DSP discloses its take, the agency discloses its fee, the verification vendor discloses its rate, and the only person positioned to total the column is the client, who was never told a total was missing.
The distortion nobody corrects for
This matters more than it sounds like it should, because working media percentage quietly distorts channel comparisons, which means it distorts the budget itself. A paid social program running at 90 percent working media will look more efficient than a programmatic program running at 65 percent, even when the programmatic buy is smarter, better targeted, and cheaper on a true cost-per-placement basis. If you're comparing channels on blended ROAS without normalizing for what share of spend actually bought media, you're not comparing performance. You're comparing overhead. And the distortion compounds annually: the channel penalized by its fee stack loses budget in each planning cycle for a crime it didn't commit, while the mix drifts toward whichever channels happen to have thin intermediation rather than whichever ones actually work. I have re-run channel comparisons on a working-media basis for clients and watched the rank order of their channels change. Same data, corrected denominator, different budget.
What I ask clients to do before signing an annual plan now, not after Q1 numbers come in soft: get the working media percentage broken out by channel, in writing, before the budget is approved. It's a one-line ask and most vendors will answer it directly if you ask directly. The ones who get vague are telling you something. And the numbers that come back become a management tool on their own: working media percentage is negotiable in a way performance never is. You cannot demand a better conversion rate from a vendor, but you can consolidate duplicated supply paths, drop the verification layer that duplicates what another layer already does, and renegotiate the data fee that predates your first-party audience build. I have found six figures in a media budget this way without changing a single targeting decision, because the money was never lost in the market. It was lost in the plumbing, and plumbing can be re-piped.
Start the year's math from the real number
None of this means fees are bad or agencies are gouging clients. Fees pay for expertise, technology, and the people managing the buy, and paying 15 percent for a well-managed program beats paying 8 percent for a neglected one every time. It means your growth math for the year needs to start from the number that actually buys media, not the number on the invoice. If the plan says two million and the real figure is one point four, build your Q1 targets off one point four, because the market only ever saw one point four. The gap between those two numbers is where a lot of "underperforming" campaigns turn out to have been correctly forecast all along: the forecast assumed dollars that were never going to reach an auction, and the campaign gets blamed for a shortfall the fee structure guaranteed at signing. Programmatic is where this gap runs widest, but every channel has one, and January, before the number is signed, is the only cheap time to ask.