How Agencies Actually Get Paid

This is business and market research, not accounting or legal advice. Contract-structure specifics (what counts as a "principal transaction," what a given state's rebate-disclosure law requires) should be reviewed by an attorney or accountant before you build a real contract around any of this.

---

#The three real models, and which one actually dominates

Commission on media spend. The agency takes a percentage — most commonly 10–20% — of whatever the client spends on media, on top of (or instead of) a separate fee. [Directional] — consistent across multiple agency-pricing sources.

Flat fee / retainer. A fixed monthly amount regardless of spend size, typically $2,500–$15,000/month for a single-channel program at the smaller end, scaling well beyond that for multi-channel, larger-budget accounts. [Directional]

Performance-based. Compensation tied to a measurable outcome (cost per lead, cost per sale, a revenue share). This is the model every client instinctively wants and the one agencies use least — it accounts for roughly 10–15% of real-world pricing arrangements, specifically because attribution complexity (which touchpoint actually caused the sale?) makes a clean, disputable-proof-free performance metric hard to construct and even harder to agree on contractually. [Directional]

Which model actually wins in practice depends on which survey you read, and the two available don't fully agree — worth flagging rather than picking whichever number sounds cleaner. One 2026 agency-pricing survey found 42% of agencies use flat fees, 31% use percentage-of-spend, 27% use hybrid models. Separately, the 4A's own 2024 Compensation Methodologies Survey found 72% of agencies use fixed fee as their primary compensation model. [Directional] for both — different survey populations and methodologies, but they agree directionally on the same conclusion: flat fee has become the dominant real-world model, and the historical "15% of spend" commission that defined the industry for most of the 20th century is now a minority structure, not the default.

#Why "15% of spend" stopped being the default

The classic commission model was eliminated over time for a straightforward reason: as media buying got more technical and more third-party ad-tech vendors inserted themselves into the supply chain, agencies increasingly earned a percentage of a number they didn't fully control or benefit from — a percentage of gross spend when a meaningful chunk of that spend was actually flowing to ad-tech intermediaries, not media. Clients pushed back, and transparency requirements pushed the market toward flat fees, where the agency is paid for its own labor and outcome, decoupled from the size of the client's media budget. [Directional]

#The commission math, worked

A "15% commission" is genuinely ambiguous unless you're specific about what it's 15% of — and the ambiguity itself has been a source of real disputes. Two versions:

• 15% of gross spend: the client pays $100,000 in total, of which $15,000 is the agency's fee and $85,000 is net media.
• 17.65% markup on net media cost: the agency buys $85,000 of net media, marks it up 17.65%, and bills the client $100,000 total — same $15,000 fee, same $100,000 total, different-sounding percentage depending on which base you quote it against. ($85,000 × 0.1765 ≈ $15,003.) [Established] as arithmetic; the two framings describe the identical dollar outcome, which is exactly why a client reading only one of the two numbers can be misled about what they're actually paying.

Spend-tier pricing. Percentage rates typically fall as managed spend rises — an agency might charge 15% managing $50,000/month but only 8–10% managing $500,000/month for the same client, because the agency's actual labor doesn't scale linearly with the size of the media budget once campaigns are built and running. [Directional]

#Worked example: what a $50,000/month retainer actually nets the agency

Take a client spending $50,000/month on media at a 15% commission, billed as a markup on net spend (the $50,000 is the client's total budget, media plus fee). Agency fee: $50,000 × 0.15 = $7,500/month, or $90,000/year, for this one account. Module 4 walks through what it actually costs to service that account (staff time, tools, overhead) against this $90,000 top-line number — the commission rate alone doesn't tell you whether the account is profitable.

#The AOR contract, structurally

An agency of record (AOR) designation is contractual, not informal — a brand names one agency its lead partner for a defined marketing remit, usually for a term measured in years, with a named account team, a defined scope, and often exclusivity within that scope. [Directional]

The financial structure of a real AOR contract commonly stacks several of the models above rather than picking one: Total AOR investment = annual retainer + (project fees × number of discrete projects) + media commission percentage. [Directional] A retainer covers the always-on strategic and account-management work; project fees cover discrete deliverables (a new creative campaign, a landing-page build) outside the retainer's defined scope; the media commission covers the ongoing execution layer. This stacking is exactly why "how much does an agency charge" doesn't have one answer — the real number is contract-specific, built from several components layered on top of each other.

#The mechanism regulators spent a decade fighting: opaque programmatic margin

This is the part of "how agencies get paid" that doesn't show up in a pricing FAQ, and it's worth understanding even if you never intend to run this way, because a client evaluating your agency has likely read about it.

The ANA's K2 Intelligence investigation (the definitive, disclosed-operator-tier report on this) found that when a holding-company trading desk acted as the actual media supplier — a structure called a principal transaction — markups on that media ranged from 30% to 90%, with some trading desks internally mandated to hit 30–50% margins. Ad-server markups in some documented cases ran 200–250%, undisclosed to the client. Rebates from media suppliers to agencies showed up as cash, free media, debt forgiveness, or even equity — one documented case involved a client spending over $5 million receiving roughly 10% of that back as free media (~$500,000), with no clear accounting for how (or whether) that value passed through to the client who generated it. [Established] — directly sourced from the ANA/K2 Intelligence report via AdExchanger's reporting, a disclosed-operator/trade-journalism tier source.

The mechanism, in one sentence: an agency that also controls the media-buying pipe can buy inventory at a price the client never sees, and keep the spread. This is structurally the same "hidden markup" logic as any intermediary business — the difference is that an agency is nominally the client's fiduciary-adjacent partner, which is what made the opacity a genuine scandal rather than an accepted cost of doing business.

As of 2026, transparency pressure has materially changed this but not eliminated it. The ANA's own Q1 2025 Programmatic Transparency Benchmark found that only 41% of programmatic ad spend resulted in quality impressions — meaning roughly 60% of spend was lost to non-quality inventory, fees, and intermediary margin somewhere in the supply chain, independent of any single agency's specific behavior. [Established] as the ANA's own reported figure; treat the underlying "quality impression" methodology as the ANA's own definition, not a universal standard every source would compute identically.

What this means for building a new, independent agency in 2026: you are not large enough to run a principal-transaction trading desk, and you shouldn't try to be — but understand that a sophisticated client (the marketing-director buyer from Module 2) has very likely read about this exact scandal and will ask pointed questions about how transparently you bill media. A flat-fee or clearly-disclosed-markup model, with the client's own ad accounts and the client able to see raw platform-level spend directly, is now a genuine competitive selling point against holding-company incumbents rather than a compliance cost — several 2026 sources frame "transparency, governance, strategy" as exactly what independent agencies are positioning against the holding companies with (see Module 4).

#The float mechanism: "playing the bank"

A separate, purely operational payment mechanism worth understanding before Module 3's capital lesson: many agencies front the client's ad spend on the agency's own credit card or credit line, then invoice the client net-15 or net-30. This captures credit-card float and points/cashback for the agency, and lets the agency control account-level billing rather than exposing the client's own card to the platform. The real risk is asymmetric: if the client is slow to pay or defaults, the agency has already paid the platform in full and eats the loss — this is a genuine, underappreciated cash-flow risk specific to agencies that "play the bank" this way, not a theoretical one. [Directional] The alternative — the client's own card or a shared corporate card product on the account — shifts that specific risk back to the client but sacrifices some of the float/points benefit and some of the billing control. Module 4 covers this as an operating risk with its own kill-switch logic.