PPC management pricing models, compared.

Every article about PPC pricing concludes that no model is perfectly aligned and you should pick the one that fits your situation. True, and useless — because it gives you no way to compare two imperfect structures. The missing step is that each misalignment has a size. You can calculate what the conflict is worth to the agency and what it can cost you, and once both numbers are on the table the choice becomes an arithmetic comparison rather than a philosophical one.

This is that comparison. For each model: what it pays the agency to do, what the divergence costs when it bites, and the one contract term that neutralizes it. Written to help you choose a structure, not to sell one — the figures throughout are market observation, not our rate card.

Percentage of spend

The fee is a share of media spend, commonly sliding from the mid-teens down as spend grows. Its virtue is simplicity. Its flaw is usually described as "it pays them more when you spend more," which understates the problem, because the interesting case is not a spend increase that works — it is one that does not.

Proposal: increase spend by $10,000/mo
Fee at 12% of spend

Agency gains, regardless of outcome        +$1,200

If that $10,000 returns at 2.5 ROAS (your breakeven):
  your contribution change                       $0
  agency gain                               +$1,200

If it returns at 1.8 ROAS (saturation):
  revenue                                   $18,000
  contribution at 40%                        $7,200
  less the spend                           -$10,000
  your loss                                 -$2,800
  agency gain                               +$1,200

The number to hold onto is the ratio in that last block: the agency captures $1,200 from a decision that costs you $2,800. This is not an accusation of bad faith — most agencies would decline that increase — it is a description of what the structure rewards when the answer is genuinely ambiguous, which at the margin it usually is. And the conflict is worst exactly where scaling decisions are hardest: near the efficiency ceiling, where nobody can be certain whether the next $10,000 lands at 2.5 or 1.8.

The neutralizing term is not a lower percentage. It is an agreed efficiency floor written into the engagement — a marginal ROAS or CPA below which spend does not increase regardless of headroom — so that "should we spend more" has an answer that does not depend on who is asking. Best fit: earlier-stage or genuinely scaling accounts with real headroom, where the ambiguous cases are rare.

Flat retainer

A fixed fee for a defined scope, independent of spend. It removes the spend-more incentive entirely and makes "we should pull back here" a costless recommendation, which is genuinely valuable. It replaces that conflict with a quieter one, and the quieter one is harder to detect: with revenue fixed, the agency's margin improves every hour it does not spend on your account.

Retainer                              $8,000/mo

At 40 hours of real work    -> $200/hour effective
At 20 hours                 -> $400/hour effective
At 10 hours                 -> $800/hour effective

The fee never changes. The service silently halves.

Percentage-of-spend conflicts announce themselves — you can see the budget recommendation. Under-service does not; the account simply gets slower to respond, tests stop happening, and nothing on the invoice changes. Which is why the failure mode of a flat retainer is usually discovered a year late.

The neutralizing term is scope in countable units rather than adjectives: number of net-new creative concepts per month, testing cadence, review frequency, named operator with a stated account load. "Ongoing optimization" is unfalsifiable and therefore unenforceable. Best fit: stable or high-spend accounts where efficiency and honest counsel matter more than raw scaling — provided the scope is genuinely enumerated.

Performance-based

Part or all of the fee is tied to results. It sounds like perfect alignment and is the hardest to implement, for a reason more concrete than "attribution is contested": the invoice now depends on a number that differs by measurement source, and the gap between sources is routinely larger than the fee itself.

Fee: 10% of attributed revenue

Platform-reported revenue           $400,000
  -> invoice                         $40,000

Back-end revenue attributable to paid
(after de-duplicating overlapping
 platform claims and brand demand)  $300,000
  -> invoice                         $30,000

Same month. Same account. $10,000 apart —
25% of the fee, decided entirely by which
number the contract names.

A gap of this size is ordinary rather than exceptional, particularly where Meta and Google both claim the same conversions and neither nets out brand demand. Every month the two sides have a financial interest in a different figure, which turns measurement — the thing you most need a shared, honest view of — into a negotiation.

There is a second cost that does not show up on any invoice: pure performance pricing selects for whatever moves the measured number fastest, and the fastest lever is almost always harvesting existing demand. Brand search, retargeting, and last-click-flattered campaigns all inflate attributed revenue without creating any. The structure pays for the appearance of incrementality.

The neutralizing term is naming the measurement source in the contract — a back-end figure such as blended revenue at a stated MER or qualified pipeline, not a platform dashboard — and keeping the performance element as a component rather than the whole fee, so a measurement dispute is an argument about part of the invoice rather than all of it.

Hybrid

A base retainer plus a smaller percentage or performance component — the most common structure past roughly $100,000 a month, and for a defensible reason: the retainer covers fixed operating cost that does not scale with budget while the variable part keeps some skin in the game. What the arithmetic above adds is that a hybrid does not remove the conflicts, it dilutes them in proportion to the variable share.

On the same $10,000 spend increase that costs
you $2,800 at saturation:

  100% percentage fee at 12%   agency +$1,200
  hybrid, 4% variable          agency   +$400
  100% flat retainer           agency     $0

So the only number that matters in a hybrid quote is the variable share, because that fraction is exactly how much of the spend-more conflict you have retained. A "hybrid" that is 80% variable is a percentage fee with a small retainer attached, whatever it is called. Ask for the split before the headline.

Matching model to situation

• →Scaling hard with real headroom, ambiguous-margin decisions rare — percentage of spend is defensible; add the efficiency floor anyway.
• →Near the efficiency ceiling, where most decisions are marginal — this is where percentage pricing is at its worst; flat or a low-variable hybrid.
• →High spend, efficiency matters, you want costless "spend less" advice — flat retainer, with scope enumerated in countable units.
• →Large account with real fixed and variable work — hybrid, and negotiate the variable share rather than the headline.
• →You want outcome alignment — a performance component inside a hybrid, tied to a back-end metric named in the contract, never the whole fee.

Where this arithmetic stops being reliable

• →The conflict sizes assume the ambiguous case actually arises. On an account with obvious headroom, the percentage-fee conflict is theoretical; near saturation it is the monthly reality. Weight each number by how often you expect the decision to be genuinely close.
• →It models incentives, not people. Most operators decline the bad spend increase and over-service the flat retainer, because reputation over a multi-year relationship outweighs a month of fee. Structure is what you fall back on when judgment is absent — insurance, not a prediction.
• →The effective-hourly framing for flat retainers ignores that experienced operators legitimately need fewer hours. Twenty hours from someone who has run fifty accounts can beat forty from someone who has not, so a rising effective rate is not automatically under-service.
• →The performance-fee gap of 25% is illustrative of a common divergence, not a measured constant; the real figure depends entirely on your channel mix and how much brand demand sits inside the attributed total.
• →None of it prices switching cost. A structurally worse model with an operator who understands your business usually beats a cleaner contract with one who does not — and the transition drag is real money, sized in how much a Google Ads agency costs.

The model sets the incentives; your economics set whether any price is worth paying at all. Size the return math with the breakeven ROAS and ad budget calculators, then run the profit-share and break-even-improvement arithmetic in how much a Google Ads agency costs and the creative-floor derivation in how much a Facebook Ads agency costs. The free audit applies the math to your account.