Most agency contracts are built on a quiet lie: that effort equals value. You pay a flat monthly retainer, the agency staffs your account, and everyone agrees to pretend that activity—reports, meetings, "optimizations"—is the same thing as results. When growth stalls, the retainer keeps clearing. That's not a partnership. That's a subscription you forgot to cancel.
Performance-based models promise to fix this by tying what an agency earns to what it actually delivers. Done right, they align everyone around the same number. Done wrong, they create perverse incentives that quietly damage your brand. The difference comes down to structure, trust, and whether both sides actually understand the economics they're signing up for.
Let's get specific about when these models work, when they blow up, and how to build one that doesn't.
Why the Flat Retainer Breaks Down
The standard retainer exists for a reason: it's predictable, easy to budget, and it pays the agency whether the market is up or down. For a lot of relationships, that stability is a feature, not a bug.
But it has a structural flaw. The agency's revenue is decoupled from your results. Once the contract is signed, the economic pressure to perform drops sharply. The agency now has an incentive to protect the relationship (keep you happy enough to re-sign) rather than to maximize your outcomes (which might require risk, experimentation, or telling you something you don't want to hear).
You see this show up in predictable ways:
- Effort theater. Decks get longer. Meetings multiply. The volume of "work" expands to justify the fee, even when the work that moves the needle is 20% of the activity.
- Risk aversion. A flat-fee agency has little upside from a bold test that doubles your results, but plenty of downside if it fails and you churn. So they play it safe.
- Scope creep in reverse. As your account grows, a fixed retainer means the agency's margin improves while your cost-per-result stays flat. Your success subsidizes their comfort.
None of this makes retainer agencies bad actors. It's just what the incentive structure rewards. And incentives, over a long enough timeline, beat intentions every time.
What "Performance-Based" Actually Means
"Performance-based" is a vague term that hides very different arrangements. Before you evaluate any agency model, get precise about which flavor you're discussing.
Commission on spend. The agency takes a percentage of ad spend (commonly 10–15%). This is technically performance-adjacent but structurally broken: it rewards the agency for spending more, not for spending well. Avoid it unless paired with hard efficiency guardrails.
Pay-per-outcome. The agency earns a fixed fee per defined action—a qualified lead, a booked demo, a completed sale. Clean and measurable, but only works when the outcome is tightly defined and attribution is reliable.
Revenue share. The agency earns a percentage of the revenue it generates. This is the model people usually mean when they talk about "aligned incentives," because the agency only wins when you win. It's also the hardest to structure fairly, because you have to answer a deceptively difficult question: which revenue is the agency responsible for?
Hybrid (base + performance). A reduced retainer covers the agency's fixed costs and keeps the lights on, plus a performance component—often a revenue share or bonus tied to targets—that rewards overperformance. In practice, this is the structure that works most often, and we'll come back to why.
The mistake founders make is treating these as interchangeable. A commission-on-spend deal and a true revenue share create nearly opposite incentives. Name the structure precisely before you negotiate it.
The Conditions That Make Revenue Share Work
A pure or heavy performance model isn't universally better—it's situationally better. It thrives under specific conditions and quietly fails without them. Before you push an agency toward revenue share, pressure-test your situation against these five requirements.
1. Clean, agreed-upon attribution. If you can't both look at the same dashboard and agree on what the agency drove, revenue share will poison the relationship within 90 days. You'll argue about whether that sale came from the paid campaign or the brand demand that was already there. Pick your source of truth—platform-reported, a measured incremental lift, or a last-touch model you both accept—and write it into the contract.
2. A product that already converts. Performance models reward the agency for finding and routing demand. They don't fix a broken offer, a leaky checkout, or a product nobody wants. If your conversion rate is in the gutter, a smart agency won't take the deal—and if they do, they're betting they can carry your product's weight, which rarely ends well. Get your fundamentals to a baseline before you ask anyone to bet on upside.
3. Enough margin to share. Revenue share has to come out of somewhere. A brand running on thin margins can't hand over 10–15% of top-line revenue without bleeding. Run the math: if your blended margin is 30% and you give away 10% of revenue, you've just surrendered a third of your profit. That might still be worth it if the agency grows the pie—but you need to know the number cold before you sign.
4. A time horizon long enough to compound. Performance models punish short engagements. The first 60–90 days are usually investment—audits, creative, testing—where the agency spends more than it earns. The payoff comes later. A three-month trial on a revenue share is a recipe for mutual resentment. Both sides need the runway for the structure to pay off.
5. Mutual trust and transparency. This is the one nobody writes into the contract and everybody needs. Revenue share means the agency sees your numbers and you see their logic. If either side is guarding information, the model turns adversarial fast.
Here's a simple test: if you can't define the result, can't measure it cleanly, and can't fund it from margin, you're not ready for a performance model. Fix those first.
The Traps on Both Sides
Aligned incentives sound great in the abstract. In practice, a poorly designed performance agency model creates new problems that a retainer never had.
Traps for the brand:
- Short-termism. An agency paid on this quarter's revenue will optimize for this quarter's revenue—even at the cost of brand equity, customer lifetime value, or next year's pipeline. Aggressive discounting, fire-sale promotions, and bottom-funnel-only spending all juice short-term numbers while quietly eroding the business. If your performance metric is purely immediate revenue, expect your brand to be mortgaged for it.
- Attribution gaming. Where the agency's pay depends on attribution, expect them to claim credit aggressively. Branded search that would have converted anyway, retargeting of people already in your funnel, email flows to customers you already owned—all of it can get swept into "agency-driven revenue" if the rules are loose.
- The free-rider problem. If your brand has strong organic demand, a revenue share pays the agency for momentum they didn't create. You end up overpaying for the easy wins.
Traps for the agency:
- Carrying the brand's weak points. The agency controls traffic and messaging but usually doesn't control the product, pricing, site experience, or fulfillment. When conversion tanks because of something outside their lane, they eat the loss anyway.
- Volatility. A client who pauses spend, has a supply problem, or hits a seasonal trough can wipe out the agency's income with no warning. Agencies with thin cash reserves can't absorb that.
- Being the first to invest, last to get paid. The upfront work is real and front-loaded. If a client churns after the investment phase but before the payoff, the agency loses twice.
Good contracts anticipate these. Caps and floors, a baseline-revenue carve-out (so the agency is only paid on incremental growth above an agreed starting point), clear attribution windows, and guardrail metrics like minimum return thresholds or brand-health checkpoints all exist to keep the structure honest.
A Framework for Structuring the Deal
If you've concluded a performance model fits, here's a practical way to build one that protects both sides. Think of it as four dials you set together.
Dial 1: The base. Decide how much fixed fee covers the agency's core costs. A pure performance deal with zero base pushes all the risk onto the agency and tends to make them conservative or desperate—neither is what you want. A modest base (enough to staff your account without panic) plus a performance kicker usually produces the best behavior. A common shape is a reduced retainer, roughly half of what a flat-fee version would be, plus upside.
Dial 2: The baseline. This is the most important and most skipped step. Agree on the revenue your business generates without the agency's efforts—your trailing run-rate before they start. The performance component should only apply to revenue above that baseline. This neutralizes the free-rider problem and makes "incremental" the operative word. Revisit the baseline on a set schedule (quarterly or semi-annually) so the agency doesn't get paid forever on growth that's become your new normal.
Dial 3: The rate and the cap. Set the revenue share percentage on incremental revenue, and consider a cap so a breakout month doesn't create an absurd invoice that makes you resent the deal. A cap feels like it limits the agency's upside, but it protects the long-term relationship—and a deal that survives is worth more than one that maximizes a single quarter.
Dial 4: The guardrails. Define the non-revenue constraints that keep short-termism in check. Minimum efficiency thresholds (e.g., don't buy revenue below a certain return), discount limits, and brand-safety standards. Add quality metrics where relevant—if you pay per lead, pay on qualified leads, with a clear definition and a clawback for junk.
To make this concrete, consider a hypothetical: a brand doing roughly $200k/month in revenue brings in an agency. They set the baseline at $200k, agree on a 10% share of incremental revenue above it, pair it with a reduced base retainer, and cap the monthly performance fee. If the agency grows the brand to $300k, they earn 10% of the $100k in incremental growth plus the base—and the brand happily pays it because every dollar of that fee came from a dollar it wouldn't otherwise have had. That's what alignment looks like when the math is clean. (Numbers here are illustrative, not a promise of results.)
How to Decide What's Right for Your Brand
Not every brand should chase a performance model, and not every agency can offer one responsibly. Use these questions to make the call.
- Can we measure the outcome cleanly, and do we both agree on how? If no, start with a retainer and build measurement first.
- Is our offer already converting at a healthy rate? If no, fix conversion before you share revenue.
- Do our margins have room to share without breaking the business? Run the actual number.
- Are we committing for long enough to let the structure compound? If you're testing an agency, use a short retainer trial first, then convert to performance once trust and data exist.
- Do we trust this agency enough to open our numbers? If not, that's your answer about the relationship, not just the model.
A reasonable default for a new relationship: start on a hybrid, with a modest base and a performance component tied to incremental revenue above a defined baseline. It lets both sides prove themselves before raising the stakes, and it hardwires alignment from day one without betting the whole relationship on it.
Your Next Steps
The point of a performance model isn't to pay less. It's to make sure that when you pay more, it's because you're making more. Alignment is the product.
If you're evaluating your current agency relationship:
- Pull your last six months of agency invoices and set them against results. If you can't draw a clear line from fees to outcomes, your incentives aren't aligned—regardless of what the contract calls itself.
- Define your source of truth for attribution before any performance conversation. Pick the dashboard, pick the model, and get both sides to sign off. This single step prevents most performance-model disputes.
- Calculate your true margin and your current revenue baseline. You can't negotiate a fair share of revenue you can't measure or afford.
- Audit your conversion fundamentals. If your offer and funnel aren't converting, no incentive structure will save you—fix that first.