Pay for Performance Marketing: When It Works and When It Doesn’t

Pay for performance marketing is a commercial arrangement where advertisers pay only when a defined outcome occurs, such as a click, lead, sale, or app install. It sounds like the ideal model: spend money only when something happens. In practice, it is more complicated than that, and the complications are worth understanding before you commit to it.

The model works well in specific contexts. It breaks down in others, often quietly, in ways that take months to surface. What determines which outcome you get is less about the channel and more about how clearly you define performance before the contract is signed.

What Is Pay for Performance Marketing, Actually?

The term covers a range of commercial models. Cost-per-click (CPC) is the most familiar. Cost-per-lead (CPL) and cost-per-acquisition (CPA) go further down the funnel. Revenue share arrangements go further still, tying vendor payment directly to the revenue generated. Affiliate marketing is the oldest and most widespread version of the model.

What they share is a risk transfer mechanism. Instead of paying for exposure or effort, you pay for a result. The vendor absorbs the cost of activity that does not convert. That sounds attractive. The catch is that vendors are rational actors. When they carry the financial risk, they optimise for what gets them paid, not necessarily for what is good for your business long-term.

If you want to go deeper on how paid channels fit together strategically, the paid advertising hub at The Marketing Juice covers the full picture across search, social, programmatic, and performance-based models.

Why the Model Is More Popular Than It Should Be

Performance marketing has grown significantly as digital channels have made conversion tracking more accessible. The logic is seductive: if you can measure outcomes, why pay for anything else? The shift in ad spend toward measurable digital channels has been consistent for two decades, and performance-based pricing has followed that trajectory.

But popularity is not the same as effectiveness. One of the things I observed running agencies is that clients often gravitated toward performance models not because they had thought through the incentive structure, but because it felt like a safer way to spend money. It removed the discomfort of paying for something uncertain. That is understandable, but it is not a strategy.

The deeper issue is that most businesses do not have measurement clean enough to make performance-based models work properly. If your attribution is broken, you are not paying for performance. You are paying for whatever your tracking system credits, which is a different thing entirely.

The Measurement Problem That Undermines Everything

I spent a significant part of my career looking at marketing measurement across dozens of businesses and hundreds of millions in ad spend. The honest conclusion is that most businesses are measuring marketing activity, not marketing impact. Those are not the same thing.

Performance marketing is only as good as the performance definition. If you define performance as a form fill, you will get form fills. Some of them will be real leads. Many will not. If you define performance as a sale, you need strong tracking from click to conversion, and you need to account for returns, chargebacks, and customers who buy once and never come back.

Understanding return on ad spend is central to any performance model. ROAS tells you how much revenue a campaign generates per pound or dollar spent, but it does not tell you about margin, customer lifetime value, or whether the revenue was incremental. A campaign can have strong ROAS and still be destroying value if it is acquiring low-margin customers at high cost.

I have seen this play out with affiliate programmes in particular. A business sets up an affiliate network, drives strong CPA numbers, and celebrates. Then someone looks at the customer data and finds that affiliate-acquired customers have half the lifetime value of customers from other channels, return goods at twice the rate, and cluster in segments that were already converting organically. The CPA looked good. The business impact was neutral at best.

When Pay for Performance Marketing Actually Works

There are contexts where the model is genuinely effective, and they share a few common characteristics.

First, you need a clear, measurable conversion event that is closely tied to revenue. E-commerce with clean purchase tracking is the obvious example. A confirmed, paid transaction is a hard outcome. You can build a CPA model around it with reasonable confidence that you are paying for something real.

Second, you need volume. Performance models work better when there is enough conversion data to optimise against. Campaigns with fewer than a few dozen conversions per month are too noisy to make performance-based pricing reliable. The vendor cannot optimise, and you cannot assess whether the model is working.

Third, the channel needs to be mature enough that the vendor has genuine control over the outcome. Paid search is a reasonable fit because experienced practitioners can meaningfully influence conversion rates through targeting, bidding, and landing page quality. Channels where the outcome depends heavily on factors outside the vendor’s control are a poor fit for performance pricing.

Early in my career at lastminute.com, I ran a paid search campaign for a music festival that generated six figures of revenue within roughly 24 hours. It was a relatively simple campaign by modern standards, but the conditions were right: high-intent search terms, a clean purchase path, and a product people wanted. That is the environment where performance models thrive. The conversion was unambiguous, the tracking was clean, and the channel had a direct line to revenue.

Where the Model Breaks Down

Performance models are a poor fit for brand-building, awareness, or any activity where the outcome is diffuse and long-term. You cannot put a CPA on brand preference. You cannot pay per acquisition for a customer who bought because they had seen your brand consistently for six months across multiple touchpoints.

The tension between brand and performance investment is not new, but performance models tend to accelerate the shift toward short-term, measurable activity at the expense of longer-term brand equity. That is fine if you are managing it deliberately. It becomes a problem when the performance model drives your entire marketing strategy by default.

Performance models also break down when the vendor’s incentives diverge from yours. An affiliate partner paid on CPA has every incentive to drive volume, including volume from low-quality sources. A paid search agency paid on leads has every incentive to generate leads, including leads that never close. These are not hypothetical problems. They are common ones, and they are structural, not a matter of vendor ethics.

When I was running an agency, we had clients who wanted to pay on leads. We had to be honest with them: if we are paid on leads, we will optimise for leads. If you want us to optimise for revenue, you need to share revenue data with us. Most could not or would not do that, which meant the performance model they thought was protecting them was actually misaligning our incentives. The better arrangement, in most cases, was a retainer with clear KPIs and shared access to CRM data.

The Affiliate Channel: A Specific Case Worth Examining

Affiliate marketing is the most mature form of pay for performance, and it has a complicated reputation for good reason. At its best, it is a scalable, low-risk acquisition channel. At its worst, it is a source of low-quality traffic, attribution fraud, and brand risk.

The mechanics of performance marketing in affiliate contexts depend heavily on network quality and publisher selection. A well-managed affiliate programme with vetted publishers and clean attribution can be genuinely efficient. An unmanaged programme on a low-quality network will generate impressive-looking numbers that do not correspond to real business value.

The critical questions for any affiliate arrangement are: who are the publishers, what traffic sources are they using, and how is attribution handled when a customer touches multiple affiliates before converting? Last-click attribution in affiliate marketing is particularly problematic because it rewards the final touchpoint regardless of where the customer’s intent was actually shaped.

Retargeting is a related area worth scrutiny. Retargeting campaigns often perform well on paper because they target people who were already close to converting. The attribution question is whether the retargeting ad caused the conversion or simply captured it. In a pay for performance model, you are paying for the capture either way.

How to Structure a Performance-Based Arrangement That Actually Works

If you are going to use pay for performance models, the structure of the arrangement matters more than the channel or the vendor. A few principles that hold across contexts.

Define performance at the business level, not the channel level. A cost per qualified lead is more useful than a cost per form fill. A cost per retained customer is more useful than a cost per acquisition. The closer the performance metric is to a real business outcome, the more aligned the vendor’s incentives are with yours.

Build in quality gates. If you are paying on leads, define what a qualified lead is before the campaign starts. Agree on a minimum qualification rate. If the vendor delivers leads that are 80% junk, the CPA is irrelevant because you are paying for volume, not outcomes.

Share data. The more visibility a vendor has into what happens after the conversion event they are paid on, the better they can optimise for quality rather than quantity. This requires some trust and some data infrastructure, but it is the difference between a performance model that works and one that generates impressive dashboards with weak commercial results.

Cap your exposure on new arrangements. A performance model with an uncapped budget and an unproven vendor is not risk mitigation. Run pilots with defined spend limits before scaling. The mechanics of PPC and performance channels reward iteration. Start small, validate the quality of outcomes, then scale what works.

The Demand Generation Question

One of the structural limitations of pay for performance models is that they are almost exclusively demand capture mechanisms. They work on people who are already in market, already searching, already close to a decision. They do not create demand.

This matters because most businesses have a finite pool of in-market buyers at any given time. Performance models compete for that pool efficiently. They do not expand it. If your growth strategy depends on reaching people who do not yet know they need what you sell, performance-based pricing will not get you there.

The relationship between demand generation and demand capture is one of the more important strategic questions in marketing, and it is one that performance models tend to obscure rather than illuminate. A business that runs entirely on performance channels is optimising for the last mile of the customer experience while ignoring the first.

I have judged at the Effie Awards, which recognises marketing effectiveness. The campaigns that consistently win are the ones that do both: they build brand and drive conversion, and they measure the contribution of each honestly. Pure performance models rarely produce Effie-level results because they are not designed to. They are designed to harvest the demand that brand activity creates.

The Honest Assessment

Pay for performance marketing is a useful tool. It is not a strategy. The businesses that use it well treat it as one component of a broader acquisition model, with clear measurement, well-defined performance criteria, and genuine visibility into what happens after the conversion event they are paying for.

The businesses that use it poorly treat it as a way to avoid thinking hard about marketing ROI. They pay on metrics that are easy to measure rather than metrics that matter. They let the performance model define their marketing strategy rather than the other way around. And they end up with impressive-looking numbers that do not correspond to commercial outcomes.

If I could give one piece of advice to any marketing leader considering a performance-based arrangement, it would be this: fix your measurement first. If you cannot independently verify that the outcomes you are paying for are real, incremental, and commercially valuable, you are not buying performance. You are buying a number on a dashboard.

For more on how paid channels work across the funnel, from search to social to programmatic, the paid advertising section of The Marketing Juice covers the strategic and tactical dimensions in detail. The performance marketing question does not exist in isolation from those broader channel decisions.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions