How to calculate incrementality and why it matters

Most marketing campaigns can show results. The harder question is whether those results happened because of the campaign.

Think about a gym owner who hires a new trainer in January and watches memberships climb 20% over the next quarter. The trainer looks like a win on paper, but January is resolution season. New members were likely coming in anyway. Without a way to compare outcomes with and without the trainer, there's no way to separate their contribution from the seasonal trend.

That's the problem incrementality in marketing is designed to solve. Rather than asking whether a marketing campaign drove results, incrementality testing asks how many of those results would have happened regardless. The difference between those two numbers is the incremental impact of your marketing activity, and it's one of the most useful figures a marketing team can work with because it impacts budget allocation, your ability to understand which marketing channels are actually pulling their weight, and whether you can avoid scaling ad spend behind results that would have happened with or without your campaign.

Key takeaways

• Incrementality testing compares a test group (or treatment group) exposed to a marketing campaign against a control group that isn't in order to isolate the results the campaign actually drove.
• Incremental lift measures how much a campaign improved conversion rate above the baseline; the incrementality percentage measures what share of total conversions were campaign-generated.
• iROAS and iCPA are the most practical profitability metrics for evaluating whether incremental results justify the ad spend.
• What counts as a "good" result varies by channel, campaign objective, and brand maturity; there's no universal benchmark.
• External factors like seasonality, regional differences, and test duration can distort your calculations.
• Incrementality testing helps you evaluate specific ad campaigns, but it captures a moment in time rather than an ongoing picture of your full marketing efforts.
• Marketing mix modeling (MMM) offers a complementary approach that measures incremental impact across all marketing activity continuously, without requiring you to pause ad campaigns to run controlled experiments.

What incrementality actually measures

Incrementality testing works by splitting your audience into two groups. The test group (also called the treatment group) sees the marketing campaign. The control group doesn't. By comparing how each group behaves during the test window, you can estimate how many conversions, purchases, or other business outcomes happened specifically because of the campaign, versus how many would have occurred through organic growth, direct intent, or other existing marketing touchpoints. This only works if the two groups are genuinely comparable: if they differ in ways beyond media exposure, you're measuring the difference between two dissimilar audiences, not the impact of a specific marketing campaign.

The core insight is that not every conversion attributed to a campaign is actually caused by it. A customer who was already planning to buy might click a retargeting ad right before checkout. Platform-reported conversions count that as a win, but would they have bought without the ad? Probably. Incrementality measurement is how you find out.

This makes incrementality testing fundamentally different from other measurement approaches. Multi-touch attribution (MTA), for example, assigns credit to each marketing touchpoint along the customer journey based on rules or statistical models. But multi-touch attribution tells you which touchpoints were present, not which ones actually changed behavior. Crediting every marketing touchpoint in a path doesn't tell you which one was doing the impactful work. Incrementality testing gets closer to the latter by comparing exposed and unexposed audiences directly within a defined test period.

The core incrementality formulas

There are two primary calculations used in incrementality measurement, and they answer slightly different questions. Both are worth understanding before you run the numbers.

Incremental lift measures how much your campaign or marketing activity improved performance relative to the baseline set by your control group.

Incremental lift (%) = ((Test conversion rate − Control conversion rate) / Control conversion rate) × 100

If your test group converts at 15% and your control group converts at 10%, your incremental lift is 50%. The campaign improved conversion rate by 50% relative to what would have happened without it.

Incrementality percentage measures what share of your test group's total conversions were actually campaign-generated, rather than conversions that would have happened anyway.

Incrementality (%) = ((Test conversion rate − Control conversion rate) / Test conversion rate) × 100

Using those same rates: (15% − 10%) / 15% × 100 = 33%. One-third of your test group's conversions were driven by the campaign while the other two-thirds would have converted without it.

Both formulas use the same inputs, but the choice of denominator changes the meaning:

• Incremental lift tells you how much better you performed with the campaign.
• Incrementality percentage tells you how much of your reported performance is real.

To calculate incremental lift or the incrementality percentage accurately, you need test and control groups that are comparable in size and composition, and a test window long enough to capture meaningful behavior.

How to evaluate the profitability of an incremental result

Positive incremental lift is encouraging, but it doesn't tell you whether a campaign is worth running. A campaign can move the needle and still be unprofitable if the cost of producing that lift is too high. These three metrics help marketing teams evaluate whether the results justify the ad spend:

Incremental revenue

Incremental revenue is the starting point for any profitability calculation.

Incremental revenue = Total revenue (test group) − Total revenue (control group, scaled for size)

This gives you the dollar value attributable to the campaign above what would have happened without it. Use this figure, not total reported revenue, as the input for the profitability metrics below.

iROAS (incremental return on ad spend)

iROAS = Incremental revenue / Campaign cost

iROAS applies the familiar ROAS framework to your actual incremental results. A campaign with a 4x reported ROAS might land at 1.5x iROAS once you strip out conversions that would have happened anyway. That gap matters significantly when you're making budget allocation decisions across marketing channels.

iCPA (incremental cost per acquisition)

iCPA = Campaign cost / Incremental conversions

iCPA tells you what you actually paid per net new customer the campaign drove. Paired with average order value or customer lifetime value, it lets you evaluate whether the economics of acquiring that customer through this specific marketing campaign hold up.

Together, these three metrics give you a clear read on whether your marketing efforts generated real value, or whether that marketing spend would produce better returns elsewhere.

What a "good" result looks like

After running their first incrementality test, most marketers want to know if their numbers are good. The honest answer is the thing everyone hates to hear: it depends. No single benchmark for incremental impact applies across all channels, objectives, or business types.

A few useful frames for evaluating your results in context:

• Brand vs. performance ad campaigns: Upper-funnel campaigns are designed to influence purchase behavior over time, not drive immediate conversions. A/B testing a brand awareness campaign against a direct response campaign will almost always show different incrementality levels, and that should be expected.
• Retargeting vs. prospecting: Retargeting campaigns target people already close to converting, so incremental lift tends to be lower. Prospecting campaigns reaching new customers in a new target audience often show higher lift, usually at a higher iCPA.
• Familiar channels vs. new ones: Marketing channels your target audience already associates with your brand may show lower incremental impact than channels where you're reaching unexposed audiences for the first time.
• Competitive environment: In crowded categories, any single marketing campaign may show lower lift because customers have more alternatives. Less competitive markets often show higher lift from the same marketing activity.
• A/B testing design: How you set up your test and control groups matters. A/B testing with poorly matched groups produces results that are hard to interpret regardless of what the numbers say.

The most reliable benchmark is usually an internal one: how does this campaign's incrementality compare to previous campaigns on the same channel, or to other marketing tactics competing for the same budget? If you're using media mix modeling alongside your incrementality tests, you can also benchmark each marketing touchpoint's incremental contribution across channels over time, which gives you a more complete picture than any single test.

What can throw off your calculation

Even well-designed controlled experiments can produce unreliable results if certain conditions aren't accounted for. Here are the most common sources of distortion when you test and control for marketing impact.

Imperfect control groups

The validity of any incrementality calculation depends on how comparable your two groups are. If the test and control groups differ in meaningful ways beyond the marketing exposure (demographic skew, purchase history, geography), then the gap in their conversion rates may reflect those underlying differences rather than your campaign's actual incremental impact.

Geo-testing is a common framework for running incrementality tests at scale, but geography introduces complications. No two regions are truly equivalent. Local economic conditions, seasonal patterns, and competitor activity vary in ways that are difficult to control for, which means differences between test and control results may reflect regional variation as much as your campaign's actual effect. This is one of the more persistent challenges in geo-level incrementality testing, and it should be factored into how you interpret the results.

External factors during the test window

If something significant happens mid-test (a competitor promotion, a supply disruption, an unexpected news event), it may affect one group more than the other and distort results. Incrementality testing assumes media exposure is the only meaningful difference between your groups. External factors make that assumption harder to hold.

Tests that are too short

Most incrementality tests run for a few weeks. The test period is often sufficient for measuring direct response behavior, but that window can miss the full effects of upper-funnel marketing activity. Some ad campaigns build awareness that converts into purchases weeks or months later, well beyond a typical test window. When measuring incremental lift for brand campaigns, plan for a longer test period than you would for direct response work. This is one area where media mix modeling has a practical advantage: because it measures ongoing marketing activity rather than a fixed window, it captures effects that a short incrementality test would miss.

Audience contamination

In geo-based tests, people move between regions. A customer who sees your ad in a test market and purchases while in a control area may not be captured correctly in either group's sales data. This is generally a small effect, but it adds noise to your calculation.

Underpowered tests

Running a test on too small an audience segment produces results with wide confidence intervals. Observed incremental lift may not be statistically reliable if the test wasn't designed with sufficient scale. This is one of the more common reasons a test appears to confirm a result that the data can't actually support.

None of these limitations mean you shouldn't measure incrementality. They mean the results deserve scrutiny before they drive major marketing spend decisions.

What to do with your results

Calculating incrementality is the easy part. Acting on the number requires more judgment. Once you measure incrementality for a campaign, you need to weigh the result against cost, context, and what else is competing for the same budget. The table below offers a starting framework based on common test and control outcomes, but we suggest you make sure you also understand the halo effects of your campaigns so you understand the full context of your marketing performance before cutting anything.

Beyond this framework, a few principles worth keeping in mind as you translate results into marketing strategy:

• One test is a data point, not a verdict. A single incrementality test captures one moment in time under specific conditions. Treat results as directional and validate them against other data sources before making permanent channel decisions.
• Results are channel-specific. A strong result for a paid social campaign doesn't mean you'd see the same incremental impact from online display or paid search. Each marketing channel and each set of marketing tactics deserves its own evaluation. Budget allocation decisions should be based on iROAS comparisons across channels, not on reported ROAS alone.
• Think beyond direct response. Most incrementality tests measure conversions within a defined window. Campaigns that shape brand consideration or influence the broader customer journey may not show their full contribution within a short test. Marketing mix modeling is better suited to capture those longer-horizon effects, since it measures marketing activity continuously rather than within a fixed window.
• Compare marketing tactics fairly. An incrementality test compares one tactic against no exposure at all. If you're comparing marketing tactics against each other, you may need separate tests for each one rather than drawing conclusions from a single test-and-control setup.
• Context matters. An incrementality test that runs during a seasonal peak or alongside a major promotion may not reflect normal performance conditions. Factor in what was happening in the market when you interpret the data.

Where Prescient comes in

Incrementality testing is a useful tool for evaluating specific ad campaigns, but it's a point-in-time method. Each test covers a defined window for a specific audience segment against a specific control group. What it can't provide is a continuous view of incremental impact across your full media mix, and running tests often requires suppressing spend to create a clean control group, which has its own cost. It also doesn't replace the broader picture you need to make confident, ongoing marketing strategy decisions.

Prescient's marketing mix modeling platform measures the incremental impact of each of your marketing channels and campaigns on an ongoing basis, with daily model updates that account for external factors like seasonality, organic growth, and cross-channel interactions. Unlike multi-touch attribution, which follows each marketing touchpoint as platforms attribute it, media mix modeling uses an external model to determine what your marketing efforts actually drove, independent of what any platform claims. For brands also running incrementality tests, Prescient's Validation Layer runs two parallel models simultaneously: one that includes your test data, and one that doesn't. Both receive accuracy scores, so you can see directly whether your incrementality data is improving or degrading your measurement. See how the platform surfaces all of this and more when you book a demo.

FAQs

What is incrementality?

Incrementality in marketing refers to the additional results that a specific marketing campaign directly produces above and beyond what would have happened without it. It's the difference between performance your campaign caused and performance that would have occurred anyway through organic growth, direct traffic, or other marketing touchpoints already at work. Incrementality measurement isolates the genuine contribution of a campaign from the baseline activity that was happening regardless, which is something multi-touch attribution and standard platform attribution alone can't do.

What is an example of incrementality?

A practical example: a brand runs a paid social campaign targeting a defined audience segment and compares the conversion rate of that test group against a matched control group that wasn't shown the ads. If the test group converts at 12% and the control group converts at 8%, the campaign's incremental lift is 50%, and roughly one-third of the test group's conversions are attributable to the campaign. The remaining conversions would have happened without the ad. This is what incrementality testing reveals that standard attribution doesn't: not just how many people converted, but how many converted because of the campaign.

What is the formula for incremental profitability?

The most common metric for evaluating incremental profitability is iROAS: iROAS = Incremental revenue / Campaign cost. A related calculation is incremental profit itself: Incremental revenue − Campaign cost. Both require starting from incremental revenue rather than total reported revenue. You calculate incremental revenue by comparing the total revenue generated by the test group against the total revenue generated by the control group (scaled for group size), giving you the net revenue the campaign actually drove above the baseline.

What is the formula for incremental conversion?

There are two versions of the incrementality calculation formula, and they answer different questions. Incremental lift (%) = ((test rate − control rate) / control rate) × 100, where "test rate" and "control rate" refer to each group's conversion rate. This measures how much the campaign improved conversion rate relative to the baseline. Incrementality (%) = ((test rate − control rate) / test rate) × 100. This measures what share of your test group's total conversions were actually campaign-generated, as opposed to conversions that would have occurred without the campaign running.