Growth Strategy
7 min read

By Michael Minter
Every ad platform reports its own results, and every platform takes credit for as many sales as it can. Meta counts a purchase if someone saw an ad. Google counts it if someone clicked a brand search ad on the way to your site. Add the platform numbers together and the total is usually larger than your actual revenue.
Incrementality testing answers a narrower question: how many of these sales happened because of the ads? This guide covers what incrementality testing is, the three common ways to run a test, how to read the results and where it fits with the numbers you already track.
What is incrementality testing?
Incrementality testing compares a group that sees your ads with a similar group that does not. The difference in conversions between the two groups is the lift your ads caused. Everything else is demand you would have captured anyway, through organic search, word of mouth, email or people who already knew your brand.
The idea comes from controlled experiments. You hold out a portion of your audience or your market, keep everything else the same, and measure the gap. Attribution models try to assign credit after the fact. A lift test measures it directly.
Why platform ROAS overstates results
Platform-reported ROAS tends to run high. Three things push it up:
Brand search. People who type your brand name into Google were already looking for you. A brand campaign often catches sales that would have arrived through the organic listing one line below.
Retargeting. Showing ads to people who already added to cart reaches buyers who were close to purchasing anyway.
View-through credit. Platforms can count a purchase made a day after someone scrolled past an ad without clicking it.
Those campaigns can still earn their place. Some brand search defends against competitors bidding on your name, and some retargeting does change behavior. You only know how much once you test it.
This is why we measure accounts on blended CAC, MER and contribution margin rather than platform ROAS alone. Those numbers come from your own revenue and spend, so no platform can inflate them. Incrementality tests then tell you which channel is driving the blended number. If you want help setting this up across channels, our paid media team runs it as part of account management.
Three ways to run an incrementality test
Geo holdout tests
A geo holdout splits your market by location. You pick a set of regions, such as states or metro areas, and turn ads off (or down) in some of them while keeping them on in comparable regions. After a few weeks, you compare sales in the two groups.
Geo tests work across any channel, including channels without built-in lift tools, such as connected TV. They measure total sales from your own store data, so they capture effects that platform tracking misses, like someone who saw an ad and later bought on Amazon or in a retail store. The tradeoff is that they need enough volume in each region to show a clear difference, and regions are never perfectly matched.
Platform conversion lift studies
Meta and Google both offer lift studies for eligible accounts. The platform randomly splits users into a test group that can see your ads and a control group that cannot, then reports the difference in conversions. Because the split is random at the user level, the groups are well matched.
The limits: the study only covers one platform at a time, it relies on the platform’s own conversion tracking, and eligibility depends on spend and account setup. It is still a strong check on whether a campaign drives sales or only collects credit for them.
On/off and budget-step tests
The simplest test is turning a campaign off for a period and watching total revenue. A gentler version steps budget down by a fixed share and tracks what happens to new customer orders and MER.
These tests are cheap and fast, but they are the easiest to fool. A holiday, a promotion, a stock-out or a viral post can move sales during the same window. Use them for quick reads on smaller campaigns, and use geo or platform lift tests for decisions about large budgets.
How to set up a geo holdout test
Pick one question. “Does our brand search campaign add sales?” is testable. “Is our marketing working?” is not.
Choose matched regions. Group regions with similar past sales trends, then randomly assign some to the holdout. Check that the two groups moved together before the test.
Decide the change. Turn the channel off in the holdout regions or cut its spend by a set amount. Leave every other channel alone.
Set the length before you start. Most tests need at least two to four weeks, longer if your buying cycle is long. Do not stop early because the first week looks good or bad.
Measure business outcomes. Use total orders, new customers and revenue from your store or CRM, not platform-reported conversions.
Freeze other changes. New promotions, price changes or big creative launches during the test make the result hard to read.
How to read the results
The output of a lift test is the extra conversions the test group produced compared with the control group. From that you can work out two numbers:
Incremental CPA: spend in the test regions divided by the extra conversions.
Incremental ROAS: extra revenue divided by spend.
Compare these with what the platform reported for the same campaign over the same period. If the platform claimed 500 purchases and the test shows 200 extra purchases, about 40% of the claimed results were incremental. You can use that ratio to discount the platform’s reporting until the next test.
Treat a single test as one data point. Results change as you scale spend, change creative or enter new markets. Retest your largest channels once or twice a year and after any major change.
Where incrementality fits with MER and blended CAC
Incrementality tests are slow and expensive to run constantly, so they work best alongside day-to-day metrics:
MER (total revenue divided by total ad spend) tells you whether the whole program is healthy.
Blended CAC (total spend divided by new customers) tells you what growth costs.
Contribution margin tells you whether that growth makes money after product, shipping and ad costs.
Lift tests tell you which channels deserve the credit, so you know where the next dollar should go.
When a channel’s platform ROAS looks strong but MER does not move as you scale it, that is the signal to test it.
An example: Owl’s Brew
Owl’s Brew, a beverage brand selling online and in retail, came to us with no clear way to measure lift from paid media and overlapping audiences across Search and Display. Part of our work was a lift measurement model: we put spend into specific geographies and tracked incremental performance there. That showed which regions were profitable to scale and helped the brand support retail growth as well as online sales. Alongside a rebuilt creative testing framework, CTR rose 800% and CPC fell 77%.
Geo testing suited Owl’s Brew because a lot of its sales happened in stores, where platform tracking cannot follow the customer. The same approach works for any ecommerce brand that sells through more than one channel.
Common incrementality testing mistakes
Testing too small. If the test regions only produce a handful of orders a week, random noise will swamp the result.
Changing several things at once. If a test also includes a new offer, the offer’s effect gets mixed into the channel’s result.
Reading platform conversions as the outcome. The point is to step outside platform tracking.
Ending the test early. Short tests miss delayed purchases and over-react to one strong or weak week.
Treating the result as permanent. Lift changes with spend level, creative and season.
Incrementality testing FAQ
What is the difference between attribution and incrementality?
Attribution assigns credit for a sale to the touchpoints before it. Incrementality measures whether the sale would have happened without the ads. Attribution answers “who touched this customer”, and incrementality answers “did the ads cause this sale”.
How much budget do I need for an incrementality test?
There is no fixed minimum. You need enough conversions in both the test and control groups to see a difference larger than normal week-to-week swings. Brands with very low order volume usually get more from MER tracking and on/off tests than from formal lift studies.
How long should a lift test run?
Usually two to four weeks, and longer for products with a long consideration period. Set the end date before you start.
Which channel should I test first?
Start with the channel where platform-reported results and business results disagree the most. For many brands that is brand search or retargeting.

want more?
read more blogs
We don’t just create campagins-we build success stories that lasts.

Landing Page Best Practices for Paid Traffic: What We Test First
Most paid traffic lands on pages that were never built for it. These are the landing page best practices we check first in every account, from message match and page speed to forms, pricing and intake flows.

Incrementality Testing: How to Measure the Sales Your Ads Cause
Ad platforms count conversions that would have happened anyway. Incrementality testing measures the sales your ads cause. Here is how it works, which tests to run and how to read the results.

Most Marketing Agencies Suck. Here's Why We Don't.
If you’ve worked with a marketing agency before, odds are it ended in frustration. Overpromising. Under-delivering. A few nice graphics and a fluffy strategy doc — and suddenly you’re locked into a contract wondering where your ROI went. We’ve heard it all before. And at Drip, we’ve built our business by being everything those agencies aren’t.

Why Your Ads Don’t Work (Hint: It’s Your Website, Not Your Budget)
You might be seeing clicks, but if your site can’t convert — that budget is going straight down the drain. That’s why at Drip, Conversion Rate Optimization (CRO) is built into every growth plan. If you want better performance, it starts with the experience after the click.

The Secret to High-Converting Creative? It’s Not Just Pretty Ads.
Creative is the #1 lever in digital advertising — and also the most misunderstood. Most brands think it’s about flashy edits, moody product shots, or trendy audio. That might get attention, but it won’t drive conversion.

How to Scale Paid Ads Across Meta, Google, and CTV Without Killing Your ROAS
Scaling ad spend is the goal — but doing it wrong can crush your margins. Most brands rush into higher budgets without fixing the leaks in their system. That’s why scaling isn’t just about adding money — it’s about scaling smarter.
Ready to elevate your business?
Contact us today to discuss how we can help you achieve your marketing goals.

© 2026 DRIP Digital Media. All rights reserved.
