Operationalizing Incrementality With Northbeam
Why It Matters
Attribution helps you understand where conversions happen, but it doesn’t tell you whether your marketing actually made those conversions happen. That’s where incrementality testing comes in. As we face more signal loss and platforms become less reliable and consistent, relying only on in-platform metrics can lead to wasted spending and misplaced confidence in how well your campaigns are performing.
Incrementality testing, particularly through techniques like geo-holdout testing, gives you a clearer picture of the true impact of your ads—it highlights the revenue that wouldn’t be there without your marketing efforts. With this valuable insight, you can feel more confident about scaling up, cutting back on wasteful spending, trying out new marketing channels, and making smarter budget decisions across all your media.
By 2026, incrementality testing will be crucial; it will be the key to figuring out whether your optimizations are based on solid facts instead of just assumptions. This approach will truly define your business impact.
Why "attribution" isn't the same as "proof" and how geo-holdout testing is changing the way serious marketers measure what's actually working.
Most digital marketers eventually confront a critical question: Is this channel actually driving meaningful results, or are we just allocating budget to customers who would have purchased regardless? While attribution platforms can illustrate clicks, conversions, and overlapping customer journeys, they fail to prove true impact. Incrementality testing fills this gap, and it’s why Northbeam has developed a comprehensive suite of incrementality tools built right into its platform.
This guide analyzes what incrementality really means, how Northbeam’s geo-holdout approach works, and what brands should know before running their first test.
Attribution assigns credit. Incrementality proves causation.
Northbeam helps brands rethink the core measurement question. Multi-touch attribution (MTA) is incredibly useful because it maps the customer journey and shows which channels were used along the way. But just because a channel gets credit doesn’t mean it actually drove the sale.
This difference matters a lot in practice. For example, a brand spending 80% of its budget on Facebook might assume that Facebook should drive 80% of purchases, but that’s almost never what the data shows. Customers may see a Facebook ad, then immediately head to Google to search for your brand and click a paid result. Standard attribution says, “Google converted them.” MTA gets you closer to the truth. But only incrementality testing can answer what would have happened if that Facebook ad had never run in the first place.
For brands with a simple media mix (say, just Meta and Google), rough back-of-envelope math can sometimes suffice. But the moment you introduce a third or fourth channel (YouTube, Connected TV, Apple Search Ads, TikTok), the interactions multiply, and the math gets unworkable. That's when a proper testing framework becomes essential.
How geo-holdout testing works
Northbeam’s approach borrows from how pharmaceutical trials work: split your audience into exposed and unexposed groups, then measure the difference in outcomes. Instead of dividing up individual audiences, they use DMAs (designated market areas) across the US. This technique is called a geo-holdout.
They hold out a certain percentage of your total reach, meaning those regions stop seeing your ads for the duration of the test. Then they compare how the holdout group behaves versus the group still seeing ads. The difference between them is used to calculate incremental cost-per-acquisition (iCPA). In plain terms: How much truly new revenue did your marketing spend create?
The three main testing methods
Geo holdouts (Northbeam's approach): Pauses ads in specific DMAs while running normally elsewhere. Comparable across platforms, repeatable, statistically robust.
Audience holdouts (used by Meta/Google): Withholds ads from a percentage of users platform-wide. Proprietary, hard to compare across channels, limits reproducibility.
Channel blackouts: Fully pause a channel to measure its absence. Option of last resort — not recommended for brands that depend on recurring revenue.
One of Northbeam’s biggest advantages is its MTA data. Since they already track detailed purchase behavior by region, they can build what they call “digital twins”—pairs of DMAs that act similarly. This means they can hold out a much smaller chunk of your audience than platforms such as Facebook (sometimes just a few percent, versus Facebook’s usual 10–15% for lift tests), and finish the test faster too.
Nerdy or not, this stuff is exactly what we’re discussing every day in the Foxwell Founders Membership. Interested in what we’re doing over there? We’d love to have you!
The question nobody asks: then what?
Too often, incrementality tests stop with a single number and a shrug. You ran the test, you got an iCPA, now what? This is the real problem with most testing: results are handed over in isolation, with no obvious next step.
Northbeam flips the script by starting with your objective. Before designing any test, the first question is: What specific business decision should this test help you make?
Baseline performance: How incremental is this channel at all? What's the true iCPA on Facebook, Google, or TikTok?
Strategy validation: Is this bidding approach, creative type, or audience strategy actually driving incremental lift, or is it just finding people who were going to convert anyway?
New channel testing: Before committing real budget to YouTube, CTV, or Apple Search Ads, does a geo-holdout test show meaningful incremental impact?
Halo effect quantification: Is your Facebook spend driving Amazon purchases? Is TikTok moving retail? These cross-channel lift effects are real and often invisible to standard measurement.
Once you have the answer to your key question, you feed that result back into your MTA model, adjusting how credit is assigned every day, based on what you now know about real causal impact. Your media mix model (MMM) also gets smarter with these incrementality inputs, making future budget forecasts more accurate. Think of it as a “triangle offense”: MTA, MMM, and incrementality, all collaborating and enhancing each other better.
The hidden enemy: test integrity
One of the most common and underappreciated problems in incrementality testing is contamination, doing something during the test period that breaks it. Launch a new creative, adjust a bid strategy, forget to apply audience exclusions to a new ad set, and suddenly weeks of work produce an inconclusive result.
Northbeam keeps a close eye on test integrity every day. Their system automatically adjusts targeting, detects budget drift, and flags anything unusual in real time. If a test starts looking inconclusive, Northbeam can step in by either tweaking the parameters and continuing the test or pausing it to take advantage of a performance spike and restarting when conditions are steadier.
Common errors that break tests
Forgetting to apply holdout exclusions to newly launched ad sets mid-flight
Confounding effects of a different, untested channel spikes and distort the baseline
Manual calculation errors in statistical projections (power levels, minimal detectable effects)
Inconclusive results from tests that ran too short or with holdouts that were too small
Which channels are most incremental in 2026?
Based on what Northbeam is seeing across its client base, the incrementality landscape has moved meaningfully in the past 12–18 months.
TikTok has become significantly less incremental for DTC brands than it was a year and a half ago. However, TikTok is increasingly driving Amazon purchases for brands even when their creative doesn't mention Amazon. The halo effect there is substantial and often goes completely unmeasured.
YouTube, particularly YouTube Shorts and Demand Gen campaigns, has improved significantly. As the world’s largest streaming platform, YouTube is increasingly defensible for brands wanting to invest in upper-funnel video, thanks to incrementality data. The measurement challenge has historically been the barrier; a solid geo test changes that conversation with your CFO.
Apple Search Ads tend to perform well incrementally for brands with longer purchase consideration windows. For fast-converting products (hours, not days), the incremental lift is often more modest. It functions more as a mid-funnel channel than as a pure acquisition channel.
Connected TV remains one of the hardest channels to measure with standard tools, making it a strong use case for geo-holdout testing before committing serious budget.
What you need before your first test
If you’re considering running an incrementality test, whether with Northbeam or on your own, the most important thing you need is a plan and the discipline to stick to it. The method gives you guardrails. Everything else is about making sure your test stays within them.
For brands spending over $100K/month on a channel, a platform-level test is well within reach. Below that threshold, tests focused on validating specific strategies rather than the whole channel often make more sense, and volume matters as much as spend. A mattress brand at $100K/month sees far fewer conversions than a pen brand at the same budget, and the minimum detectable effect calculations look very different.
For brands already using Northbeam, the incrementality suite integrates directly with test design, holdout management, daily monitoring, and result interpretation, all happening in one place. But you don't have to be an existing customer to run a test; the value just compounds when it sits alongside your MTA and MMM data.
The bottom line: With signal loss, walled gardens, and platform numbers that can’t be compared across channels, incrementality testing has gone from “nice to have” to a must-have for brands investing in multiple channels. The real question isn’t if you should do it, it’s how to do it in a way that gives you clear decisions, not just more numbers.
and get blogs just like this one sent straight to your inbox.
This content is brought to you in partnership with Northbeam. Read more about our sponsored content policies here: https://www.foxwelldigital.com/sponsored-content-policies