Quick Summary
Summarize this article instantly with your preferred AI model.
How to Run a LinkedIn Ads Incrementality Test
How to Run a LinkedIn Ads Incrementality Test
An incrementality test measures the pipeline your LinkedIn Ads actually caused — the deals that wouldn’t have happened without them — by comparing accounts you advertised to against a holdout group you deliberately didn’t. It answers the question attribution can’t: not “which touchpoints did the ads touch,” but “did the ads make a difference at all.” This matters because attribution, whether last-click or multi-touch, credits ads for conversions that might have happened anyway, so it can over- or under-state real impact. A holdout test is the closest thing to proof. This guide covers how incrementality differs from attribution, how to run a holdout test on LinkedIn, and the discipline it requires to be valid.
Key takeaways
- Incrementality measures the pipeline ads caused — what wouldn’t have happened without them.
- It answers a different question from attribution: “did the ads make a difference,” not “what did they touch.”
- The method is a holdout: advertise to a test group, withhold from a comparable control, compare outcomes.
- The difference between the groups is the incremental lift your ads actually drove.
- Valid tests need comparable groups, enough volume, and enough time — discipline matters.
What is incrementality, and how does it differ from attribution?
Attribution assigns credit for a conversion to the touchpoints that preceded it — LinkedIn gets credit because someone saw or clicked an ad on the path to converting. Incrementality asks whether that conversion would have happened anyway, without the ad. They’re different questions, and they can give very different answers.
The gap matters because attribution can mislead in both directions. It over-credits when it assigns conversions to ads that reached people who would have bought regardless, and it under-credits when it misses influence that didn’t leave a trackable click. Incrementality cuts through this by comparing what happened with ads to what happened without them — measuring cause, not correlation.
| Attribution | Incrementality | |
|---|---|---|
| Question | Which touchpoints preceded the conversion? | Would it have happened without the ads? |
| Measures | Credited influence | Caused lift |
| Method | Tracking touchpoints | Comparing test vs holdout |
| Risk | Over- or under-crediting | Needs disciplined test design |
How do you run a holdout test on LinkedIn?
By splitting a comparable audience into a group that sees your ads and a group that doesn’t, then comparing outcomes:
- Define the population — a set of accounts or a geography that fits your ICP and is large enough to measure.
- Split it into test and control — a test group that will see your ads and a comparable holdout group that won’t. The two must be genuinely similar for the comparison to be fair.
- Run the campaign to the test group for long enough to matter, keeping the control unexposed.
- Compare outcomes — conversion rate, pipeline, or revenue — between the two groups over the same window.
- Measure the lift — the difference between test and control is the incremental impact your ads caused.
If the test group converts or generates pipeline at a higher rate than the comparable control, that gap is what your ads actually drove — not just influenced, but caused.
Why is incrementality worth measuring?
Because it tells you whether your spend is doing anything, which attribution alone can’t. A channel can look great in attribution — credited with lots of conversions — while adding little incremental value, because it was reaching people who’d have converted anyway. Or it can look weak in last-click while driving real incremental pipeline that the tracking missed. Only a holdout reveals which. For an expensive channel like LinkedIn, knowing the incremental truth is the difference between confidently scaling spend that works and pouring money into spend that merely gets credited. It’s the most rigorous answer to “is this actually working,” which is why sophisticated teams use it to validate what attribution suggests.
The incrementality framework
Run a test that produces a trustworthy answer:
- Ensure comparable groups. Test and control must be genuinely similar, or the difference reflects the groups, not the ads.
- Get enough volume. Small groups produce noisy results; you need enough accounts and conversions to detect a real difference.
- Run long enough. Give the test enough time for outcomes — especially pipeline, which lags — to develop.
- Keep the control clean. The holdout must genuinely not see the ads, or the comparison is contaminated.
- Read the lift honestly — the gap between groups is your incremental impact; if there’s no gap, the ads weren’t adding what you thought.
What makes an incrementality test invalid?
Broken comparability or insufficient scale. If the test and control groups differ in ways that affect outcomes — different account sizes, industries, or stages — the difference you measure reflects those differences, not your ads. If the groups are too small or the test too short, random variation swamps any real effect, and you’ll read noise as signal (or miss a real effect). And if the holdout isn’t genuinely unexposed — if control accounts see your ads through another channel or campaign — the comparison is contaminated. The discipline of a valid test is exactly the discipline of any experiment: comparable groups, adequate scale, enough time, and a clean control. Cutting corners on any of these produces a number that looks like proof but isn’t, which is worse than no test at all.
Frequently Asked Questions
Q1. What is an incrementality test for LinkedIn Ads?
An incrementality test measures the pipeline your ads actually caused — deals that wouldn’t have happened without them — by comparing accounts you advertised to against a comparable holdout group you didn’t. It answers whether the ads made a difference, not just which touchpoints they touched, making it the most rigorous way to prove real impact.
Q2. How is incrementality different from attribution?
Attribution credits ads for conversions they preceded; incrementality asks whether those conversions would have happened anyway without the ads. Attribution measures credited influence and can over- or under-state impact; incrementality measures caused lift by comparing a test group to a holdout. They answer different questions and can give very different answers.
Q3. How do you run a holdout test on LinkedIn?
Define a population that fits your ICP, split it into a comparable test group (which sees your ads) and control group (which doesn’t), run the campaign to the test group for long enough, and compare outcomes like conversion and pipeline between the two. The difference in outcomes is the incremental lift your ads actually drove.
Q4. Why measure incrementality instead of just attribution?
Because attribution can’t tell you whether spend is doing anything. A channel can be credited with many conversions while adding little incremental value, because it reached people who’d have converted anyway — or drive real lift that last-click missed. Only a holdout reveals which, which matters most for an expensive channel like LinkedIn.
Q5. What makes a valid incrementality test?
Comparable test and control groups, enough volume to detect a real difference, enough time for outcomes like pipeline to develop, and a genuinely clean control that doesn’t see the ads. If groups differ in ways affecting outcomes, or are too small, or the control is contaminated, the result reflects those flaws rather than your ads’ true impact.
Q6. How big does an incrementality test need to be?
Large enough that random variation doesn’t swamp the effect you’re trying to detect — small groups produce noisy, unreliable results. The exact size depends on how big a lift you expect and your conversion volume; smaller expected effects need bigger samples. Ensure both test and control have enough accounts and conversions to make the comparison statistically meaningful.
Q7. What can invalidate an incrementality test?
Non-comparable groups (differing in size, industry, or stage), insufficient scale (too few accounts or too short a run), or a contaminated control (accounts that see your ads through another channel). Any of these makes the measured difference reflect the flaw rather than your ads, producing a number that looks like proof but isn’t — worse than running no test.
Q8. Can you use incrementality and attribution together?
Yes, and sophisticated teams do. Attribution gives ongoing, granular visibility into which touchpoints influence conversions, while periodic incrementality tests validate whether that influence is real and caused. Use attribution for day-to-day measurement and incrementality to confirm the channel is genuinely driving lift, so you’re not scaling spend that attribution credits but doesn’t actually cause.
More Featured Blogs
View All Blogs
How to Tier Accounts for LinkedIn ABM
How to Use Intent Data With LinkedIn Ads
Lead vs MQL vs SQL From LinkedIn Ads: What's the Difference?