The Google Ads dashboard shows you plenty of conversions, but not all of them are down to Google. A good share can be sales that were going to happen anyway: people who were already searching for your brand, or demand that your social media advertising actually created and that Google claims when the user finishes the purchase by searching for you. The result is a Google ROAS that looks great and may be taking credit for sales it didn't generate.
When you're deciding how much to invest, the number that matters is how many new customers Google genuinely brings you, the ones you wouldn't have had without that spend, rather than how many conversions the platform reports. That's called incrementality, and the only way to know it is to measure it with tests instead of trusting the platform's own attribution.
In this article you'll find six ways to check whether your Google Ads are incremental, from the simplest to the most rigorous, plus a final summary of which metric answers each business question.
Why can Google Ads take credit for sales you already had?
There are three ways Google takes credit it doesn't fully deserve. The first is brand searches, when someone who already knows you types your name into Google and clicks your ad instead of the organic result sitting right below it. That sale would have happened anyway, but Google counts it as its own and has charged you for the click on top.
The second is demand created by other channels. Someone sees your ad on Meta or TikTok, gets interested, and later searches for you on Google to buy. Paid social lit the spark, but because the last click was on Google, Google gets the conversion. The third is Performance Max, which, by mixing all the inventory together, tends to pick up that brand and organic demand and present it as its own result. In all three cases the pattern is the same: Google measures what happens inside Google very well, but it can't see what happens across the rest of your channels, so it has no way of knowing what would have converted without it.
Raise your brand spend and read the total clicks
The first test is the most accessible, and it shows you the cannibalisation between what you pay for and what you already had for free. Before you change anything, pull your paid brand clicks and your organic clicks into a single report so you can see them as one total. Then triple your brand budget and watch what happens to total clicks over about six weeks, ignoring the CPA the platform reports.
The usual outcome is an unpleasant surprise. In one test we ran, we tripled brand spend and gained around 2,500 extra paid clicks, but at the same time lost about 5,000 organic ones, because people were clicking the ad instead of the free result sitting right below it. Put another way, we paid for traffic that was already ours and ended up with fewer clicks overall. Also keep an eye on when your brand searches spike, because if that peak lines up with your Meta spend, paid social created that demand, not Google.
Geo holdout: switch the channel off in half your markets
The geo holdout is one of the most reliable tests, because it measures against real sales and doesn't depend on what the platform's attribution says. The idea is simple. You split your markets or regions into two halves that resemble each other, leave the channel on in one and switch it off completely in the other. After a few weeks you compare total sales across the two groups, and that difference is what the channel really contributes: the sales you wouldn't have had without it.
Setting it up takes more work than the other tests, but it has an advantage that's hard to match: there's no way for the platform to inflate the result, because you're comparing one group's real sales against the other's.
Switch the channel on and off week by week
If you can't split by geography, the alternative is an on/off pulse. You alternate the channel, one week on and the next off, for at least six weeks, then compare total conversions in the on weeks against the off weeks. For the contrast to read clearly, it helps to spend twice as much in the active weeks, so the swing is obvious.
The most famous example of this idea came from eBay in 2014. It ran the extreme version of the test, switched its brand ads off entirely, and found that its traffic held almost steady, because anyone searching for eBay arrived through the organic result anyway. It's the classic demonstration that a large part of brand spend may not be incremental.
Exclude your brand in Performance Max
Performance Max is a black box that tends to claim brand demand for itself, so there's a specific test to expose it. Add brand exclusions to your Performance Max campaigns and watch two numbers at once: PMax conversions and total account conversions. If PMax drops but the account total holds steady, PMax was claiming demand that was going to convert anyway through another route.
The same test works with Search Partners, the network of associated sites where Google also places your ads. It's worth checking how much budget they take and the quality of that traffic, because there have been cases where Search Partners ate close to 30% of a campaign's spend on clicks or calls that never turned into customers.
Customer list holdout (Customer Match)
This test fits when you want to measure advertising aimed at people you already know, your customers or your subscribers. You upload your customer list to Google with Customer Match, show the ads only to a randomly chosen half and leave the other half untouched as a control group. At the end you compare the conversion or repeat-purchase rate of the two groups, and the difference between the one that saw the ads and the one left out is the real gain the channel gave you.
It's especially useful for retargeting and campaigns aimed at your base, where it's very easy to claim sales that were going to happen anyway because these are customers who already buy from you. It's the same holdout logic as the earlier tests, applied to a specific customer list rather than to markets or weeks.
Google's lift test
Google has its own incrementality study inside the platform, Conversion Lift, which randomly splits users into a group that sees the ads and one that doesn't, to measure the difference in conversions between them. The good news is that it has become far more accessible: the spend threshold to run it has dropped a great deal, from levels only large accounts could reach to around $5,000, so today almost any advertiser can use it.
It has one drawback worth keeping in mind. It's Google measuring Google, with its own control group and its own way of calculating the result, so it's still judge and jury. Use it, because it's convenient and accessible, but cross-check it against one of the earlier tests that measure against your total sales and don't depend on the platform.
Which metric answers which question
Each type of measurement serves a different decision, and mixing them is what leads to the wrong conclusions. First-click attribution helps you split budget across channels, because it tells you who started the journey. Google's data-driven attribution helps you know which ad performs best inside Google, but only inside Google. And holdout tests are the only ones that answer the question that really matters: whether a channel brings you new customers.
The underlying conclusion is that Google's models can't measure their own incrementality, because they can't see what happens across the rest of your channels. That's why a Google Ads management that measures incrementality with its own tests, not just with the dashboard, is what separates scaling on real data from scaling on attribution that takes credit for other channels' work.
Frequently asked questions
What does it mean for a Google ad to be incremental?
It means it brings sales you wouldn't have got without it. An incremental conversion is one that exists thanks to the ad, rather than one that was going to happen anyway because the customer was already searching for you or had already decided. That's what should really guide how much you invest, above the ROAS the dashboard reports.
Why does Google Ads ROAS look better than it is?
Because the dashboard counts conversions that would have happened anyway, above all those from brand searches and demand created by other channels like paid social. Google measures what happens inside Google very well, but it can't see the rest of your channels, so it takes credit for sales it merely closed.
How do I know if my Google brand ads are worth it?
You can raise your brand spend and look at total clicks, adding paid and organic together over a few weeks. Another option is to switch your brand ads off for a while and see whether your traffic holds. If paying for brand clicks costs you almost as many organic ones, that spend isn't being incremental.
Is it worth excluding your brand in Performance Max?
Yes, it's one of the most revealing tests. By adding brand exclusions to Performance Max and watching PMax conversions and whole-account conversions at the same time, you see whether PMax was generating sales or just picking up brand demand. If PMax conversions drop but the account total holds, it was taking credit for sales that were going to happen anyway.
Can I trust Google's lift test?
It's useful and now accessible, because Google has dropped the spend threshold to run it to around $5,000. The caveat is that it's Google measuring Google, with its own control and its own calculation, so it's best used alongside a holdout test of your own that measures against total sales and doesn't depend on the platform.
What's the difference between attribution and incrementality?
Attribution splits the credit for a sale among the touchpoints the platform can see, and it's useful for deciding budget or for knowing which ad wins inside a channel. Incrementality measures something different: how many sales exist thanks to advertising and wouldn't have happened without it. Only the second tells you whether a channel brings in new customers.
Closing (no heading in the published article)
The Google Ads dashboard is useful for optimising, but it doesn't prove you're winning new customers. Confusing the two is one of the most expensive ways to scale. Any of these tests, from the simplest to the most rigorous, gives you something attribution never can: the certainty of how much of your Google spend brings real business. If you want someone to build that measurement and set your budget on real incrementality, at STRAT we work as a paid media agency for ecommerce, measuring every decision against real profit.
