If I turned off my best channel tomorrow, would revenue actually drop?
Efri · 4 min read- 1Fix your tracking first. No test works if you cannot tell which channel a customer came from.
- 2Audit what each platform claims. Compare their numbers against your actual revenue to see the gap.
- 3Run a full blackout test. Turn all marketing off in a small segment and measure what revenue remains.
- 4Test your biggest channels one at a time. Hold out each major channel and measure the real drop.
- 5Reallocate based on incremental ROAS, not platform-reported ROAS.
Here is the step-by-step process for knowing which channel to cut and which to scale.
Step 1: Fix your tracking foundation
Before you can test anything, you need to know where your customers came from. Capture every touch point with consistent UTM parameters on every link.
Capture ad platform click IDs: gclid for Google, fbclid for Meta, li_fat_id for LinkedIn, msclkid for Microsoft, ttclid for TikTok.
These click IDs are the only keys that let you match a session on your site to a specific billed click inside a platform account.
Connect your CRM to your ad data. When a deal is created, copy the original touchpoint data onto the deal record. This is the only way to tie a channel to actual revenue, not just a click.
Reconcile against your bank account. Add up what all your platforms claim. Compare it to what you actually collected. The gap is your starting point.
Step 2: Understand your baseline. Before testing individual channels, you need to know how much revenue comes from marketing at all. Run a complete holdout of all marketing in a small segment of your market. Ten percent is a safe starting point. Measure revenue in that segment versus a matched control segment that still sees ads. The difference is how much revenue marketing is actually driving. Everything else is baseline demand that would have happened anyway.
Step 3: Test your biggest channels one at a time
Identify your top three spending channels. Run a holdout test for each, one month at a time. Hold that channel out from a segment of your audience or geography. Keep everything else running normally.
Compare conversion rates between the holdout group and the control group. The difference is the incremental lift for that channel.
For geo holdouts, split your markets into test and control. Run the channel in control. Turn it off in test. Compare revenue between them. A premium fashion retailer ran a geo holdout on Google branded search.
With ads off in test markets, it lost just seven orders. The channel was claiming far more credit than it was actually creating.
For audience holdouts, hold back 10 to 20 percent of your audience from seeing ads on that platform. Compare conversion rates between the exposed group and the holdout group.
Step 4: Measure incremental ROAS per channel
Once you have the incremental lift for each channel, divide incremental revenue by the cost of that channel. That gives you incremental ROAS. This is the only ROAS number that matters for budget decisions.
If a channel has an incremental ROAS of 4x, every dollar you spend returns four dollars in revenue that would not have happened without that spend. If a channel has an incremental ROAS of 1.5x, it is barely breaking even.
Dropbox ran geo-level blackout experiments across search and mobile. They found mobile advertising had substantially lower incrementality than click-based attribution implied. Search displayed more durable causal performance.
Applying these results, Dropbox reallocated approximately $25M of performance marketing spend away from low-incrementality segments. Portfolio incremental LTV:CAC improved by 53 percent versus baseline.
Step 5: Compare against your target ROAS
Set your target ROAS before you test. For example, if your business needs three dollars for every dollar of ad spend, then any channel with incremental ROAS below 3x is a candidate for reduction. Any channel above 3x is a candidate for scaling.
Do not average the platform numbers. If three tools each claim a sale, the average still includes claims that may not be real. You just get a smaller inflated number.
Step 6: Reallocate budget
Cut spend from channels below target incremental ROAS. Increase spend on channels above target. Stay budget neutral where possible. Reallocate, do not just increase total spend.
Let the new allocation sit for a couple of months to re-establish saturation and baseline. Then measure your blended MER to see if overall performance improved. Re-test the full contribution of media to confirm your baseline has shifted.
Step 7: Re-test quarterly. Incrementality changes. Saturation, competition, and audience fatigue all shift the numbers. Repeat the testing cycle quarterly. Run five incrementality tests a year minimum. More complex brands run twenty or more.
What you need to run this
You need a way to split your audience or geography before campaigns run. You need clean tracking so you can measure outcomes in each segment. You need CRM integration so you can tie revenue back to the test.
Tools that support this include geo holdout platforms like Haus and Measured, incrementality testing through AppsFlyer or Kochava, and open-source options like Meridian GeoX.
The bottom line. Platforms will never tell you which channel to cut. Each one claims the sale. The only way to know is to test what happens when you turn it off. Build the tracking foundation. Run the tests. Reallocate based on incremental ROAS. Re-test every quarter. That is how you stop guessing and start knowing.