How to pause an ad channel for a test, step by step

Usually in eight steps. List every campaign that buys the channel, decide which stand-ins may keep running, and pick the regions that go dark. Switch the channel off there, check that spend really stops, and log every other change. Then compare total sales by region, and keep counting after the restart.

This walkthrough covers the pause itself: the switches, the leaks and the checks. For pairing regions and sizing the test, use the geo holdout walkthrough. For dates, use the test length guide. The menu paths come from the Google Ads, Meta, Shopify and GA4 help pages, read on 1 October 2026.

Step by step

1. List every campaign that buys the channel. A channel is rarely one campaign. Google's Performance Max is served on Search, Maps, YouTube, Discover, Gmail and the Google Display Network. So it may buy the very channel you plan to pause. Path: Google Ads > Campaigns menu > Campaigns > Insights and reports > Channel performance.
2. Decide which stand-ins may keep running. Pausing one campaign can hand its auctions to another. Leave a stand-in on if you would keep it after a cut, and the test prices that cut. Exclude the dark regions from it as well if you want the paused channel's whole effect. Path: Google Ads > Campaigns menu > Campaigns > Settings > Locations.
3. Pick the dark regions from your own customers. Count new customers by region over recent months. Then split the regions into two groups that looked alike. Shopify places each new customer at their most recent shipping location. Path: Shopify admin > Analytics > Reports > Category filter > Customers > Customers by location.
4. Switch the channel off in the dark regions. In Google Ads, open each campaign's Settings, expand Locations, search for each dark region and click Exclude. Choose Presence under Location options, as Google's geo guide does for exclusions. On Meta, exclude the same areas in every ad set. Path: Google Ads > Campaigns menu > Campaigns > Settings > Locations > Exclude; Meta Ads Manager > ad set > Audience > Locations.
5. Check that dark really went dark. Two days in, look at spend by location. Google's Matched locations report shows performance by the locations that matched your ads. Meta's Ads Reporting can break spend down by Region. For website conversions, its only geographic breakdown is Country. Path: Google Ads > Campaigns > Insights & reports > When and where your ads showed > Matched locations. Meta: Ads Reporting > breakdowns > Geography > Region.
6. Log every change while the test runs. Google Ads keeps a change history that shows when a campaign was paused and what its budgets were. Add promotions, emails and price changes to the same log, with dates. Path: Google Ads > Campaigns menu > Change history.
7. Compare total sales by group, then look for leaks. Compare how each group's sales moved against its own weeks before the test. The extra drop in the dark group is the channel's work. In GA4, compare the groups in Traffic acquisition, and watch Cross-network and Organic Shopping too. Path: GA4 > Reports > Acquisition > Traffic acquisition > comparison icon > + Create new > Dimension.
8. Restart at the old budget and keep counting. Remove the exclusions on the date you planned, at the budgets the campaigns had before. Google's geo guide says not to raise a go-dark campaign's budget during the cooldown. Count sales until your slowest buyers have had time to finish. Path: Google Ads > Campaigns menu > Campaigns > Settings > Locations.

A worked example

For illustration, with round numbers made up for the example. Say a store spends about €6,000 every four weeks on a YouTube video campaign, and also runs Performance Max. It would keep Performance Max whatever happens, so it lets that campaign run.

Say the store's regions split into two groups that each sold about €25,000 a week before the test. YouTube goes dark in one group for four weeks. Both groups are counted for two more weeks after the restart.

For illustration, the dark group's sales fell €4,500 short of the lit group during the dark weeks. Stop counting at the restart and that looks like the whole story. In this worked example, another €1,500 went missing in the two weeks after YouTube came back, as interrupted journeys failed to finish. So the worked example's shortfall is €6,000, not €4,500.

The saving needs the same care. In this worked example, the pause held back €3,000 of YouTube spend, but Performance Max spent €600 more in the dark regions. So the worked example's net saving is €2,400.

For illustration, €6,000 of sales against €2,400 of net saving is 2.5x. The yardstick is one store's Break-even sheet, where a 40% margin gives a break-even ROAS of 2.5x (1 divided by 0.40). In this worked example, YouTube roughly pays for itself, so keep it or test a smaller budget.

The careless reading tells another story. For illustration, counting only the dark weeks and ignoring Performance Max gives €4,500 on €3,000, which is 1.5x. That version would cut a channel that pays its way.

What to check when the report looks wrong

• Spend still shows in a dark region. A trickle is normal. Meta decides by current and home location, so a visitor can still see your ads there. More than a trickle means a campaign from step 1 is missing the exclusion.
• Region rows don't add up to the campaign total. Google says metrics below country level may not add up. Some impressions match only at country level. Compare trends by region, not sums.
• The lit regions dipped on day one. If you swapped country targeting for a list of regions, Google's geo guide says overall traffic falls. It can't place every user that precisely. Compare the groups with each other, not with last month.
• Cross-network rose in the dark regions. GA4 files Performance Max visits there, so it picked up some of the paused channel's buyers. That is expected if you let it run. Count its extra spend against the saving, as in the worked example.
• Sales kept falling after the restart. That is the slow lane catching up, not a broken restart. Keep counting until it settles before you read the result.

What to do this week

1. Open the Matched locations report for one campaign. In Google Ads, click Campaigns, then Insights & reports, then When and where your ads showed. Select Matched locations. Pass: spend sits in the regions you meant to target. Fail: a visible share lands elsewhere, so fix the campaign's locations before any test.
2. Find the Region breakdown in Meta Ads Reporting. Add Region under Geography and read Amount spent for last month. Pass: spend shows for each region you might pause. Fail: one region takes most of the spend, so split it by city before you plan the test.
3. Write the restart plan before the pause. Note each campaign's daily budget today, the restart date and who will restore it. In Google Ads, the budget sits under Budget and bidding optimization in the campaign's settings. Pass: all three are written down. Fail: the restart date floats, which is how a test turns into a cut nobody decided.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: About Performance Max channels (Google Ads Help); About the channel performance report for Performance Max (Google Ads Help); Customers reports (Shopify Help Center); Exclude ads from geographic locations (Google Ads Help); Implement campaigns for geo experiments (Google Ads Help); Use location targeting (Meta Business Help Center); View matched locations and distance reports (Google Ads Help); About breakdowns, metrics and filtering in Meta Ads Reporting (Meta Business Help Center); About change history (Google Ads Help); Traffic acquisition report (Google Analytics Help); Apply comparisons to reports (Google Analytics Help); Default channel group (Google Analytics Help).

Related answers

• All Incrementality questions
• How do I pause a channel to test it?
• How should an electronics store pause a channel to test it?
• Geotargeting, in the glossary
• Incrementality testing, explained