How to compare lift and attribution, step by step

Usually in seven steps. Pick one campaign and fixed dates, then write down what each attribution rule credits it with in GA4, Google Ads or Meta. Run a lift test on the same campaign and dates. Put the numbers side by side: the attribution figures move when you change the rule, the lift figure does not.

You need admin access to GA4 and the ad account, one campaign with steady spend, and a spreadsheet with one row per rule. The output is a small table: what each rule claims, what the test found, and the ratio between them.

Step by step

1. Pick one campaign and fix the dates. Choose the campaign whose attributed numbers move the most budget. Set test dates that cover your buyers' lag: Google typically recommends lift studies of more than 14 days. Path: Google Ads > Campaigns menu > Campaigns > Experiments > Lift studies tab.
2. Write down GA4's rule. Note the reporting attribution model and the key event lookback window. For purchases the default lookback is 90 days. A model change rewrites history too, since it applies to past and future data. Path: GA4 > Admin > Data display > Events > Attribution settings.
3. Read GA4's credit under two models. Add an Include filter on User campaign so the report shows only your campaign. Pick purchase as the key event, compare Data-driven with Paid and organic last click, and note key events and revenue under each. Path: GA4 > Advertising > Attribution > Attribution models > Add filter > Acquisition > User campaign.
4. Read Google Ads' own claim. For a Google campaign, check which attribution model and windows its purchase action uses. Google's default click-through window is 30 days, and its view-through window is 1 day. Path: Google Ads > Goals icon > Conversions > Summary > your purchase action > Edit settings.
5. Line up Meta's windows. For a Meta campaign, show the same conversions under 1-day click, 7-day click, 28-day click and 1-day view. Then add Incremental, Meta's modelled count of the conversions it considers incremental. Path: Ads Manager > Columns: Performance > Compare attribution settings; then Compare attribution models > Incremental > Apply.
6. Run a lift test on the same campaign and dates. Meta's Experiments tool sets up a Conversion Lift test step by step. In Google Ads, choose Conversion Lift, then Based on users, and add the campaign. Change as little as you can until the test ends. Path: Meta Experiments > Conversion Lift; Google Ads > Lift studies tab > plus button > Conversion Lift > Based on users.
7. Divide, then date the ratio. When the results land, divide the incremental conversions by each attributed figure for the same dates. Write the date beside every ratio, because Google frames incrementality at a certain point in time. Path: Google Ads > Goals menu > Lift measurement > study name; Meta Experiments > Learn > View report.

A worked example

Round, invented numbers, for illustration. Say a Meta prospecting campaign ran a four-week Conversion Lift test, with nothing else changed.

For illustration, the test found 140 purchases the ads added. In the worked example, the usual setting of 7-day click and 1-day view claimed 280, so the ratio is 0.50. Half the claim held up.

The ratios also show which rule sits nearest the test for this campaign. In the worked example, 1-day click claimed 130, close to the test's 140. If your weekly report must use one column for this campaign, 1-day click is the closest here. Don't carry that answer to another campaign: retargeting and prospecting can land far apart.

In the worked example, GA4 credited 120 purchases, fewer than the test found. GA4 counts the visits it records. Its models give direct visits no credit unless the whole path was direct. A buyer who saw the ad and later typed your address lands in Direct.

Windows explain most of the spread between the Ads Manager rows. Some buyers take longer than a week. In one store's Journeys sheet, journeys with 2 to 3 touches took 12.5 days to buy. Journeys with 4 to 9 touches took 16.9 days, and those with 10 or more took 16.0 days, in the same Journeys sheet.

If an ad click opened journeys like those, the sale could land after a 7-day click window had shut. The 28-day click column could still count it. So could the lift test, which counts conversions relative to the start and end dates of the test.

One caution on the division itself. Meta says lift results are not meant to be compared with ad campaign results in Meta Ads Manager. So read the ratio as a rough exchange rate for one campaign at one budget, not a correction factor for the whole account.

What to check when the numbers look wrong

• Every rule claims more than the test found. Expected when a campaign reaches people who were going to buy anyway. Read the size of the gap, not its existence, and test the next campaign before you generalise.
• The test found more than GA4 credited. GA4's models exclude direct visits from credit unless the whole path was direct. Buyers who saw the ad and came back by typing your address count in the test, and in GA4's Direct row.
• The 28-day click column looks thin. Meta says that view only holds partial data until a campaign has been live for 28 days. Compare windows the campaign has fully lived through.
• The Incremental column is empty. Meta shows no incremental attribution results for date ranges before April 1, 2025. Keep the comparison to dates after that.
• GA4's history changed overnight. Someone switched the reporting attribution model, which applies to historical and future data. Save the model's name with every GA4 figure you export.
• Google Ads and GA4 disagree on the same campaign. Google Ads counts inside each conversion action's windows, while GA4 uses its own model and lookback window. Write both rules beside both numbers before you compare either with the test.

What to do this week

1. Pull Meta's window spread for your biggest campaign. In Ads Manager, open the Columns: Performance menu, select Compare attribution settings and pick 1-day click, 7-day click and 1-day view. Pass: the columns sit close together, so the claim barely depends on the window. Fail: 7-day click or 1-day view adds a lot on top of 1-day click, so test that campaign first.
2. Write GA4's rule on the report you share. In GA4, go to Admin, click Events under Data display, then Attribution settings. Copy the reporting attribution model and the lookback window onto the report. Pass: everyone who reads it knows which rule made the numbers. Fail: nobody can say, so no comparison with a test will mean much yet.
3. Save a dated baseline before any test. Once the comparison columns are set in Ads Manager, export the table for the last four weeks and date the file. Pass: you hold the claims a future test result will be held against. Fail: the export lacks the comparison columns, so add them and export again.

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: Set up Conversion Lift based on users (Google Ads Help); Select attribution settings (Analytics Help); Key event attribution models report (Analytics Help); About conversion windows (Google Ads Help); Compare attribution settings in Meta Ads Manager (Meta Business Help Center); How to view results for incremental attribution in Meta Ads Manager (Meta Business Help Center); Best practices to get started with Experiments (Meta Business Help Center); About Conversion Lift (Google Ads Help); Understand your Conversion Lift based on users measurement data (Google Ads Help); View and understand holdout test results across Meta technologies (Meta Business Help Center); Differences between Conversion Lift test results and other reporting tools (Meta Business Help Center); Get started with attribution (Analytics Help)

Related answers

• All Incrementality questions
• What is the difference between lift and attribution?
• How to set up a conversion lift test, step by step
• Incrementality testing explained