Measuring Incrementality: Do Your Ads Really Bring Additional Leads?

Your brand campaign reports a ROAS of 12, your retargeting campaign a cost per lead of $15. Both look great. But how many of those leads would have come anyway, because people already knew you and would have searched for you?

No attribution model answers that question. Only a test with a control group does. In October 2026, Google made these tests available for Search and Performance Max without a Google rep. This article explains what incrementality means, what changed at Google and how to check in lead generation, even without a huge budget, whether your ads really bring additional leads. As of November 2026.

Key points

• Incrementality is the share of conversions that wouldn't have happened without the ad. Attribution only shows who gets credit for a conversion.
• Google Conversion Lift for Search and Performance Max can be set up by eligible accounts themselves since October 2026. Before, it required a Google rep.
• Requirements according to Google: at least 1,000 observed conversions and a campaign budget of at least $5,000, a duration of at least 7 days, 14 or more recommended.
• Results are incremental conversions and cost per incremental conversion (iCPA), and with conversion values also incremental ROAS.
• In lead generation, a lift test usually measures incremental leads, not incremental customers. Few accounts reach 1,000 conversions with deals.
• With low volume, simpler tests help: pause regions, switch campaigns off and on deliberately and always compare against leads and deals in your CRM.

Attribution or incrementality?

Attribution answers the question: which ad gets credit for this lead? Whether it's first-click attribution, last-click attribution or an ad platform's attribution window, every model distributes conversions that happened across channels and ads.

Incrementality asks something else: how many of those conversions did the ad cause? Google describes incrementality as the way to measure the causal impact of ads (Google Ads Help).

The difference is small for some campaigns and huge for others:

On top of that, how many conversions get credited also changes with the platform's measurement method. A current example is Meta on-device measurement, which is set to count more conversions from November without any ad doing more.

If you allocate budget by credited conversions alone, you often strengthen exactly the campaigns that add the least.

How a lift test works

A conversion lift test randomly splits the audience into two groups. The test group sees your ads, the control group doesn't. After the test period, the platform compares how many conversions happened in each group. The difference is the lift, the conversions caused by the ads.

A simplified example with a Search campaign over four weeks:

Leads credited by Google               = 400
Leads in the test group                = 400
Expected leads without ads             = 260   (projected from the control group)
Incremental leads                      = 400 - 260 = 140
Budget                                 = $8,000

Cost per lead according to platform    = $8,000 ÷ 400 = $20
Cost per incremental lead              = $8,000 ÷ 140 ≈ $57

So the campaign does bring additional leads, but each one costs almost three times what the cost per lead in the ad account suggests. When you decide whether more budget pays off, the second number is the right one.

What changed at Google

In October 2026, Google revised its help page on user-based Conversion Lift studies. According to trade press, Search and Performance Max no longer require a Google rep (PPC News Feed, Search Engine Roundtable). There was no official announcement, and other Google pages still refer to the rep. So check whether the option shows up in your account.

According to Google, user-based studies work like this (Google Ads Help):

• Campaign types: Search, Performance Max, Display, Video, Demand Gen and App (excluding iOS).
• Minimums: 1,000 observed conversions and a campaign budget of at least $5,000.
• Control group: 1 to 50% of users don't see the ads.
• Duration: at least 7 days, 14 or more recommended. With a long time to conversion, short studies can understate the lift by up to 17%, according to Google.
• Results: incremental conversions and iCPA or, with conversion values, incremental conversion value and iROAS.
• Where: under Campaigns in Experiments, tab Lift studies.

A campaign can only be in one lift study at a time.

Why lift tests are hard in lead generation

An online store measures purchases that happen minutes after the click. Lead generation has two hurdles:

• Leads, not customers: Most accounts only reach 1,000 conversions in a study with leads. A lift test then shows whether the campaign brings additional form fills. Whether they turn into additional customers stays open.
• CRM data only counts partly: According to Google, enhanced conversions for leads count as directly measured. Imported offline leads, however, only feed in as modeled supplementary data and don't count toward the 1,000 conversions (Google Ads Help).

Then there's the sales cycle. If there are three weeks between lead and deal, a 14-day study is over before most deals are decided.

A lift test on leads is still worth it if you keep an eye on lead quality: after the test, check in your CRM how many leads from the test period were qualified and closed. How to measure real ROAS from click to deal is covered in the guide on ROAS in lead generation.

Checking incrementality with low volume

If you don't reach the minimums, you can still check incrementality roughly. These tests are less precise than a lift test, but they often show the direction:

1. Region test

   Pause a campaign in a few comparable regions, for example two states, and keep it running everywhere else. Over four to six weeks, compare leads and deals per region with the period before. Google also offers geo-based studies officially, but they still require a rep.

2. On-off test

   Switch off a campaign where you suspect low incrementality, such as brand search or retargeting, for two to four weeks. If total leads in your CRM barely drop, most of the credited leads weren't additional. Watch for seasonal effects and don't change anything else during that time.

3. Compare totals, not platform numbers

   Judge every test by the leads and deals that actually reach you, not by platform reports. The ad platform keeps crediting everything to itself during a test.

4. Feed the result into budget planning

   Calculate with the cost per additional lead or deal, not the cost per lead from the ad account. Repeat the test once or twice a year, which Google describes as typical.

Common mistakes

• Too short a test: A one-week test shows little for leads with a sales cycle.
• Several changes at once: If you introduce new creatives, bids and landing pages during the test, you won't know what made the difference.
• Looking only at leads: A campaign can bring additional leads that never buy.
• Adding up platform results: Meta and Google often credit themselves with the same leads. A lift test on one platform says nothing about the other.
• Testing once and never again: Incrementality changes with budget, season and competition.

How LeadMetrics helps

LeadMetrics doesn't run lift tests, but it provides the numbers you measure every test against:

• Leads and deals per campaign: Lead tracking attributes every lead to a campaign and ad with first-click or last-click attribution. Deals from HubSpot and Close and payments from Stripe, CopeCart and Digistore24 are matched to the original lead.
• Compare test periods: In the dashboard you see leads, deals, cost per acquisition and revenue for any date range, including for channels you've just paused.
• Better signals for Google: The Conversion API sends leads and won deals with value to Google Ads, Meta, TikTok and LinkedIn, so the platforms optimize for deals instead of just forms.