How to measure Demand Gen creative impact with asset uplift tests
Google’s Demand Gen campaigns have quickly become a cornerstone of modern full-funnel marketing strategies. By leveraging high-intent surfaces like YouTube (including Shorts), Discover, and Gmail, these campaigns allow brands to reach over 3 billion monthly active users. However, as with many top-of-funnel or mid-funnel initiatives, measuring their true impact has historically been a challenge for digital marketers.
The primary hurdle is often referred to as the “attribution illusion.” Because Demand Gen operates across visual and social-style feeds, many users may view an ad, feel an emotional connection to the brand, but choose not to click immediately. When that user later converts via a direct search or a branded organic link, standard attribution models may struggle to accurately assign credit. This leads to a fundamental question for advertisers: Did the Demand Gen creative actually cause the conversion, or would the user have purchased anyway?
To solve this, Google introduced asset uplift experiments in November 2025. This feature provides a scientific framework for measuring the incremental impact of your creative assets. By moving beyond traditional reporting and embracing asset uplift tests, you can stop guessing and start scaling based on hard data.
Why attribution doesn’t equal incrementality
In the world of digital advertising, “attributed” conversions are not always “incremental” conversions. Attribution is a reporting mechanism that connects a conversion event to a specific touchpoint based on a set of rules (such as last-click or data-driven attribution). Incrementality, however, measures the causal lift—the conversions that happened specifically because the ad was shown.
Consider a scenario where a user sees a compelling video ad for a new pair of running shoes on YouTube Shorts. They don’t click the ad because they are busy scrolling, but the creative stays in their mind. Two days later, they search for the brand on Google and complete a purchase. In this instance, Google Ads may attribute partial or full credit to the Demand Gen campaign. While this shows a correlation between the ad view and the sale, it doesn’t prove causation unless you know what that user would have done if they had never seen the ad.
This is where the scientific method becomes essential. Asset uplift tests allow you to establish a baseline by withholding specific assets from a segment of your audience. By comparing a “treatment group” (those who see the ad) against a “control group” (those who do not), you can isolate the variables and identify the exact percentage of lift generated by your creative. This approach is the only way to prove the real-world value of your marketing spend to stakeholders.
What you need before testing creative uplift
Before jumping into the Google Ads experiment interface, it is vital to ensure your account and campaigns are prepared for a rigorous test. Running an experiment without the proper foundation often leads to inconclusive results, wasting both time and budget. There are three primary prerequisites to consider: conversion volume, budget consistency, and creative isolation.
Conversion volume
Statistical significance is the backbone of any valid experiment. If your sample size is too small, a few random conversions can skew the results, leading you to believe a creative is performing better or worse than it actually is. Google recommends a minimum of 50 conversions across both the treatment and control arms of the test during the experiment period.
If your primary conversion—such as a completed purchase or a high-value lead—does not reach this volume, you should consider optimizing the test around high-intent micro-conversions. For example, “Add to Cart” or “Lead Form Initiated” can serve as reliable proxies for success. These micro-conversions provide the data density needed for the algorithm to find a winner more quickly.
Budget minimums
An experiment is only as good as the data it collects, and that data must be collected consistently. If your Demand Gen campaign is frequently “limited by budget,” your ads may stop showing halfway through the day. This creates “noise” in the data because the control and treatment groups may not be receiving a representative sample of daily traffic.
Ensure that your campaign has a sufficient budget to run without interruption for at least four weeks. This duration allows the test to account for weekly fluctuations in consumer behavior, such as the difference between weekday and weekend shopping patterns.
Creative isolation
The golden rule of A/B testing is to change only one variable at a time. If you launch a test where you change the video creative, the headline, and the audience targeting simultaneously, you won’t know which change drove the result. To measure the impact of a specific video or image, keep all other campaign elements—including bidding strategies and standard assets—identical across both arms of the test.
How to run an asset uplift test in Google Ads
Google has streamlined the process of setting up creative experiments, making it easier for advertisers to deploy tests without needing a deep background in data science. Follow these steps to build a sound experiment within the platform.
1. Define a clear hypothesis
A successful test starts with a question, not just a curiosity. Instead of simply “seeing what happens,” define a specific outcome you expect. A weak hypothesis might be: “I want to see if our new video is good.” A strong, actionable hypothesis would be: “Adding a 15-second testimonial-style video to our Demand Gen asset group will result in a 15% increase in incremental conversions compared to our current lifestyle-focused imagery.”
2. Navigate to the Experiments interface
To begin, log in to your Google Ads account and locate the Campaigns tab on the left-hand navigation menu. From there, select Experiments. Click the blue plus (+) icon to create a new experiment and select Asset tests provided by you. Ensure you specifically select the Demand Gen campaign option to access the relevant testing tools.
3. Configure a 50/50 split
When setting up the experiment arms, you will be asked how to split your traffic. For the most accurate results, a 50/50 cookie-based split is the industry standard. This method ensures that once a user is assigned to either the control or treatment group, they stay in that group for the duration of the test. This prevents “cross-contamination,” where a user might see ads from both arms, which would invalidate the incrementality data.
Assign your current, live campaign as the “Control” and the version containing your new creative assets as the “Treatment.”
4. Lock your variables
Once the test is live, it is crucial to maintain a “hands-off” approach. Discipline is key. Do not adjust your bids, change your targeting, or add new keywords to the campaigns involved in the test. Even minor tweaks can reset the learning phase of the algorithm and introduce variables that make it impossible to determine if the creative was the true driver of performance.
5. Set the duration
Patience is required when measuring creative lift. Google Ads experiments generally require a minimum of four weeks to reach a conclusion. The first week is typically a “learning period.” During this time, the Google AI is adjusting to the new audience split and understanding how users interact with the new assets. Actionable, statistically significant data usually begins to emerge between weeks two and four. If you are in a B2B industry with long sales cycles, consider extending the test to six or eight weeks to capture the full customer journey.
What your experiment results actually mean
Once the experiment concludes, Google provides a detailed dashboard within the Experiments tab. This report highlights the performance of each arm and provides a “confidence interval.” This interval tells you how certain the system is that the results weren’t just a result of chance. Here is how to interpret the three most common outcomes.
Outcome 1: Positive lift (statistically significant)
If the treatment group shows a positive lift with at least 95% confidence, your hypothesis has been validated. This is the green light to scale. However, don’t just look at the conversion count; look at the incremental cost per acquisition (iCPA). You can calculate this by taking the total ad spend of the treatment group and dividing it by the number of conversions that occurred above the control group baseline. This iCPA is your true north metric for determining the profitability of your creative strategy.
Outcome 2: Negative lift
It can be discouraging to see that a new, expensive video asset actually performed worse than the original. However, this is a highly valuable result. A negative lift suggests that the new creative may be causing friction, perhaps by being too disruptive or failing to resonate with the target audience. In some cases, a high skip rate on a video can signal to the algorithm that the content is irrelevant, leading to reduced delivery. If you see a negative lift, pause the treatment arm immediately and use the insights to pivot your creative direction.
Outcome 3: Inconclusive result
If the results show no clear winner after four weeks, it usually means one of two things: either you didn’t have enough conversion volume, or the creatives were too similar. If the assets you are testing are only slightly different (e.g., a different color button or a slightly modified headline), they are unlikely to drive a statistically significant difference in a high-volume environment like Demand Gen. For your next test, try a “radical differentiation” strategy—test a highly polished cinematic video against a raw, user-generated content (UGC) style video to see which direction the audience prefers.
Prove creative impact with incrementality testing
In an era where AI-driven bidding and automated targeting have leveled the playing field, creative has become the most significant lever a marketer can pull to drive performance. However, creative is also one of the most expensive and time-consuming parts of the marketing process. You cannot afford to rely on “gut feelings” or “vanity metrics” like likes and views when deciding where to allocate your production budget.
Demand Gen is a powerful vehicle for visual storytelling, but its true value is often hidden behind the complexities of modern user journeys. By utilizing asset uplift experiments, you provide your stakeholders with the scientific evidence needed to justify continued investment. These tests move the conversation from “We think this video looks great” to “We know this video drives a 12% incremental lift in revenue.”
Start by establishing your baseline today. Use these experiments to build a roadmap of what works, and let the data guide your creative evolution. In the competitive landscape of digital advertising, those who master the art of measuring incrementality are the ones who will ultimately win the market.