Demand Gen Campaigns in Google Ads

A buyer rarely comes “out of nowhere.” First, they see a video on YouTube Shorts, then a banner in Gmail, then a product card in the Discover feed—and only a week later do they enter a search query. Google Demand Gen campaigns focus specifically on this first stage: they reach people right when they’re searching for a product and spark their interest.

Let’s break down how the format works, who it’s best suited for, and how to set up a Demand Gen campaign in Google Ads without blowing your budget.

What Is Google Demand Gen?

Demand Generation is a type of campaign in Google Ads built on visual ad formats and machine learning algorithms. The format has been available to advertisers since October 2023 and replaced Discovery Ads. The focus is on visuals, automation, and generating demand—not just responding to existing demand.

The key difference from traditional search ads is the targeting logic. Search ads capture ready-made demand: a person types “buy a Nike tracksuit” and sees an ad. Demand Gen relies on behavioral signals and interests: video views, clicks in Gmail, and activity in the Discover feed. The user wasn’t searching for anything—the ad found them at the right moment.

Essentially, this is Google’s response to Meta’s advertising model: the same approach to reaching a cold audience, visual creatives, and a recommendation feed. The only difference is the ad placements—Google’s own.

Key Features of Google’s Demand Generation Campaigns

Google demand-generation campaigns stand out significantly from other tools. They use a different logic for audience selection, have their own set of platforms, and feature proprietary tools for measuring effectiveness. There are five key features of this tool you should understand before launching a campaign—your strategy and expectations for results depend on them.

A unique set of ad placements

A single campaign covers several areas of the Google ecosystem at once:

• In-Stream—ads that play before, during, and after other videos;
• In-Feed — ads in the main feed and video search results;
• YouTube Shorts — ads within short vertical videos;
• Discover feed — recommendations on Google’s mobile homepage;
• Gmail — ads in the “Promotions” and “Social” tabs.

No other type of advertising offers this combination. The audience for these platforms is enormous: more than 50% of users regularly watch reviews on YouTube before purchasing a product or service, and another 39% do so from time to time. The ad appears native here—as part of the feed, rather than as an intrusive banner.

Important: You cannot disable individual ad placements. The system’s algorithms automatically distribute impressions to where the likelihood of a target action is highest. You can only influence this distribution indirectly—through your selection of creatives. For example, if you don’t include vertical video in your setup, Shorts won’t be included either.

Targeting a “cold” audience

The format was originally designed to reach people who haven’t heard of your business yet. Hence the main tools: lookalike audiences, interest-based targeting, and broad reach. The system searches for users whose behavior resembles that of your customers—and introduces them to your product.

Targeting is the key advantage. Lookalike audiences (similar segments) are available here, which are not available in other types of Google Ads campaigns following the discontinuation of classic similar audiences.

Here’s how it works: you upload a source segment—a customer base, website visitors, or channel subscribers—and the system finds users with similar behavior. There are three reach levels: narrow (about 1%, the most similar people), balanced (~5%), and broad (10%+). Narrow provides accuracy but little traffic; broad does the opposite.

Tip: The quality of the Lookalike audience depends directly on the source audience. A segment of 100 random website visitors will yield poor results. Build your audience from buyers or those who have submitted a request—at least 1,000 people.

In addition to Lookalike, you can target based on interests and intent, demographics (gender, age, income), your own data for remarketing, and life events such as weddings, moving, etc. Don’t forget about exclusions—remove existing customers so you don’t pay for those who have already purchased.

This also changes the requirements for ad creatives. A cold audience isn’t ready for a direct “Buy Now” message. Approaches that focus on utility, emotion, or curiosity get the best response: “5 Mistakes to Avoid When Choosing a Mattress,” “What a 150,000 UAH Kitchen Renovation Looks Like.” First, spark interest—then make the sale.

Focus on the upper stages of the sales funnel

This format addresses the objectives of the top of the funnel: reach, brand awareness, and generating interest. Expecting instant sales at the cost of search ads is a mistake—one that causes most tests to fail.

The proper role of Demand Generation in this strategy: the top stage is product awareness; the middle stage is nurturing interested prospects; and the bottom stage is the purchase via retargeting and search. A person saw an offer today but made a purchase two weeks later using a brand-related keyword—and this is a normal scenario, not a failure.

AI-Based Reporting

The classic problem with media traffic is, “We spent money—but what did we get in return?” Here, that problem is solved. To achieve this, we use:

1. Brand Lift. It measures the impact of advertising on brand perception. The system divides the target audience into two groups—those who saw the ad and a control group—and surveys both directly on YouTube. Comparing the responses demonstrates an increase in ad recall, brand awareness, and purchase intent. These aren’t guesses—they’re measurable percentages.
2. Search Lift. It works similarly but measures something else: how much the number of searches for your brand on Google and YouTube has increased among those who saw the ads. You launched a campaign—brand-related searches increased by 18%—that’s a proven effect you won’t see in a click report.

One caveat: to get accurate Brand Lift and Search Lift measurements, you’ll need to invest a significant amount—small data sets won’t be enough to compare groups. Check with Google Ads support for the current thresholds specific to your market.

Combining Google Shopping and Display Advertising

For online stores, Demand Generation combines the capabilities of shopping and display ad campaigns. Connect a feed from Merchant Center—and your ads will turn into a storefront: product cards with photos, prices, and names will appear in YouTube, Discover, and Gmail feeds.

The result is a hybrid: the reach and visual engagement of display ads, plus specific product details. A user scrolls through a feed of recommendations, sees a selection of sneakers with prices in hryvnias—and clicks directly on the product card, bypassing the homepage.

What ad formats and creatives are available in Google Demand Gen?

Demand Gen ad campaigns support three ad types: video ads, single-image banners, and carousels consisting of 2–10 cards. For e-commerce, there’s a fourth option—feed-based product ads.

Rule of thumb: Upload creatives in all aspect ratios. Horizontal 16:9 for In-Stream, square 1:1 for feeds, and vertical 9:16 for Shorts and Discover. Did you only upload a horizontal banner? You’ve missed out on the cheapest impressions in vertical placements.

A few practical tips on creatives:

• The first 3 seconds of a video are everything. Put your offer right at the beginning—don’t drag out the intro.
• Prepare at least 3–5 variations of images and headlines. Google will generate ads from the content you provide and will show combinations with higher performance more often.
• Banner text should be large and concise. Small fonts are unreadable on a smartphone screen.
• Show the benefits specifically: not “great terms,” but “delivery throughout Ukraine starting at 70 UAH” or “500 UAH off your first order.”

How does Google Demand Generation differ from Performance Max?

This is the question PPC specialists are asked most often. The formats are similar: both are automated, and both use artificial intelligence. But their objectives are different.

Simply put: PMax targets those who are already ready to buy, while Demand Gen in Google Ads brings new people into the funnel. These are complementary tools, not competitors.

Advantages of Demand Generation campaigns over Google Discovery

Demand Generation isn’t just a rebranded Discovery—it’s an improved version with expanded functionality. Google automatically migrated all Discovery campaigns to the new format back in 2024, so there’s essentially no choice left. But it’s helpful to understand the difference—especially for those familiar with the old format who are carrying their expectations over to the new one.

What’s new:

• Video ads. Discovery used only images. Demand Gen supports video ads, including In-Stream and Shorts—and that’s a huge source of low-cost traffic.
• New placements. In-video and short-form video placements have been added to all previous options.
• Lookalike audiences. Previously, there was no Lookalike tool—targeting was limited to interests and your own data.
• More bidding strategies. “Maximize Clicks” and target return on ad spend (tROAS) have been added to “Maximize Conversions” and tCPA.
• A/B tests. Test creatives and audiences within ad campaigns—Discovery didn’t support this.
• Carousel and product feeds from Merchant Center.

In short: Discovery allowed you to display “banners in feeds,” while the new format has become a full-fledged powerhouse for the top of the sales funnel—with videos, carousels, product cards, and smart reporting.

How do you create a Demand Gen campaign in Google Ads?

Before launching, check your setup: conversion tracking is enabled on your website, remarketing audiences are set up, and creatives are ready in all variations. Next—the actual setup of Demand Gen campaigns, step by step.

Setting Up an Ad Campaign

In your Google Ads account, click “+ Create Campaign” and select a goal—“Sales,” “Website Traffic,” or the option without specifying a goal (which offers more flexibility in settings).

Campaign type: “Demand Generation.” In the Russian-language interface, the format is called exactly that.

Set conversions for optimization. Choose meaningful actions—such as purchases or lead submissions—rather than button clicks or page views. The algorithm optimizes based on what you specify—if you set a meaningless objective, you’ll get meaningless traffic.

Set up geotargeting and language. For Ukraine, you typically set the location to “Ukraine” and three languages: Ukrainian, Russian, and English—the device’s interface doesn’t always match the user’s language. Keep this in mind from the start: once the campaign is live, you won’t be able to change the language or location at the ad group level. If you make a mistake, you’ll have to recreate the ad group.

Check the ad schedule and dates. For limited-time promotions (“discounts through July 31”), set the end date right away.

Setting Up Your Target Audience

Create an ad group and add a target audience. The main rule is one audience per group. This way, the system will receive a clear signal, training will happen faster, and you’ll be able to see which segment is more effective.

A working framework to get started: three ad groups:

• Lookalike based on your customer or buyer database;
• Interests + demographics (specific segments based on competitors’ keywords also go here);
• retargeting: website visitors, video viewers, email list.

Let’s say you’re promoting an English language school in Kyiv. The first group is a lookalike segment based on your student database. The second is “people interested in learning English,” ages 18–44. The third is remarketing to those who watched your videos but didn’t sign up for a trial lesson. After a month, the statistics will show you where to reallocate your budget.

The “Optimized Targeting” option allows the system to go beyond the specified segments if it detects a likelihood of conversion. It’s best to turn this off at the start—first test your audiences in their pure form, then expand later.

Creating Ads

Upload as much content as possible to the ad group: up to 5 headlines, up to 5 descriptions, images in all aspect ratios, videos, and a logo. Google will compile these elements into an ad and will show combinations that generate orders more frequently.

On the right side of the interface, you’ll see a preview of how your ad appears on different platforms and an ad quality score—aim for an “Excellent” rating by adding any missing formats. Write headlines with different angles: one focusing on the benefit (“English course starting at 2,400 UAH/month”), another on the pain point (“Do you only understand TV shows with subtitles?”), and a third on the result.

For online stores, connect the product feed from Merchant Center at this step—your ads will turn into product cards with prices.

Need to test a second audience? Copy the ad group via the three-dot menu and replace the segment—it’s faster than building one from scratch.

Bids, Budget, and Strategies

Available bidding strategies:

• Max Clicks—for reach and traffic goals;
• Maximum Conversions—the best starting point for most projects;
• Target Cost Per Acquisition (tCPA)—once you have enough data;
• Target Return on Ad Spend (tROAS)—for e-commerce with end-to-end analytics.

At the very beginning, it’s best to enable the “Maximum Conversions” option without setting a target cost, so the algorithm can collect as much data as possible, and then work toward strict targets. Did you set the tCPA too low right from the start? The campaign won’t gain traction: there aren’t enough impressions, and the algorithm needs time to learn.

Now, regarding the minimum budget allocation. Google recommends setting a daily budget based on at least 15 expected conversions—that is, 15 × your target CPA. For the Ukrainian market, a realistic starting point is 400–600 UAH. You can go lower, but the learning phase will take longer, and you won’t have enough data to draw conclusions.

After setting your bids, double-check everything and publish the ad campaign. It will appear in your account’s general list.

Tips for Setting Up Demand Gen Ad Campaigns

1. Give the ad campaign at least 30 days. The first two weeks are spent training the algorithms—don’t change the basic settings during this period. Any sudden change resets the training, and the countdown starts over. A specialist who pauses an ad on the 14th day because there are no leads is cutting the test short halfway through.
2. Evaluate performance across the entire account, not just based on the last click. Case study: A campaign with a daily budget of $80 showed a CPA twice as high as search ads, but the total number of orders increased by 23%, and the ROAS for the search ad campaign rose from 4.1 to 5.3. The campaign was kept running because it increased the advertiser’s overall revenue, even though it appeared unprofitable in the report itself.
3. Feed the algorithm with creative assets. If you use only one banner, you’re guaranteed to hit a performance ceiling: the system has nothing to test. Keep 3–5 images and at least 2 videos in rotation per ad group, and update them every 4–6 weeks—creatives lose their appeal, CTR drops, and traffic costs rise.
4. Don’t narrow your target audience down to micro-segments. A segment of 5,000 people won’t give the AI enough room to work. Target broad audiences and manage quality through exclusions: existing customers, irrelevant demographics, and those who have purchased in the last 30 days.
5. Spread the funnel stages across different ad sets. Cold audience, warming up viewers with video ads, and retargeting website visitors—these are three distinct objectives with different ad variations and bids. In a single campaign, they compete for the budget, and the system almost always diverts funds to remarketing: conversions are cheaper there, but you get zero new customers.
6. Build touchpoint sequences. A working sequence: a cold audience sees a helpful video → viewers see a carousel of products → website visitors see a banner with a 300 UAH promo code. Each subsequent step targets those who responded to the previous one.
7. Use A/B tests instead of making changes “by eye.” Want to test a new landing page or a set of creatives? Run an experiment and compare the numbers under equal conditions. This is more reliable than changing everything at once and guessing what exactly made a difference.

And one piece of honest advice to wrap things up: don’t run Demand Gen campaigns for urgent services. Locksmiths, tow trucks, boiler repairs in winter—there’s no time to build demand here; only search ads work. This format works best where the decision to buy takes days or weeks to mature: clothing, cosmetics, courses, electronics, home repairs, and travel.

How do you evaluate the results of Demand Gen campaigns in Google Ads?

Metrics depend on the stage of the sales funnel. For awareness campaigns, look at CPM, video view percentage, Brand Lift results, and growth in brand-related search queries. For conversion campaigns, look at CPA, conversion rate (the standard for this format is 0.5–3% of clicks), and the bounce rate on the landing page. A high bounce rate is a sign that the creative promises one thing, but the page delivers another.

One month after launch, compare the following: Have brand-related search queries increased? Has the CPA in search ad campaigns changed? Have associated conversions involving Demand Generation appeared? If there’s no change across all metrics, change your creatives and target audience. If at least two metrics have shifted, scale your marketing budget in 20% increments to avoid disrupting the algorithm’s learning process.