Demand Gen campaigns explained: Discovery ads evolved

In 2026, one of the most significant shifts in Google Ads is the full transition from Discovery ads to Demand Gen campaigns. What started as a relatively straightforward visual ad format on Gmail, Google Discover and the YouTube homepage has evolved into a versatile, AI-driven campaign type that combines visual reach with concrete conversion objectives. In this article, AdBrains explains exactly what Demand Gen campaigns are, how they differ from the old Discovery ads, when to use them, and how our own AI technology takes this campaign type to a structurally higher level.

From Discovery ads to Demand Gen: the evolution

Discovery Ads vs. Demand Gen campaigns

What changed in 2026?

Discovery Ads (old)

• Static images and carousels only
• No video ads
• Limited audience controls
• No lookalike audiences
• Only Gmail, Discover & YouTube Home
• Limited per-placement reporting
• Manual budget management

Demand Gen (new)Recommended

• Images, carousels and video assets
• YouTube Shorts, in-stream and in-feed video
• Advanced audience segmentation
• Lookalike audiences based on first-party data
• Gmail, Discover, YouTube and Google Display
• Granular reporting per channel and asset
• Integrated Smart Bidding (tCPA/tROAS)

Discovery ads were introduced by Google in 2019 as a way to reach users at moments when they were open to discovering new ideas and inspiration. Placements included Gmail, Google Discover (the personalised feed on mobile) and the YouTube homepage. The format worked well for brand awareness but had clear limitations: no video ads, limited audience controls, and little visibility into per-placement reporting.

In 2023, Google announced the successor: Demand Gen campaigns. By mid-2024, Discovery ads were fully phased out and replaced by this new campaign type. Demand Gen is not simply a rebrand but a fundamental evolution. The biggest differences lie in three areas: expanded placement options (including YouTube Shorts and in-feed video), advanced audience options such as lookalike audiences based on first-party customer data, and deeper integration with Smart Bidding strategies like Target CPA and Target ROAS.

For advertisers still operating with a Discovery ads mindset, it is crucial to understand that Demand Gen requires a completely different approach. You are combining the visual power of social media with Google's intent signals, which creates unique opportunities for both e-commerce and lead generation.

What exactly are Demand Gen campaigns?

Demand Gen is a Google Ads campaign type specifically designed to create demand in the upper and middle funnel stages, while simultaneously generating measurable conversions. The campaign type reaches users across the following Google-owned properties:

• YouTube in-stream ads: ads shown before or during YouTube videos, skippable or non-skippable.
• YouTube in-feed ads: ads appearing in YouTube search results and the recommended videos sidebar.
• YouTube Shorts: short, vertical video ads in the Shorts feed, now one of the fastest-growing formats.
• Google Discover: the personalised news feed on Android and iOS.
• Gmail: sponsored messages in the Promotions and Social tabs.
• Google Display Network: image ads on millions of partner websites and apps.

Within a single Demand Gen campaign, images, carousels and video assets can all be combined. Google's AI determines which combination of asset type and placement offers the best conversion opportunity for a specific user at a specific moment. This makes the campaign type significantly more powerful than its predecessor.

Audience options: lookalike audiences and segmentation

One of the most impactful additions in Demand Gen compared to Discovery ads is the ability to use lookalike audiences. Advertisers can upload a list of existing customers or converters, after which Google's AI identifies similar users based on behaviour, interests and intent signals. This is comparable to Meta Ads' lookalike functionality, but powered by Google's enormous data capabilities.

Beyond lookalike audiences, Demand Gen provides access to the full breadth of Google's audience solutions:

• Custom intent audiences: users who have recently searched for specific terms or visited relevant websites.
• In-market audiences: users Google has classified as actively interested in a particular category.
• Remarketing lists: existing website visitors or app users you want to re-engage (the RLSA principle).
• Customer Match: targeting based on uploaded customer email addresses.
• Life events: users going through a significant life event such as moving house or graduating.

The combination of all these audience options with the broad placement mix makes Demand Gen particularly well-suited for advertisers who want to reach new users as well as reactivate existing customers.

Average performance improvement Demand Gen vs. Discovery

2026 benchmarks — advertisers who fully migrated

More reach via additional placements

41%

Higher CTR through video assets

34%

Lower CPA via Smart Bidding integration

27%

More tracked conversions (lookalike)

31%

Increase in view-through conversions

38%

Assets and creative requirements

Demand Gen is an asset-based campaign type, comparable to Responsive Search Ads (RSA) for search ads or Performance Max (PMax) for cross-channel campaigns. You supply multiple versions of images, headlines, descriptions and videos, after which Google's AI assembles the best combination per user and placement.

For most advertisers, Target CPA or Target ROAS is the most effective choice once the campaign has accumulated sufficient conversion data. In the early phase, we recommend starting with "Maximise conversions" without a tCPA restriction, giving the algorithm enough room to learn.

Demand Gen in practice: e-commerce and lead generation

Demand Gen works for a wide range of business types, but the application differs per model. Consider ToetsJeKennis.nl, an online platform for exams and courses with an average order value of around 50 euros. For this type of e-commerce advertiser, Demand Gen works excellently as a mid-funnel activation: potential students using Google Discover or YouTube who have previously visited the website are re-engaged with visually compelling carousel ads showing specific courses. By building a lookalike audience based on existing buyers, new and previously unknown users with a similar profile can also be reached.

For a lead generation advertiser like Clima-Active.nl (specialist in air conditioning and heat pump installation), the focus is different. Here, Demand Gen is used to stimulate quote requests from users in the consideration phase. Video assets with a brief explanation of the benefits of heat pumps are shown to in-market audiences for "energy saving" and "home renovation". The combination of informative video and a sharp call-to-action generates quality leads at a competitive CPL.

A third example: E-4motion.com, the webshop for new electric folding bikes, uses Demand Gen for both objectives simultaneously. Short YouTube Shorts videos show the bikes in action (e-commerce activation), while a separate ad group with a "Book a test ride" message is aimed at generating test ride requests (lead generation).

Demand Gen funnel: from impression to conversion

Average e-commerce benchmark — percent mode

100% Impressions

Full audience reached across all placements

5% Clicks (CTR)

Average 5% CTR with optimised video+image mix

3% Engaged sessions

60% of clickers show active engagement on the landing page

0.4% Conversions

Average 0.4% end-to-end conversion rate for e-commerce

A fully optimised Demand Gen funnel delivers strong results at every stage.

How AdBrains AI optimises Demand Gen campaigns

Demand Gen campaigns may seem straightforward to set up, but real performance is determined by the quality of optimisation after launch. This is precisely where AdBrains distinguishes itself with our own AI technology, developed specifically for managing Google Ads and Meta Ads campaigns.

Our multi-agent verification system plays a key role in Demand Gen management. Every optimisation decision, from pausing a poorly performing asset to adjusting a tCPA target, is first reviewed by four independent AI agents before the change is actually implemented. This prevents impulsive adjustments that disrupt the learning phase and ensures stable, consistent campaign performance.

For the creative side, AdBrains applies an RSA improvement system also to Demand Gen assets. Our AI continuously analyses which headlines, descriptions and visual combinations generate the highest Ad Strength score and the best CTR. Assets with POOR status are automatically flagged for replacement, and our system generates concrete improvement suggestions based on landing page analysis and industry-specific language patterns.

Our automated tCPA/tROAS optimisation adjusts bidding strategies daily based on current conversion volume and the client's margin targets. For e-commerce clients like ToetsJeKennis.nl, the tROAS is continuously adjusted as seasonal patterns or promotions change. For lead generation clients like Clima-Active.nl, the system optimises on CPL while accounting for lead quality measured via CRM integration.

Our audience management automation ensures that PROD audiences, incubator audiences and RLSA lists are automatically updated weekly. For Demand Gen, this is crucial: lookalike audiences are only precise when the source list is current and sufficiently large. AdBrains' AI monitors list sizes and automatically signals when a list has become too small for effective lookalike generation.

Finally, our server-side signal enrichment via a proprietary sGTM infrastructure provides enriched conversion signals in Demand Gen campaigns. By linking first-party data to conversion events, Google's Smart Bidding algorithm receives richer signals about the value of each conversion, resulting in better audience modelling and ultimately lower CPAs and higher ROAS values than advertisers relying solely on standard conversion tracking.

Demand Gen versus Performance Max: when to choose which?

A frequently asked question is how Demand Gen relates to Performance Max. Both campaign types use AI and multiple placements, but they have a fundamentally different purpose. Performance Max is designed for maximum conversion efficiency across all Google channels, including Search, Shopping, Display, YouTube and Gmail. Demand Gen is specifically focused on visual channels and places greater emphasis on audience targeting and creative control. For advertisers already running Performance Max, Demand Gen is a valuable complement, not a replacement.

Frequently asked questions about Demand Gen campaigns

What is the difference between Demand Gen and Discovery ads?

Discovery ads were the predecessor of Demand Gen. The biggest difference lies in available placements and asset types. Demand Gen adds YouTube Shorts, in-stream video and the Google Display Network to the existing Discovery placements (Gmail, Google Discover, YouTube homepage). Additionally, Demand Gen offers lookalike audiences, deeper Smart Bidding integration and granular reporting per channel. Discovery ads were fully phased out by mid-2024.

What budget is suitable for Demand Gen?

Demand Gen can work effectively from a daily budget of 20 to 30 euros, but for optimal Smart Bidding learning performance, a minimum of 50 euros per day is recommended. The campaign type needs sufficient conversion data to train the algorithm. For smaller budgets, we recommend starting with "Maximise conversions" without a tCPA target, and only setting a Target CPA or Target ROAS once 50 or more conversions per month have been tracked.

Can I run Demand Gen and Performance Max simultaneously?

Yes, and for many advertisers this is actually the most effective account structure. Performance Max focuses on all Google channels with a strong emphasis on conversion efficiency, while Demand Gen handles the visual mid-funnel component. Ensure clear audience segmentation and budget allocation to avoid overlap and budget cannibalisation. In AdBrains' approach, we continuously analyse overlap and automatically adjust campaign settings to maintain clean separation between the two campaign types.

How do I measure the success of a Demand Gen campaign?

Beyond standard conversion KPIs (CPA, ROAS, conversion rate), there are Demand Gen-specific metrics that deserve attention. View-through conversions measure users who saw an ad but did not click directly, yet converted later. Engaged-view conversions (specific to YouTube) measure conversions after a meaningful video interaction. Ensure conversion tracking is correctly set up and Enhanced Conversions are activated for the most complete measurement. AdBrains' server-side tracking ensures conversions are captured even when they would otherwise be lost to ad blockers or cookie restrictions.

Are lookalike audiences in Demand Gen as effective as on Meta Ads?

Lookalike audiences in Demand Gen are based on Google's own intent and behavioural data, which differs in nature from Meta's social graph data. Both have their strengths: Meta's lookalikes are powerful for broad consumer segments and social behaviour, while Google's lookalikes are particularly effective for users who are actively in the research and consideration phase. For most advertisers, both platforms deliver complementary results, and an integrated approach across both Google and Meta provides the broadest coverage.