Google Ads: GDN Campaigns Move Into Demand Gen

What changed

Standalone Google Display Network campaigns are done. Google confirmed on May 26, 2026 that GDN inventory is moving entirely into Demand Gen, following the same playbook it used to absorb Video Action Campaigns earlier. Starting June 2026, advertisers can migrate existing campaigns with a dedicated migration tool. At some point after that, new standalone Display campaigns can no longer be created at all, and by 2027 any unmigrated campaigns get moved automatically.

The GDN inventory itself, roughly 2 million sites, videos, and apps, isn't going away. It's still accessible inside Demand Gen through channel controls set at the ad group level. Those controls let you pick which surfaces an ad group serves on: GDN, YouTube, Discover, Gmail, Maps, or some combination. If you want GDN-only behavior, you can have it. The difference is that you're now managing it inside Demand Gen's creative, audience, and reporting framework rather than inside the simpler standalone Display interface.

Who it affects

Anyone running standalone GDN campaigns. That's a wide group: direct-response advertisers using Display for lower-funnel retargeting, e-commerce brands running prospecting on GDN at scale, and agencies managing a large portfolio of Display-heavy accounts. If your account has active Display campaigns and you haven't touched them in a while, set-and-forget GDN retargeting is common, this is a forced migration with a deadline, and the settings that come out the other side depend on whether you do it yourself or let Google do it for you.

Why it matters

The mechanism here is consolidation into AI-driven cross-surface buying. Demand Gen is built around Google's audience and creative optimization signals across YouTube, Discover, Gmail, Maps, and GDN together. When you fold GDN into that system, the algorithm gets more surface area to find your conversions, which is the logic behind Google's 9.5% average ROI increase claim. More inventory, more signals, more paths to a conversion.

That said, "average" is doing a lot of work in that number. It comes from Google, not a third-party study. Accounts with tightly controlled GDN retargeting lists, where you've spent months pruning placements and exclusions, may find that Demand Gen's looser, ML-driven targeting erodes efficiency before it improves it. The GoFood case study (24% CPA decrease, 19% higher conversion volume) is a single data point from a high-volume app advertiser, which is exactly the type of account that benefits most from multi-surface ML optimization.

The subtler issue is creative. Demand Gen has different asset requirements than classic Display. It expects image and video assets built for visual-first surfaces: square and vertical crops, motion assets, format variety. Teams running GDN with a small library of static display banners will hit creative friction immediately after migration if they don't prepare.

Reporting changes too. Demand Gen surfaces in Performance Max-adjacent reporting structures. Impression share, placement-level transparency, and frequency controls work differently than they did in standalone Display. If you're reporting GDN performance to a client or a board on those classic metrics, the numbers won't look the same post-migration even if underlying performance is identical.

The play

1. Audit your active Display campaigns now. Pull placements, exclusions, audience lists, and bidding targets from every active GDN campaign. Export them. Demand Gen's migration tool is designed to carry settings over, but documented settings you control beat assumed settings you inherit.

2. Migrate voluntarily in June. Use the migration tool as soon as it's available for your account. Self-migration preserves your ability to sanity-check the output. Auto-migration in 2027 puts Google in control of how ambiguous settings get resolved.

3. Test GDN-only channel controls against your Display baseline. Set up a Demand Gen ad group with channel controls locked to GDN, match the audience and budget to your last standalone campaign, run for two weeks or 50 conversions, and compare CPA. This tells you whether the new container changes performance before you broaden to multi-surface.

4. Build Demand Gen-native creative assets before migration. At minimum: 1:1 and 1.91:1 image crops, a 16:9 video under 60 seconds, and headlines written for Discover feed context (conversational, not banner-ad copy). Test whether adding video assets to your ad group changes CTR (click-through rate) by more than 10% versus static-only, that gap is the creative efficiency you're leaving on the table if you migrate with banners alone.

5. Set up placement exclusion lists at the Demand Gen level. GDN's placement exclusion logic applies in Demand Gen, but it has to be configured at the right level. Don't assume your existing exclusions carried over cleanly, verify them in the first week post-migration.

Watch-outs

The 9.5% ROI uplift average conceals variance. Accounts with highly refined GDN retargeting, strict placement exclusion lists, and low-funnel audience targeting may see CPA go up, not down, when those signals get pooled into a broader Demand Gen model that wants to optimize across surfaces. Watch your first-party retargeting CPAs closely in the first 30 days.

Frequency capping is less granular in Demand Gen than in standalone Display. If you're running retargeting at high frequency against a small list, the blended cross-surface environment can burn that audience faster than you expect. Set frequency caps at the ad group level on day one, not after you see the bill.

Finally: Demand Gen's reporting doesn't break out GDN placements with the same transparency classic Display did. If placement-level reporting is part of your brand safety workflow or client reporting, rebuild that workflow before migration, not after you've lost the historical comparison point.

The WhyItWon angle

Every consolidation like this creates a new creative unknown: same inventory, new algorithm, new surfaces, new asset requirements. The creative mix that was winning in standalone GDN may behave completely differently inside Demand Gen's multi-surface model, and the only way to know what works in the new environment is to test before you commit budget at scale. WhyItWon reads what's already winning in your account, what's working for competitors, and what creative signals are resonating with your customers, then scores the next ad before you spend to find out. When the campaign type underneath your budget changes this materially, that kind of pre-spend scoring isn't a nice-to-have, it's how you avoid paying for the learning curve with real CPA.