Is Google Display Network being replaced by Demand Gen?

No. The GDN inventory still exists with its 2 million-plus sites, apps, and videos. What's retiring is the standalone Display campaign type. You'll manage GDN placements inside Demand Gen alongside YouTube, Discover, Gmail, and Maps. The network isn't dead—the campaign wrapper around it is.

How is Demand Gen different from traditional Display campaigns?

Two critical differences. First, Demand Gen uses Smart Bidding powered by audience signals and conversion data—you can't opt into manual bidding like you could with Display. Second, Demand Gen runs across YouTube, Discover, Gmail, Maps, and GDN, not just GDN placements. The bidding model and the surface mix both change.

Does the GDN-only toggle in Demand Gen work like old Display campaigns?

No. The GDN-only toggle is a placement filter, not a bidding-model filter. It restricts where your ads appear to GDN inventory, but Demand Gen still uses Smart Bidding based on audience signals and conversion data. The auction dynamics are fundamentally different from standalone Display. Flipping this toggle and expecting old Display performance to replicate will disappoint you.

How do you measure Demand Gen performance without last-click attribution?

Run incrementality tests. Geo-lift testing compares total conversions across all channels in exposed regions vs. matched control regions. Holdout audience testing splits your remarketing pool and compares conversion rates between the ad-exposed group and the held-back group. Both methods measure whether Demand Gen drove conversions that wouldn't have happened otherwise—which last-click attribution structurally cannot detect for upper-funnel spend.

What is the 14-day relearn window when migrating Display to Demand Gen?

Elevarus observed that performance metrics—particularly cost-per-lead for lead-to-sale funnels—take roughly 14 days to stabilize after migrating Display campaigns into Demand Gen. This is agency-observed, not Google-documented. The instability comes from Smart Bidding recalibrating to a new campaign structure, audience signal set, and placement mix. Don't make optimization decisions during this window.

How should I structure a full-funnel Demand Gen campaign sequence?

Three stages. Stage 1 (awareness): 40-50% of budget, broad demographics + interest signals, problem-awareness creative, all surfaces, relaxed CPA targets. Stage 2 (consideration): 30-35% of budget, custom intent + Stage 1 engagement audiences, product-feature creative, moderate CPA targets. Stage 3 (conversion): 15-20% of budget, remarketing lists, direct-offer creative, aggressive CPA targets. Each stage feeds audience signals into the next. Don't blend cold and hot audiences in one campaign—Smart Bidding can't optimize for awareness reach and conversion efficiency simultaneously.

Google Demand Gen 2026: Full-Funnel Architecture Guide

Most advertisers are about to waste their GDN budgets by treating Demand Gen like Display 2.0. The GDN-only toggle doesn't recreate old Display bidding. Smart Bidding runs regardless. Last-click attribution will declare the channel dead before it scales. And the migration forcing this recklessness is already underway — Google retired standalone Display campaigns on May 26, 2026, absorbing 2M+ sites, apps, and videos into Demand Gen through 2027.

Headlines say "Display is dead." Wrong. The Google Display Network still exists. What's dead is the campaign type you've used for 15 years to run it.

This isn't a cosmetic change. Demand Gen doesn't just change where you click "create campaign." It changes how bidding works, how audiences are targeted, and how you should measure results. Treating it like Display 2.0 is the fastest way to waste budget.

I've managed over $50M in ad spend across Meta and Google. I've run incrementality tests that proved platform-reported ROAS regularly overstates true incrementality by 20-40%. Most advertisers are about to misuse Demand Gen in exactly the same way they misused GDN — measuring it with last-click attribution, running it without audience sequencing, and expecting it to perform like a direct-response channel when it's built for demand creation.

What actually changed. What most marketers are getting wrong. And how to build a Demand Gen setup that drives real conversions — not just cheap impressions.

Google Killed the Display Campaign Type (Not the Network)

On May 26, 2026, Google announced that standalone Display campaigns are retiring. GDN inventory stays accessible, but you'll manage it inside Demand Gen alongside YouTube, Discover, Gmail, and Google Maps placements (Google, May 2026). Search Engine Journal confirmed the same day that this affects campaign structure, management, and optimization across the next year (SEJ, May 2026). PPC Land reported the migration tool rollout is phased through 2027 (PPC Land, June 2026).

The network isn't gone. The campaign wrapper around it is.

This sounds like a UI change. It isn't.

Demand Gen Is Not Display 2.0

Old Display campaigns let you choose manual bidding, set placement exclusions, and target audiences the way you wanted. Demand Gen uses Smart Bidding powered by audience signals regardless of which placements you select. You don't get to opt out of that.

The GDN-Only Toggle Misconception

Google added a "GDN inventory only" setting inside Demand Gen. Most advertisers see this and think: great, I can recreate my old Display campaigns.

No. The toggle is a placement filter. It limits where your ads appear. It does not change the bidding model. Demand Gen still uses Smart Bidding based on audience signals and conversion data even when you restrict placements to GDN only.

Elevarus flagged this clearly: the GDN-only setting filters placements, not bidding behavior (Elevarus, May 2026). Your ads still compete in a Smart Bidding auction optimized for feed and short-form video engagement metrics. The auction dynamics are fundamentally different from what you had in standalone Display.

If you flip the GDN-only toggle and expect your old Display performance to replicate, you'll be disappointed. Elevarus also observed a roughly 14-day relearn window for lead-to-sale metrics before cost-per-lead stabilizes after migration (Elevarus, May 2026). This is agency-observed, not Google-documented, but it aligns with what Smart Bidding needs when campaign structure changes.

Smart Bidding Changes Everything

In standalone Display, you could bid manually on impressions or clicks. You could target broad audiences and pay for reach. In Demand Gen, Smart Bidding optimizes toward your conversion goal using audience signals, historical conversion data, and real-time contextual signals.

Three consequences:

Your conversion tracking setup directly affects how Demand Gen bids. If your conversions are poorly defined or attributed via last-click, Smart Bidding optimizes toward last-click conversions. Disaster for upper-funnel spend.
Audience signals matter more than placement selection. Demand Gen decides who sees your ad based on signal quality, not just which website they're visiting.
You can't separate "who" from "where" the way you could in Display. The system ties them together.

Google reports that advertisers using Demand Gen see an average ROI increase of 9.5%, with GoFood achieving a 24% CPA reduction and 19% conversion increase (Google, May 2026). These are Google-selected case studies, not independent benchmarks. But they indicate what the system optimizes toward when you feed it good data.

Last-Click Attribution Will Murder Your Demand Gen Budget

Most marketers measure Demand Gen with last-click attribution. They look at the ROAS column in Google Ads and decide the campaign is underperforming because last-click ROAS looks terrible for upper-funnel placements.

Last-click attribution over-credits branded search and undervalues every channel that creates demand. I've seen this firsthand running incrementality and geo-lift tests on Meta and Google campaigns. Platform ROAS regularly overstates true incrementality by 20-40%. The gap is even wider for upper-funnel channels because last-click assigns zero credit to the ad that made someone aware of your product.

Demand Gen is a demand creation tool. It runs on YouTube, Discover, Gmail, and GDN placements where people aren't searching for your product. They're watching videos, reading articles, checking email. Your ad interrupts them and creates awareness.

If you measure that interruption with last-click attribution, you'll conclude it doesn't work. You'll cut the budget. You'll never build the demand that your Search and PMax campaigns later convert.

This is exactly what happened with GDN for years. Marketers ran Display campaigns, saw terrible last-click ROAS, killed the budget, and then wondered why Search volume kept declining. They were destroying the demand pipeline and measuring the destruction with a metric designed to ignore it.

Kill last-click attribution for Demand Gen. Use incrementality testing instead. I've written about this in depth in my guide to incrementality testing for paid media — read that if you need the full framework. Applied to Demand Gen specifically:

Building a Full-Funnel Demand Gen Architecture

Stop running one Demand Gen campaign with a blended audience and hoping Smart Bidding figures it out. Build a sequence.

Stage 1: Cold Audience Awareness

Campaign goal: reach new people who don't know your brand exists.

Bid strategy: Maximize clicks or Target CPA with a relaxed target — set 2-3x your normal CPA to give the system room to learn.

Audiences: Broad demographic + interest signals. No custom intent or remarketing lists.

Creative: Video (15-30 second Shorts and in-stream) + image carousels for Discover and Gmail. Focus on problem-awareness messaging. Don't pitch the product. Pitch the problem.

Placements: All surfaces. Don't restrict to GDN only. YouTube and Discover deliver better awareness performance than GDN-only placements.

Budget allocation: 40-50% of total Demand Gen spend.

This stage creates the demand pool. Measure it with reach, frequency, and brand lift — not last-click ROAS. Run an incrementality test after 4-6 weeks to validate that exposed regions show higher downstream Search and PMax conversion volume.

Stage 2: Warm Audience Consideration

Campaign goal: move people who saw your awareness ads toward active interest.

Bid strategy: Target CPA or Target ROAS with moderate targets.

Audiences: Custom intent segments + website visitors (last 30 days) + YouTube viewers from Stage 1 campaigns. Build audience lists from Stage 1 engagement signals.

Creative: Longer video (60s) showing product features, comparison content, customer stories. Image ads with stronger product messaging.

Placements: All surfaces. The system will naturally shift spend toward higher-engagement placements based on your audience signals.

Budget allocation: 30-35% of total Demand Gen spend.

Measure with engagement rate, click-through rate, and downstream conversion lift. Run incrementality tests comparing consideration-exposed audiences against cold-exposed-only audiences to see if the consideration layer actually moves people closer to purchase.

Stage 3: Hot Audience Conversion

Campaign goal: convert people who engaged with Stage 1 and Stage 2 but haven't purchased.

Bid strategy: Target CPA or Target ROAS with aggressive targets aligned to your actual conversion goals.

Audiences: Remarketing lists — website visitors who viewed product pages, cart abandoners, YouTube subscribers, people who engaged with Stage 2 ads. Layer purchase intent signals.

Creative: Direct offer messaging. Pricing, promotions, urgency triggers, product-specific calls to action.

Placements: All surfaces. Smart Bidding will concentrate spend on placements where your hot audience is active.

Budget allocation: 15-20% of total Demand Gen spend.

This is the only stage where last-click ROAS gives you a remotely useful signal. Even here, incrementality testing is more accurate. But if you must use platform attribution for quick optimization decisions, the conversion stage is where it's least misleading.

The Sequencing Logic

Each stage feeds audience signals into the next. Stage 1 creates the pool. Stage 2 filters for engagement. Stage 3 converts the engaged.

Without this sequence, you're dumping cold and hot audiences into one campaign and letting Smart Bidding guess who to reach. Smart Bidding is good, but it can't simultaneously optimize for awareness reach and conversion efficiency in the same campaign. The objectives conflict. Separate them.

How to Actually Measure Demand Gen

I've covered the full incrementality framework in my post on incrementality testing for paid media. Applied to Demand Gen specifically:

Geo-Lift Testing for Demand Gen

Pick 5-10 geographic regions as test groups. Pick matched control regions with similar population, income, and historical conversion rates. Run Demand Gen in test regions. Hold back in control regions. Measure the difference in total conversions — not just attributed conversions — across all channels: Search, PMax, Direct, Organic.

If test regions show 15% more total conversions than control regions, that's your incremental lift. Divide incremental conversions by Demand Gen spend to get true incremental CPA. Compare that against what Google Ads reports as attributed CPA. The gap will surprise you.

I've run these tests across Meta and Google campaigns managing over $50M in spend. Platform attribution consistently overstates incrementality by 20-40%. The overstatement is worst for upper-funnel channels like Demand Gen awareness campaigns.

Holdout Audience Testing

For consideration and conversion stages, create audience holdouts. Split your remarketing pool. Serve Demand Gen ads to half. Hold back from the other half. Compare conversion rates between the two groups over 14-30 days.

This is simpler than geo-lift but only works for remarketing audiences large enough to split meaningfully. You need at least 10,000 people per group for statistical significance.

What to Do with the Results

If incrementality testing shows positive lift, scale the stage that drives the most incremental conversions. Feed the incrementality data back into Smart Bidding by adjusting CPA/ROAS targets to reflect true incremental cost, not attributed cost.

If testing shows zero or negative lift, kill the campaign. Don't rationalize. Don't adjust creative. Don't "give it more time." If the ad didn't create demand that wouldn't have existed otherwise, the spend is waste.

Where Demand Gen Fits in the 2026 Google Ads Stack

Demand Gen isn't an isolated campaign type. It's the demand creation layer that feeds your conversion-focused campaigns.

The 2026 Google Ads stack for a full-funnel paid media strategy:

Demand Gen — creates demand across YouTube, Discover, Gmail, GDN, and Maps. Upper and mid-funnel. Measured by incrementality.
Google AI Max for Search — captures demand from people searching after seeing your Demand Gen ads. I cover this in my post on Google AI Max for Search as the DSA replacement that automates search coverage for queries you'd miss with manual keywords.
Performance Max — converts demand across all Google surfaces using your strongest creative and audience signals. Lower funnel. I've written about controlling PMax in my guide to Google Performance Max — the asset group and audience signal architecture matters.

The flow: Demand Gen creates awareness. AI Max for Search captures the resulting search demand. PMax converts it.

If you run PMax without Demand Gen feeding it demand, you're relying on existing brand awareness and organic search volume. That works until it doesn't. Search volume declines when upper-funnel spend stops. PMax efficiency drops. And you wonder why ROAS is falling while you "optimized" by cutting Display.

Migration Prep: What to Do Before Your Display Campaigns Force-Migrate

The migration is phased through 2027. You have time, but not unlimited time. Set up now.

Audit your current Display campaigns. Document which campaigns use manual bidding vs. automated bidding. Manual bidding campaigns will behave differently under Smart Bidding. Flag them for close monitoring during migration.

Clean up your conversion tracking. Demand Gen's Smart Bidding relies on conversion data quality. If you're tracking micro-conversions that don't correlate with revenue, Smart Bidding will optimize toward cheap micro-conversions. Define conversions that represent real business value.

Build your audience lists. Demand Gen audience targeting is signal-based, not placement-based. Start building custom segments, customer match lists, and website visitor audiences now. The richer your audience signals, the better Smart Bidding performs from day one.

Plan your 14-day relearn window. Based on Elevarus's observed pattern (May 2026), expect roughly 14 days of unstable performance metrics after migrating Display campaigns into Demand Gen. Don't make optimization decisions during this window. Let the system learn.

Set up incrementality testing before you scale. Don't wait until you've spent $50k on Demand Gen to discover last-click ROAS looks bad. Run a geo-lift test from the start. You'll have real data within 4-6 weeks.

Don't flip the GDN-only toggle and call it done. The toggle restricts placements but doesn't recreate old Display campaign dynamics. Use it only if you have a specific reason to exclude YouTube, Discover, Gmail, and Maps — not as a default that makes Demand Gen "feel like Display."

The migration is happening whether you're ready or not. You'll either use Demand Gen as a demand creation engine — or waste it measuring last-click ROAS on a channel designed to create demand that last-click can't see.

Frequently Asked Questions

Is Google Display Network being replaced by Demand Gen?
No. The GDN inventory still exists with its 2 million-plus sites, apps, and videos. What's retiring is the standalone Display campaign type. You'll manage GDN placements inside Demand Gen alongside YouTube, Discover, Gmail, and Maps. The network isn't dead—the campaign wrapper around it is.
How is Demand Gen different from traditional Display campaigns?
Two critical differences. First, Demand Gen uses Smart Bidding powered by audience signals and conversion data—you can't opt into manual bidding like you could with Display. Second, Demand Gen runs across YouTube, Discover, Gmail, Maps, and GDN, not just GDN placements. The bidding model and the surface mix both change.
Does the GDN-only toggle in Demand Gen work like old Display campaigns?
No. The GDN-only toggle is a placement filter, not a bidding-model filter. It restricts where your ads appear to GDN inventory, but Demand Gen still uses Smart Bidding based on audience signals and conversion data. The auction dynamics are fundamentally different from standalone Display. Flipping this toggle and expecting old Display performance to replicate will disappoint you.
How do you measure Demand Gen performance without last-click attribution?
Run incrementality tests. Geo-lift testing compares total conversions across all channels in exposed regions vs. matched control regions. Holdout audience testing splits your remarketing pool and compares conversion rates between the ad-exposed group and the held-back group. Both methods measure whether Demand Gen drove conversions that wouldn't have happened otherwise—which last-click attribution structurally cannot detect for upper-funnel spend.
What is the 14-day relearn window when migrating Display to Demand Gen?
Elevarus observed that performance metrics—particularly cost-per-lead for lead-to-sale funnels—take roughly 14 days to stabilize after migrating Display campaigns into Demand Gen. This is agency-observed, not Google-documented. The instability comes from Smart Bidding recalibrating to a new campaign structure, audience signal set, and placement mix. Don't make optimization decisions during this window.
How should I structure a full-funnel Demand Gen campaign sequence?
Three stages. Stage 1 (awareness): 40-50% of budget, broad demographics + interest signals, problem-awareness creative, all surfaces, relaxed CPA targets. Stage 2 (consideration): 30-35% of budget, custom intent + Stage 1 engagement audiences, product-feature creative, moderate CPA targets. Stage 3 (conversion): 15-20% of budget, remarketing lists, direct-offer creative, aggressive CPA targets. Each stage feeds audience signals into the next. Don't blend cold and hot audiences in one campaign—Smart Bidding can't optimize for awareness reach and conversion efficiency simultaneously.

Google Demand Gen 2026: Display Ads Are Gone—Here's How to Build Full-Funnel Demand That Actually Converts