Demand Gen campaigns: complete 2026 guide

Demand Gen replaced Discovery Ads in 2024 and is now Google's primary upper/mid-funnel campaign type — running across YouTube (in-feed and Shorts), Discover, and Gmail. The major upgrades vs Discovery Ads: YouTube Shorts inventory (the most-engaged surface), lookalike-style audience modeling, AI-generated creative, and improved incrementality measurement. Existing Discovery campaigns were migrated automatically; new advertisers can only create Demand Gen.

This guide gives the 2026 playbook: what Demand Gen actually is, when to use it, how to brief creative, how to set audience signals, how to measure (and why in-platform attribution overstates contribution by 30-60%). We cover three vertical applications — B2B SaaS, e-commerce, local services — and the common mistakes that turn Demand Gen into wasted upper-funnel spend.

Updated 2026-05-08 with current creative spec, audience signal options, and measurement architecture.

What Demand Gen is (and isn't)

Demand Gen IS: Google's automated campaign type that runs visual ads (image + video) across YouTube in-feed, YouTube Shorts, Discover (Google Discover feed), and Gmail. ML-driven targeting based on audience signals + Google's user behavioral data. Uses Smart Bidding (Maximize Conversions, Max Conversion Value, tCPA, tROAS).

Demand Gen IS NOT: A direct-response channel for cold conversions on the same level as Search. Search captures existing demand (someone already searching the query); Demand Gen tries to create demand (capture intent before active search). The metrics game is different — comparing Demand Gen CPA to Search CPA is comparing apples to oranges.

Demand Gen IS NOT: YouTube TrueView in-stream (the pre-roll/mid-roll skippable ads). TrueView remains separate, optimized for different goals (CPM-priced reach vs Demand Gen's conversion-priced engagement).

Demand Gen IS NOT: Performance Max. PMax is fully automated cross-inventory; Demand Gen is upper-funnel-focused and surface-restricted. Some advertisers run both in parallel.

For PMax positioning, see our Performance Max 2026 guide.

Surfaces, creative spec, formats

Best practice: provide assets in all three aspect ratios (9:16, 1:1, 16:9). Demand Gen will combine and serve to the right surface. Withholding aspect ratios limits inventory and degrades Smart Bidding learning.

Creative volume target. 8-15 unique creative variants per asset group; refresh ~30% monthly. AI-generated assets (via Google's Asset Studio) accelerate volume; combine with human-curated brand-on creative for quality balance.

Hook patterns that work for Demand Gen video:

• Question (3-second open question that targets the ICP)
• Bold claim (counter-intuitive statistic or result)
• Problem-statement (specific pain that the product solves)
• Transformation (before/after, with-and-without, etc.)

For AI creative, see our AI creative guide.

Audience signals: how Demand Gen targets

Audience signals seed Demand Gen's targeting. They aren't hard constraints — Demand Gen extrapolates from signals using lookalike-style modeling. Quality of signal drives quality of Demand Gen results.

Signal types:

• Customer Match — your existing customers. Strongest signal; produces lookalike-quality reach.
• In-market segments — Google's high-intent audiences for active researchers.
• Affinity segments — broader interest categories.
• Custom segments — keyword/URL-based audience definitions for niche targeting.
• Detailed Demographics — life events, education, parental status (where allowed).
• Lookalike of converters — audience modeled on past converters (auto-generated).

Best practice signal mix for Demand Gen:

• 1× Customer Match (existing customers) for lookalike modeling
• 1-2× In-market or custom segment for active-research audiences
• 1× Affinity for broader awareness reach (optional; only at higher budgets)

Signals are observation by default in Demand Gen — Demand Gen extrapolates beyond the signal. To restrict to the signal, use 'targeting' mode, but expect 60-80% reach reduction.

For audience strategy, see our audiences guide and Customer Match guide.

Bidding strategies for Demand Gen

Standard 2026 progression: Maximize Conversions for first 30 days → tCPA once stable conversion volume reached → tROAS for e-commerce with strong conversion-value tracking. Don't switch strategies more than once per 21-30 days; each switch resets Smart Bidding learning.

Demand Gen vs Performance Max vs Search

Most accounts run Search + PMax + Demand Gen in parallel. Search handles existing demand; PMax scales catalog/audience-driven acquisition; Demand Gen creates new demand and reaches users not yet in active search. The three layers are complementary, not competing.

Measurement: incrementality vs in-platform

The single most important measurement principle for Demand Gen: in-platform attribution overstates Demand Gen's contribution by 30-60%, because Demand Gen surfaces capture credit for conversions that would have happened anyway. Real measurement requires incrementality testing.

Three approaches:

1. Conversion Lift Study (Google's built-in tool). Randomized test exposing some users to Demand Gen and holding out others. Reports incremental conversions vs holdout. Requires sufficient volume (typically $30k+/month spend); Google manages the methodology.

2. Brand Lift Study (Google's built-in tool). Surveys ad-exposed vs unexposed users for brand awareness, consideration, and purchase intent. Useful for upper-funnel campaigns where direct conversion attribution is weak.

3. Geo-lift testing (custom). Turn off Demand Gen in some matched regions, leave on in others; measure baseline change. Most rigorous; requires multiple geographic markets and statistical analysis. Used by enterprise advertisers for media-mix modeling.

Without incrementality testing, the safest assumption: discount in-platform Demand Gen attribution by 40%. A campaign reporting $80 CPA with Demand Gen in-platform credit likely costs $130-$140 incrementally.

Demand Gen for B2B SaaS

B2B SaaS use case: drive demo requests and free-trial signups by reaching ICP-relevant prospects on YouTube Shorts and Discover before they're actively searching.

• Audience signal: Customer Match of existing customers + relevant in-market segments (e.g. "Business Software", "Cloud Computing").
• Creative: Founder-led video, customer-story video (60-90s), product-demo Shorts (15-30s).
• Bidding: Maximize Conversions for 30 days, then tCPA at 1.5-2× Search CPA target.
• Measurement: Conversion Lift Study after 60 days; expect 35-55% in-platform overstatement.
• Budget: 10-15% of B2B SaaS Google Ads budget; secondary to Search at this funnel stage.

For B2B SaaS strategy, see our SaaS B2B strategy guide.

Demand Gen for e-commerce

E-commerce use case: drive incremental demand and remarketing reach on YouTube Shorts + Discover, especially for new product launches or seasonal campaigns.

• Audience signal: Customer Match (past purchasers) + in-market segments matching product category.
• Creative: Product-demo Shorts (lifestyle, not catalog), unboxing or transformation video, sale-promotion creative.
• Bidding: tROAS at 60-75% of Search tROAS target initially (Demand Gen has lower direct-response efficiency).
• Catalog feed integration: enable retail-feed integration to dynamically generate ads from product feed.
• Budget: 15-25% of e-commerce Google Ads budget after Search and PMax are stable.

Demand Gen for local services

Local services use case: drive awareness and lead generation in defined service areas. Useful for high-LTV services where prospects need education before searching.

• Audience signal: Customer Match + relevant local in-market segments + custom geo-targeting.
• Creative: Service-explainer video, before/after transformation, customer-testimonial Shorts.
• Bidding: tCPA at 1.5-2× Search target.
• Geo-targeting: Strict — only the unit's service area; exclude other unit territories for franchise chains.
• Budget: 10-20% of local services Google Ads budget.

Common mistakes & FAQ

Mistake 1 — Treating Demand Gen as cold direct-response. Compares unfavorably to Search at first-touch CPA; the channel is for incremental demand, not click-arbitrage.

Mistake 2 — No creative refresh. Demand Gen creative fatigues fast on Shorts; CTRs drop 30-50% after 4-6 weeks without refresh.

Mistake 3 — Single aspect ratio assets. Limits inventory access; provide 9:16 + 1:1 + 16:9 always.

Mistake 4 — Audience signal too narrow. Restricting Demand Gen to a single 1k-user audience starves the algorithm; use signals as seeds, not hard targeting.

Mistake 5 — No incrementality measurement. Trusting in-platform CPA leads to over-investment; measure lift after 60-90 days.

Mistake 6 — Skipping Demand Gen because of low direct ROAS. Many advertisers cut Demand Gen after 30 days because direct ROAS underperforms — missing the upper-funnel demand it creates that converts via Search 30-60 days later. Always measure cross-channel attribution.

For complementary reading, see our Discovery Ads incrementality (legacy), our Display vs Meta Ads, and our YouTube Shorts vs TikTok. To audit how Demand Gen fits into your overall account, run our free audit.

Sources

Official sources consulted for this guide:

• blog.google
• support.google.com
• developers.google.com
• support.google.com

FAQ