What is Google Ads Demand Gen?
Demand Gen is Google's visual, creative-led campaign type built to reach people before they're actively searching. It runs exclusively on YouTube (including Home, Watch Next, Shorts, and in-stream), Google Discover, Gmail, and Maps. It replaced Discovery campaigns in 2023 and has since picked up capabilities Discovery never had, including video ad support, Lookalike audience segments, built-in A/B creative testing, and, as of the January 2026 update, Shoppable connected-TV placements on YouTube's living-room experience.
Where Demand Gen stands apart from most other Google Ads campaign types is control. You choose which of the four surfaces your ads run on, build audiences from custom intent signals or your own first-party lists, exclude specific YouTube channels or Gmail placements, and layer in either conversion-based or value-based Smart Bidding once you have enough data.
What is Google Ads Performance Max?
Performance Max (PMax) is Google's fully automated, full-funnel campaign type. From a single campaign, it can serve across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps, and it's currently the only campaign type eligible to appear inside Google's AI Overviews. You supply creative assets, a conversion goal, a budget, and Google's machine-learning models decide which channel, placement, and audience to show each ad to, in real time, to hit that goal.
PMax has historically been criticized as a "black box," but 2026 has brought the biggest transparency upgrade since launch: campaign-level negative keywords (up to 10,000), brand exclusions, channel-level reporting, landing page performance reports, and asset-level strength scores are all now available. It still leans heavily on automation: you're guiding the model, not hand-picking placements the way you can in Demand Gen or Search.
Demand Gen vs Performance Max at a glance
Control over coverage
- Runs on YouTube, Discover, Gmail, Maps only
- Manual audience building: custom segments, Lookalikes, remarketing
- Placement, channel, and video exclusions available
- Best for upper-to-mid funnel awareness and consideration
Coverage over control
- Runs on Search, Shopping, Display, YouTube, Discover, Gmail, Maps
- Audience signals guide, but don't restrict, the algorithm
- Negative keywords, URL and brand exclusions now available
- Best for bottom-funnel conversions and revenue at scale
| Attribute | Demand Gen | Performance Max |
|---|---|---|
| Funnel stage | Upper to mid funnel | Full funnel, conversion-focused |
| Inventory | YouTube, Discover, Gmail, Maps | Search, Shopping, Display, YouTube, Discover, Gmail, Maps |
| Manual control | Audiences, placements, creative format | Budget, goal, assets, exclusions only |
| Ideal minimum data | Works with smaller conversion volumes | 30–50+ conversions per month recommended |
| Exclusive features | Lookalike segments, Shoppable CTV | AI Overviews eligibility, Shopping ads |
| Creative formats | Image, carousel, video (1:1, 16:9, 9:16) | Text, image, video, product feed assets |
| Reporting depth | Placement and audience-level | Channel-level as of 2026, still less granular than Search |
Demand Gen: pros and cons
Pros
- Real control over which of the four surfaces your budget reaches
- Lookalike segments are exclusive to Demand Gen, available nowhere else in Google Ads
- Placement and channel exclusions protect brand safety
- Supports both conversion- and value-based Smart Bidding as data accumulates
- New Shoppable CTV format opens the living-room screen for commerce
Cons
- Results live or die on creative quality, since there's no keyword layer to fall back on
- Smaller total inventory than Performance Max, which caps volume potential
- Lookalikes are shifting from a strict audience to an AI "signal" through 2026
- Segments need a meaningful seed list before they perform well
- Attribution often needs view-through and assisted-conversion context, not last-click CPA alone
Performance Max: pros and cons
Pros
- Single campaign reaches every major Google surface, including Search and Shopping
- Only campaign type currently eligible for AI Overviews placements
- 2026 updates added real negative keyword and brand controls (up to 10,000 negatives)
- Channel-level and landing-page reporting finally show where budget is going
- Built-in asset A/B testing lets you validate creative changes with control groups
Cons
- Needs meaningful conversion volume, usually 30 to 50+ per month, to optimize well
- Still less transparent than a dedicated Search or Shopping campaign
- Can compete with and cannibalize existing Search campaigns if not managed carefully
- Over-adding negative keywords too early can choke the algorithm's ability to find new converters
- Needs a 2–3 week learning phase before tight CPA or ROAS targets should be applied
Performance Max vs Demand Gen: cost per conversion by industry
Google doesn't publish official cost-per-conversion data split by both industry and campaign type, so no chart claiming exact figures for both would be honest. What we can show responsibly: each industry's overall 2026 benchmark cost per conversion, which these days is driven mostly by Performance Max and Search, plotted against Demand Gen using the real cost gap documented between the two campaign types. Read the second line as a modeled estimate, not a published benchmark.
Performance Max figures reflect 2026 all-channel industry benchmarks (WordStream/LocaliQ, SearchLab), which today are weighted heavily toward Performance Max and Search since those carry the bulk of spend. Demand Gen figures are a modeled estimate built from the real cost gap reported between the two campaign types, most concretely in Piperocket's 2026 B2B SaaS data, where Performance Max averaged $25 cost per lead against $315 for Demand Gen in the same accounts. That gap exists because Demand Gen reaches people earlier in the funnel, before purchase intent exists, so a higher cost per last-click conversion is expected and often still worth it for the pipeline and awareness it builds. Treat the Demand Gen line as directional, not a guarantee for any specific account.
How to optimize Demand Gen properly
Demand Gen rewards accounts that treat it as a creative and audience program, not a "set it and forget it" automation layer. These are the practices actually moving performance in 2026.
Layer your audiences by warmth
Start with remarketing and customer match lists to prove the funnel converts, then add Lookalike segments seeded from your best customers, then custom intent segments to reach cold, in-market users at scale.
Build creative in all three aspect ratios
Upload square (1:1), landscape (1.91:1 image / 16:9 video), and vertical (4:5 image / 9:16 video) assets. A single 16:9 cut leaves Shorts placements awkward or invisible.
Seed Lookalikes properly
Google's minimum is a small seed list, but segments built from a few thousand engaged users consistently expand more accurately than bare-minimum lists.
Move to value-based bidding once you have data
Conversion-based Smart Bidding is fine to start, but switching to value-based bidding once volume supports it lets the algorithm prioritize your higher-value customers rather than treating every conversion equally.
Prune placements regularly
Review and exclude underperforming YouTube channels, apps, or Gmail placements on a recurring cadence rather than only at launch.
Decide your Lookalike-transition stance
Google is phasing Lookalike segments from a hard targeting constraint into an AI-guided signal through 2026. If precise, narrow prospecting matters to your strategy, plan for the opt-out option as it rolls out; otherwise let the AI-signal version run and monitor reach and conversion quality.
Feed your product data for ecommerce
Connecting Merchant Center lets Demand Gen serve shoppable formats, including the new CTV placements, directly from your existing product feed.
How to optimize Performance Max properly
2026's transparency updates changed what "good" PMax management looks like. The goal is no longer to fight the black box: it's to feed the model better inputs and use the new controls without over-correcting.
Get conversion tracking and feed quality right first
Every optimization downstream depends on accurate conversion data and a clean, complete product feed. Fix this before touching bids or exclusions.
Launch on Maximize Conversions, not a strict target
Let the campaign gather two to three weeks of data before layering in a Target CPA or Target ROAS. Applying tight targets too early limits what the model can learn.
Add negative keywords without over-sculpting
Campaign-level negative keywords (up to 10,000) and brand exclusions stop irrelevant and internally-competing traffic. Add them based on real search query review, not preemptively, or you risk turning PMax into a worse version of Search.
Exclude low-value landing pages
Use URL exclusions under Asset Optimization to keep budget off blog posts, career pages, and other informational content that rarely converts, unless you've tested and confirmed it assists conversions via remarketing.
Force new-customer prospecting when needed
Apply first-party audience exclusions alongside the High-Value New Customer bidding mode so the algorithm actively looks for net-new buyers instead of re-serving your existing customer base.
Expand your search themes
Search themes now support up to 50 per asset group, up from 25. Use the extra room to describe intent more precisely, especially in categories your feed or landing pages don't fully capture.
Run structured asset A/B tests
Keep a stable set of common assets as your baseline and test one meaningful creative variable at a time in the treatment group, such as lifestyle photography versus Shorts-style video, rather than changing several variables at once.
Review channel and placement reports weekly
The 2026 reporting overhaul finally shows spend by channel and by landing page. Use it on a weekly cadence during any transition period, not just at campaign launch.
Segment campaigns by product or service line
Distinct product lines with different margins, lead times, or intent, like a monogrammed product versus a standard SKU, behave differently and generally deserve their own PMax campaign structure rather than being lumped together.
Which one should you run, and how much budget goes where?
For most accounts, this isn't an either/or decision. Google's own recommended 2026 account structure, sometimes called the "Power Pack," runs Performance Max, Search (including AI Max), and Demand Gen together, each with a different job.
Budget allocation like this is a starting point, not a fixed rule. Accounts with strong existing Search performance and a volume ceiling tend to shift more toward Demand Gen for new-audience growth; new or low-data accounts often lean harder on Search and PMax until they have enough conversions to support upper-funnel spend.
Choose Performance Max when conversions are your primary KPI and you have the historical data to support Google's automation across every surface. Choose Demand Gen when you need to build new demand with audiences who aren't searching yet, or when you want direct control over creative and placements. Run both when you have the budget to support each, just manage the overlap on YouTube, Discover, and Gmail with exclusions and regular reporting reviews.
Frequently asked questions
What is the main difference between Demand Gen and Performance Max?
Can I run Demand Gen and Performance Max at the same time?
Which campaign type is better for lead generation versus ecommerce?
How much conversion volume do I need before launching Performance Max?
Do Demand Gen and Performance Max compete for the same auctions?
What creative assets do Demand Gen and Performance Max each require?
Is Demand Gen a good fit for B2B advertisers?
How do I stop Performance Max and Demand Gen from cannibalizing each other?
What changed with Lookalike segments in Demand Gen for 2026?
Demand Gen and Performance Max reward the accounts that manage them closely
Every practice on this page takes ongoing attention: weekly search term reviews, careful negative keyword management, creative testing, and budget shifts as each channel proves itself. Getting it wrong wastes real budget, and most in-house teams don't have the hours to stay on top of both campaign types at once. That's exactly what our PPC management team handles every day.
Vanessa Lyles is the founder of eGr0wth and a digital marketing strategist obsessed with turning clicks into customers. From SEO and Google Ads to ecommerce and conversion optimization, she helps businesses stop wasting marketing dollars and start generating measurable growth.
