Google Demand Gen for B2B: Test Paid Reach Beyond LinkedIn

Google quietly turned its most social-feeling ad product into a serious top-funnel option for B2B — and most teams that live inside LinkedIn haven't noticed. Demand Gen, the campaign type that replaced Discovery Ads, now ships with the two things B2B performance marketers always demanded before they would test a new channel: control over where ads run, and a way to pour first-party data into the targeting. If you have been treating "paid social beyond LinkedIn" as a someday experiment, the setup friction that justified the delay is mostly gone.

What actually changed

Demand Gen launched in 2023 and fully replaced Discovery Ads by March 2024, serving across YouTube (including Shorts), Discover, Gmail, Maps, and now the Google Display Network (Google Ads Help). In July 2025 it also absorbed Video Action Campaigns, making it the single home for visual, engagement-driven advertising across Google's surfaces (groas.com 2026 guide).

The unlock for B2B is channel controls, which rolled out as a global beta starting March 2025. You can now precisely choose where ads appear — YouTube, YouTube Shorts, Discover, Gmail — instead of accepting blended cross-channel delivery (Google Ads & Commerce Blog, Jan 2025). That single change converts Demand Gen from a black box into something you can structure as a clean experiment.

The second unlock is audience flexibility. Unlike Performance Max, which offers no hard targeting, Demand Gen supports Lookalike segments and audience customization layered on your own seed lists (Google Ads Help). Pair that with Customer Match — which lets you use online and offline first-party data to reach people across Search, Gmail, YouTube, and Display — and you have a credible way to model net-new accounts off your existing pipeline (Google Ads Customer Match best practices).

Why it matters for B2B

The reflex objection is reach quality: Google's visual surfaces feel consumer-first. But the scale is real and the intent layer is improvable. Demand Gen can reach up to 3 billion monthly active users, and adding Display inventory extends it to over 90% of the global internet population across more than 3 million sites and apps (Google Ads & Commerce Blog, Jan 2025). For B2B, the point is not raw reach — it is feeding that reach with your signal so the algorithm models buyers who look like your closed-won accounts, not generic in-market consumers.

There is also a performance argument Google is leaning on hard. Based on a Nielsen analysis, Demand Gen delivers on average 58% higher ROAS than the Video Action Campaigns it replaced (Google Ads & Commerce Blog, Jan 2025). And per Google's own internal analysis (Global, Apr 2024–Dec 2025), advertisers that adopted at least 3 of the 4 Demand Gen best practices — audiences, bid and budget, creative, and data strength — saw on average over 40% more conversions (Demand Gen February Drop). Treat those as vendor-reported directional numbers, not a guarantee for your account — but they tell you where Google is pointing the optimization.

This is the same logic behind feeding warehouse data into ad platforms and treating YouTube as a demand-capture surface, not a brand afterthought.

Who should test this — and who shouldn't

This is for B2B performance marketers and demand gen leaders who are over-indexed on LinkedIn and want incremental reach without abandoning measurement discipline. It is a strong fit if you already run Search profitably, have a Customer Match list of meaningful size, and own creative that works in motion.

It is a poor fit if you expect bottom-funnel efficiency on day one. Demand Gen is top of funnel — budget, benchmarks, and patience all need to reflect that. If your only success metric is last-click cost per MQL, you will kill the test before it can prove incremental reach over your existing channels.

What to test next

Run it as a controlled experiment, not an always-on line item:

1. Isolate the channel. Use channel controls to start with YouTube in-feed plus Discover only, holding Shorts back as a separate cell. This keeps creative-format and placement effects from blurring together.
2. Seed with first-party data. Upload a fresh Customer Match list of closed-won and high-fit accounts, then build Lookalike segments from it. Stale lists limit what Google can model, so refresh on a schedule (Customer Match best practices).
3. Set top-funnel bids. Use tCPA or value-based bidding aligned to a mid-funnel action (demo request, high-intent content), not a raw lead, and size budget with Performance Planner (Demand Gen February Drop).
4. Hold the channel constant for the learning period. Don't change placements or audiences mid-flight, or you can't read the result against your intent-signal baseline.

Metrics and failure signals

Watch incrementality, not vanity reach. The metrics that prove this worked: net-new qualified accounts touched, view-through and assisted pipeline influence, and Conversion Lift or Brand Lift where you have access. The metrics that flag a problem: cheap clicks with no downstream pipeline, audience overlap with your existing Search and LinkedIn buyers, and a cost-per-qualified-account that drifts above your blended target. If you see volume without qualified pipeline after the learning period, the channel mix or seed data is wrong — not the platform. Demand Gen finally gives B2B teams the controls to find out which.