Google Demand Gen Campaigns: Ecommerce Guide

Google Demand Gen campaigns are the most powerful mid-funnel format on Google's ad network — and most ecommerce brands are either ignoring them entirely or running them with stale Discovery-era creative that misses the point. Demand Gen places your ads across YouTube (including Shorts), Gmail, and Google Discover simultaneously, with full creative control and custom audience targeting that Performance Max strips away. For brands that have maxed out Meta and need to unlock Google's mid-funnel, this is the gap.

What Is Google Demand Gen and How It Replaced Discovery Ads

Google Demand Gen campaigns replaced Discovery Ads in 2024. If you ran Discovery campaigns before, the inventory is the same — YouTube feed, Gmail Promotions, Google Discover — but the format has been rebuilt with stronger creative options (including native YouTube Shorts placements), improved audience tools, and deeper integration with Google's AI optimization layer.

The name signals the intent. Discovery Ads were designed to surface products to passive audiences. Demand Gen is designed to create demand: to reach people who don't yet know they need your product and move them toward intent. That's a meaningfully different goal, and it requires a different creative approach.

Demand Gen combines YouTube in-feed ads, YouTube Shorts ads, Gmail sponsored ads, and Discover feed placements in a single campaign with a unified budget. Google's AI allocates spend across those placements in real time based on predicted conversion probability — but unlike Performance Max, you can see and control what creative runs where, who it targets, and how it bids.

For ecommerce brands, Demand Gen sits in the part of the funnel that's historically been hardest to own on Google: awareness and consideration. Search captures intent that already exists. Shopping closes it. Demand Gen is what builds it. According to Google Think (Jan 2026), AI-powered campaigns with high-quality creative asset libraries adapt ads to each consumer query in real time — and Demand Gen is the format that gives you the most control over what those assets are.

Where Demand Gen Ads Show Up: YouTube, Shorts, Gmail, Discover

Understanding the placements inside Demand Gen matters because the creative requirements for each are genuinely different.

YouTube In-Feed Ads appear in the YouTube homepage feed, search results feed, and Watch Next section. These are thumbnail + headline units, not pre-roll. The user chooses to engage. This format rewards strong visual curiosity — the thumbnail needs to stop the scroll, and the headline needs to be specific enough to earn the click.

YouTube Shorts Ads are vertical (9:16), full-screen, and appear between organic Shorts content. They autoplay with sound on. This is the closest Google equivalent to TikTok or Instagram Reels ad placement. The first 2–3 seconds are critical — there's no "skip" hesitation here; users swipe immediately. Native-style content dramatically outperforms polished studio video in this placement.

Gmail Sponsored Promotions appear in the Gmail Promotions tab as collapsed units that expand when clicked. These benefit from offer clarity: a clear discount, a specific product category, or a value proposition that reads well in a subject-line-like format. Gmail placement tends to have high click-through but lower purchase intent than YouTube.

Google Discover appears in the Google app and mobile Chrome new tab feed — a personalized content stream based on the user's search history and interests. Discover ads look like editorial content cards: a strong image, a clean headline, and a source label. The audience here is highly intent-signal-rich because Discover infers interests from actual search behavior.

Demand Gen vs Performance Max: When to Use Each

This is the question every ecommerce advertiser should answer before allocating Google budget. According to research by Fluency (Feb 2026), brands that retain creative control via Demand Gen outperform those relying solely on PMax for mid-funnel acquisition.

Performance Max is the right choice when you want Google to manage the full funnel — Search, Shopping, YouTube, Display — and you have a healthy product feed and conversion history. The tradeoff is control. PMax is a black box for creative performance: you can't see which headline paired with which image is driving conversions.

Demand Gen is right when you need explicit creative control, when you want to test messaging and visual concepts, or when your goal is building mid-funnel volume rather than capturing existing demand. Many mature Google advertisers run both — PMax for bottom-funnel capture, Demand Gen for mid-funnel prospecting — with distinct budgets and no shared attribution targets.

For newer brands or those expanding beyond Meta, start with Demand Gen. You'll learn more about what creative and messaging resonates than you will from a PMax campaign that auto-generates everything.

Audience Targeting in Demand Gen: Custom Segments, Lookalikes, and Intent

Audience targeting is where Demand Gen earns its competitive edge over Performance Max. You have direct control over who sees your ads, and the available targeting types are meaningfully different from what Meta offers.

Custom Segments let you target users based on their Google Search behavior — the specific keywords they've searched. This is a direct view into purchase intent. A user who searched "best running shoes for flat feet" in the past 30 days is a different prospect than someone Google categorizes broadly as "interested in fitness." Build custom segments around your high-value search terms, competitor brand names, and category searches that indicate real purchase consideration.

Google Lookalike Audiences (called "Similar Segments" in the interface) model new prospecting audiences based on your existing customer lists, website visitors, or YouTube viewers. Feed it your customer email list and Google builds a lookalike pool from users with similar behavioral and interest profiles. These audiences perform well for brands with clear customer profiles and enough seed data (500+ matching users minimum; 1,000+ for quality).

In-Market Audiences are Google's behavioral categories — users identified as actively researching a purchase in a specific category. These are broad but useful as a starting layer, particularly for brands newer to Google advertising who don't yet have first-party data to build from.

Customer Match lets you upload your customer email list and target or exclude those users directly. Useful for suppressing existing customers from prospecting campaigns, or for reaching lapsed customers with a win-back offer.

The most effective Demand Gen targeting structure: run separate ad groups for custom segments (high intent), lookalike audiences (proven profile), and in-market audiences (broad reach), each with distinct creative tuned to where that audience sits in the consideration journey.

Creative Requirements: What Works on YouTube Shorts vs Gmail vs Discover

Demand Gen campaigns require more creative investment than a standard Google Shopping campaign — and that's exactly why brands that put in the work outperform those that don't. Each placement has different creative norms.

For YouTube In-Feed: Use a striking thumbnail with clear product visibility. The headline should be specific and benefit-driven — "Why 40,000 customers switched to [Product]" outperforms "Shop our sale." Video thumbnails with a person's face or a clear product in use consistently outperform generic lifestyle shots. Keep headlines under 15 words.

For YouTube Shorts: Vertical 9:16 video, 15–30 seconds. The first 3 seconds must carry the hook — assume the user will swipe if you haven't made a case for watching by second three. UGC-style footage (creator-shot, phone-quality, native to the platform) outperforms studio-produced video by a significant margin. Show the product in real use, not on a white background. This is the same principle that drives results for brands running Meta Advantage+ Shopping campaigns with native-style creative.

For Gmail: Write the headline like a subject line — specific, curiosity-driven, offer-forward. "20% off for new customers — expires Sunday" outperforms "Discover our collection." The expanded Gmail ad should function like a landing page in miniature: product image, benefit statement, CTA button.

For Discover: High-quality lifestyle imagery that looks native to a content feed. Avoid anything that screams "ad." Clean compositions, natural lighting, a product in context. Discover users are in a passive browsing mode — intrusive creative gets swiped past; editorial-quality imagery earns the pause.

Best practice: build at least 3 creative variants for each major placement type. Google will allocate toward top performers, but you need enough variety to learn what's working before the algorithm locks in.

Bidding Strategy: Maximize Conversions vs Target CPA for Ecommerce

The bidding decision matters most in the first 30–60 days of a Demand Gen campaign.

Start with Maximize Conversions (no Target CPA set). This tells Google to spend your budget and get as many conversions as possible without a cost constraint. The algorithm needs data before it can optimize efficiently — a Target CPA set too aggressively in week one will cause Google to underspend and starve the campaign of the learning data it needs.

Once the campaign has accumulated 50+ conversions, evaluate your actual CPA and set a Target CPA at 20–30% above your observed CPA. This gives the algorithm headroom to continue learning while steering toward efficiency. Tighten the target incrementally as the campaign matures.

Avoid Target ROAS in the first 60 days of a Demand Gen campaign. ROAS targets require more conversion data to optimize reliably than CPA targets, and setting them too early causes the campaign to underspend against high-value audiences while chasing efficiency metrics it doesn't have enough data to accurately predict.

One important nuance for ecommerce: if you're using Demand Gen for awareness/consideration rather than direct conversion, consider optimizing for a micro-conversion (product page view, add-to-cart) rather than purchase. This gives the algorithm a higher-frequency signal to learn from, and you can layer in purchase conversion optimization once the campaign has established reach patterns.

How to Read Demand Gen Results Without Misleading Yourself

Demand Gen results require context that the interface doesn't always provide clearly. The three most common misreads:

Attribution overlap is real. If you're running Performance Max and Google Search campaigns alongside Demand Gen, expect significant conversion credit overlap, particularly under data-driven attribution. A user who sees a Demand Gen ad on Discover and then converts after clicking a branded Search ad will often be attributed to both. Don't read Demand Gen's reported conversions at face value without running a holdout test or incrementality study.

View-through conversions inflate numbers. By default, Demand Gen reports view-through conversions — users who saw your ad but didn't click, then converted later. These inflate your headline conversion count. Filter to click-through conversions only for a cleaner read on direct response performance, and treat view-through conversions as a supporting signal, not a primary metric.

Benchmark against Meta, not Shopping. Demand Gen occupies the same funnel position as Meta prospecting. Compare CPMs, CPCs, and cost per acquisition against your Meta top-of-funnel campaigns, not against your Google Shopping campaigns, which operate at a fundamentally different point in the purchase journey. For a deep dive on Shopping campaign strategy, see our Google Shopping ads strategy guide.

The real test of Demand Gen's impact is whether branded Search volume increases and whether assisted conversions show up in your attribution path reports. A working Demand Gen campaign creates demand that surfaces in Search — this is the signal that it's actually moving people through the funnel, not just generating cheap clicks.

How Atlas Builds and Manages Demand Gen Campaigns That Drive Revenue

At Atlas, we run Demand Gen campaigns as a mid-funnel layer in a broader Google advertising strategy — not as a standalone experiment. The structure that consistently works for our ecommerce clients:

Phase 1 (Weeks 1–4): Launch with 3 ad groups — custom segment (high-intent keywords), lookalike (built from customer list), and in-market (broad reach). Three creative variants per ad group minimum. Maximize Conversions bidding. Budget set at 15–20% of total Google spend.

Phase 2 (Weeks 5–8): Review placement-level performance. Cut underperforming placements or creatives. Introduce Target CPA at 125% of observed CPA. Begin building a second wave of creative based on what's working.

Phase 3 (Month 3+): Optimize to a stable Target CPA. Refresh creative every 4–6 weeks. Expand to new lookalike seed audiences as customer data grows. Evaluate incrementality via geo holdout or matched-market test before scaling budget significantly.

The brands we see fail at Demand Gen are almost universally making one of three mistakes: running it with Performance Max-era creative (static product images, not built for video placements), setting a Target CPA too early and starving the algorithm, or reading reported conversions without accounting for attribution overlap.

Our performance marketing team manages the full Google Ads funnel — Shopping, Performance Max, Search, and Demand Gen — as an integrated system, not as isolated campaigns competing for credit. If you're running Google Ads without a Demand Gen layer, you're leaving mid-funnel volume on the table that your competitors are capturing.

Frequently Asked Questions