Demand Gen campaigns for e-commerce: when they earn their budget

Demand Gen campaigns for e-commerce: when they earn their budget

Demand Gen campaigns are Google's answer to social advertising: image, video and product ads on YouTube, Discover, Gmail and the Google Display Network, bought on audiences rather than keywords. For an e-commerce brand that has already captured most of its search demand, they look like the obvious next step.

Many accounts already have one, whether they planned it or not. Google moved Video action campaigns into Demand Gen during 2025, and from June 2026 it began offering a tool to move Display campaigns there too. So the question for most marketing directors is not whether to try Demand Gen, but whether the budget it already takes is earning its keep.

This article covers where Demand Gen fits next to Search, Shopping and Performance Max, how we set it up for e-commerce, and how to tell whether it is adding sales or only claiming them.

What Demand Gen campaigns can do for the business

Search and Shopping capture demand that already exists. Sooner or later, growth means reaching people before they search, and that is the job Demand Gen is built for. It gives you visual, scrolling placements where people discover products, which suits categories bought on looks or inspiration, such as fashion, home and travel.

The cost of getting it wrong is easy to miss. Demand Gen reaches people who were not searching, so many of the conversions it reports would have happened anyway. If a campaign mostly reaches existing customers and people who were already in the market, its platform ROAS can look healthy while it adds little to total sales.

The benefit of getting it right is new customers you would not have reached through search, at a cost you can measure. That is worth having, but only if you measure it properly.

Where it fits next to Search, Shopping and Performance Max

Google's own comparison is a useful starting point. It says Performance Max focuses on finding converting customers across all of Google's channels, while Demand Gen gives more control over creative testing, audience targeting and placements on YouTube and Discover.

Our rule of thumb is simple. Search and Shopping come first, because they meet existing intent. Demand Gen gets budget when non-brand Search and Shopping are close to the point where extra spend stops paying back, which we check through impression share lost to budget and the marginal return on recent budget increases. It also earns a place when the business needs a controlled test of a new audience, product range or creative idea, which Performance Max does not easily allow.

We also stop Demand Gen and Performance Max chasing the same people. If both campaigns reach recent site visitors, they compete for the same sale and each claims it in its own report. Giving Demand Gen prospecting audiences, with customer lists and recent visitors excluded, keeps the two jobs apart and makes each one easier to judge.

How we set up Demand Gen campaigns for e-commerce

The set-up decisions matter more in Demand Gen than in Search, because there is no keyword to anchor the campaign. These are the steps we follow.

  1. Give each campaign one job. We separate prospecting for new customers from remarketing, and keep product launches and seasonal pushes in their own campaigns. Google's customer acquisition goal supports Demand Gen in its new customer value and new customer only modes, using your customer lists and conversion tags to tell new buyers from existing ones.
  2. Connect the product feed. Demand Gen can show products from Merchant Center in image and products, video and products, or products only formats. Google recommends at least four products and ideally 50, square images and item_group_id for variants. We filter products by custom label, so prospecting spend goes on products with margin and stock, which is why feed quality comes first.
  3. Build audiences from first-party data. We seed Lookalike segments with high-value customers rather than every purchaser, and exclude existing customers from prospecting. Google is moving Lookalikes in Demand Gen from a hard similarity limit (2.5%, 5% or 10% of the target location) to a signal during 2026, and offers an opt-out to keep them as a constraint. We opt out when we need a clean test of a defined audience.
  4. Make creative for every placement. Google's benchmark is three vertical, three square and three horizontal images, plus a video in each orientation, with vertical video needed for YouTube Shorts. Feed-only ads look like a catalogue, so we pair product feeds with short lifestyle or demonstration video made for the first few seconds.
  5. Set budgets the system can learn from. Google recommends a daily budget of at least 10 times target CPA, waiting for 50 conversions before judging performance, and moving bids and budgets by no more than 15% a week. We usually start on Maximise conversions or conversion value and add targets once there is enough data.
  6. Choose channels on purpose. Demand Gen lets you prioritise channels and choose placements, such as YouTube Shorts only. Since Display inventory now sits inside Demand Gen, we check where spend lands and separate placements when one surface is taking budget without results.

What changes by sector

Fashion brands tend to get the most from Demand Gen, because the product is visual and new ranges give a steady reason to reach new people. Our fashion e-commerce playbook covers how this sits with organic and Shopping activity.

Travel is different. Consideration windows are long and the first touch can come weeks before a booking, so we allow longer conversion windows and judge tests over a full planning cycle. The travel and hospitality playbook sets out how we time activity around booking seasons.

How to measure Demand Gen without trusting view-through claims

Demand Gen reporting gives the campaign generous credit. Google treats Demand Gen as an exception to its usual rule that view-through conversions are for reporting only, and offers bidding that optimises for them, which is off by default. Its Platform Comparable columns add credit for views and engaged views of video to match how social platforms report, though Google says they are not used for bidding and should not be used to compare Demand Gen with Search or Performance Max.

A view-through conversion is a claim, not proof. Someone saw an ad and later bought, but most of those people may have bought anyway. So we judge Demand Gen on incrementality testing: showing ads to one group, holding them back from another, and comparing total sales.

Google's Conversion Lift does this inside Google Ads, with a holdback group of 1% to 50% and a recommended minimum of 14 days. It is not available in every account and has conversion and budget thresholds, although Google said in September 2025 that it now works at lower spend. Where it is not available, we run a geographic test, running Demand Gen in matched regions and holding it off in others, then compare sales in your own order data.

  • Incremental revenue and gross profit per pound spent, from the latest lift or geo test.
  • Cost per incremental new customer, using your own order data rather than platform conversions.
  • Share of Demand Gen conversions from new versus returning customers.
  • Total online sales in test and control regions, not only the conversions the platform reports.

What to report upwards

Leadership should see one line for Demand Gen: what the last test showed it added in sales and new customers, at what cost, and how that compares with the marginal return on more Search or Shopping spend. If no test has been run, say so, and treat the platform numbers as unproven. For how this fits with wider measurement, see our article on marketing mix modelling, attribution and incrementality.

When Demand Gen does not fit, and the mistakes we see

Demand Gen is the wrong next step when Search and Shopping still have profitable room to grow, when there is no capacity to make video and images regularly, or when the account cannot reach the conversion volume Google asks for before judging results. In those cases the budget usually works harder elsewhere.

  • Judging Demand Gen on platform ROAS that includes view-through credit.
  • Prospecting campaigns that are really reaching existing customers, because nobody excluded them.
  • Changing budgets and targets every few days, so the campaign never settles.
  • Running product feed ads with no video or lifestyle images.
  • Leaving Lookalike behaviour on the default without knowing it changed.

Questions to ask your team or agency

  • What share of our Demand Gen conversions are view-through, and how much are new customers?
  • Have we run a lift or geo test, and what incremental revenue did it show?
  • Are existing customers excluded from prospecting campaigns?
  • Which products does Demand Gen promote, and are they the ones with margin and stock?
  • Did any of our Video action or Display campaigns move into Demand Gen, and what changed when they did?

How we help

We plan and test Demand Gen as part of our PPC management, with incrementality built into the plan from the start. Our e-commerce visibility playbook shows where it sits in the wider search and visibility mix.

Frequently asked questions

What are Demand Gen campaigns in Google Ads?

Demand Gen campaigns are a Google Ads campaign type that shows image, video, carousel and product ads on YouTube, including Shorts, Discover, Gmail and the Google Display Network. They are targeted by audience, such as Lookalike segments, customer lists and interests, rather than by keyword. They suit brands that want to reach new customers before they search, and they can use a Merchant Center product feed.

Should e-commerce brands use Demand Gen or Performance Max?

They do different jobs. Performance Max spreads budget across all Google channels with a high level of automation and is usually the main product campaign alongside Shopping. Demand Gen gives more control over audiences, creative tests and placements on YouTube and Discover. We use Performance Max to capture demand and Demand Gen to test and build new audiences once Search and Shopping are close to their profitable limits.

How much budget do Demand Gen campaigns need?

Google recommends a daily budget of at least 10 times your target CPA when you use target-based bidding, and says not to judge performance until a campaign has 50 conversions. It also advises changing bids and budgets by no more than 15% a week. In practice this rules out very small tests, so plan a budget that can reach those levels within a few weeks.

How do you measure whether Demand Gen is incremental?

Compare people or regions that saw the ads with those that did not. Google's Conversion Lift holds back a share of users, from 1% to 50%, and Google recommends running it for at least 14 days, although it is not available in every account. A geographic test, running ads in some regions and not others and comparing total sales, works when Conversion Lift is not available.

Did Demand Gen replace Video action campaigns?

Yes. Google stopped the creation of new Video action campaigns in April 2025 and began upgrading existing ones to Demand Gen from mid-2025, with the last due by April 2026. From June 2026 Google also began offering a migration tool to move Display campaigns into Demand Gen, and says new campaigns will later only be created there.

Sources

Let’s create your future growth plan

Get My Growth Plan
More Insights

More articles we
think you’ll like