What Demand Gen Actually Is
Demand Gen Google Ads is the campaign type that replaced Discovery. According to Google’s own help documentation, it serves ads on YouTube (including Shorts), Discover, Gmail and the Google Display Network, in image, video, carousel and product-feed formats. One campaign, one set of assets, cut for feeds rather than for results pages.
Two things make it different from the rest of the account. The first is that nobody asked. A Demand Gen ad interrupts somebody watching a video or reading their inbox, which means it behaves much more like paid social than like Search. The second is audience. Demand Gen is where Google offers lookalike segments, built from your own first-party data, so you can reach people who resemble your customers rather than people who typed a keyword.
It is also getting bigger. Google is folding standalone Display campaigns into Demand Gen: the migration tool opened to eligible advertisers in June 2026, and campaigns not moved by hand are being migrated automatically through 2027. If your account still has a Display campaign, it is going to become a Demand Gen campaign whether or not you plan for it. Better to plan for it.
| Search | Demand Gen | |
|---|---|---|
| Who starts it | The buyer, by searching | You, by interrupting a feed |
| Who it reaches | The few already in market | The many who will be |
| Where it runs | Google results pages | YouTube, Shorts, Discover, Gmail, Display |
| What decides results | Keywords, match types, negatives | Audience seed and creative |
| What it should be judged on | Cost per qualified opportunity | Lift: branded search, direct traffic, pipeline over quarters |
Where It Fits in a B2B Account
Most B2B buyers are not in market most of the time. A company replaces its accounting platform, its ERP or its agency every few years, and in the quarters between, nothing you bid on in Search will reach them because they are not searching. Search catches the narrow slice that is. Demand Gen is one of the few paid tools Google gives you for the rest.
That is exactly how we scope it on our own Google Ads work. For a B2B company with a long sales cycle, the plan we publish is Search on high-intent terms only, plus YouTube and Demand Gen for the long middle where nobody is searching yet. The two jobs are different, so they get different budgets, different measures and different patience.
Three signs an account is ready for Demand Gen Google Ads:
- Search is already tidy. Conversion tracking verified, a negative list maintained, the search terms report read weekly. Demand Gen spends money creating searches; if the Search campaign leaks, you are paying twice to lose the buyer you created.
- You have a point of view worth repeating. A clear problem you want to be known for. Without one, Demand Gen has nothing to say except your logo.
- You have creative, or the means to make it. We will come back to this, because it decides more than the budget does.
If none of those is true yet, the honest answer is that the first dollar belongs in Search and the tracking under it. A Demand Gen campaign launched on top of a broken account produces a lovely reach number and nothing you can act on.
Seed It With the Accounts You Want
The audience is where B2B teams can get far more out of Demand Gen than a consumer brand can, because they usually know exactly who they want to reach. The tool for it is the lookalike segment.
Google builds a lookalike from a seed list of your first-party data: a Customer Match list, or people who have interacted with your website, app or YouTube channel. It then finds similar users in your target location at one of three reach levels, as Google describes them: narrow aims for about 2.5% of users there, balanced (the default) about 5% and broad about 10%. The segments refresh every one to two days.
The quality of the seed is the whole game. A list of everyone who ever visited your site teaches the model to find people who visit sites. A list of closed-won customers, or of the contacts at opportunities that reached a sales stage, teaches it something worth learning. For B2B, three seeds tend to work:
- Closed-won customers from the CRM, as a Customer Match upload. The strongest signal you have, and usually the smallest list.
- Contacts at open opportunities, if the list is big enough. These are the people inside accounts that already look like buyers.
- High-intent site visitors: pricing page, case studies, the demo page. Not the blog, which is full of students, job seekers and competitors.
Start narrow. A broad lookalike in a B2B market reaches a lot of people who resemble your customers only in that they own a phone.
One change to watch. Google has said the default behaviour of Demand Gen lookalikes will evolve toward a suggestion mode, where the seed acts more as a signal and the system may reach beyond the reach level you chose. For a consumer brand that may be fine. For a B2B team running Demand Gen Google Ads against a defined buyer profile it can quietly widen the audience, and Google provides an opt-out form to keep lookalikes working as a strict targeting constraint. Decide which you want on purpose rather than discovering it in the placement report.
If you sell to a named list of accounts rather than a broad market, treat Demand Gen as one layer of an account-based marketing programme rather than the whole of it. Google cannot target a company the way LinkedIn can. What it can do is put video in front of people who look like that company’s buying committee, at a fraction of the cost per impression, while LinkedIn does the precise work.
The Creative Is the Targeting
In Search, the keyword does the selecting: whoever typed it is the audience. In a feed there is no keyword, so the ad has to do the selecting itself. The first two seconds of the video, the headline on the image, the problem it names: those decide who stops and who scrolls past. That is why weak creative is where Demand Gen leaks most, and it is why most Google-only agencies quietly avoid the campaign type. It is a visual, interruption-based channel, and the rules that govern paid social apply.
What good Demand Gen creative looks like for a B2B advertiser:
- It names the problem in the first line or the first two seconds. “Your month-end close is two days late because of three spreadsheets” selects a finance lead. “Innovative solutions for modern teams” selects nobody.
- It comes from a person. A founder or a practitioner explaining something on camera is remembered in a way a logo animation is not.
- It is cut for the placement. Vertical for Shorts, square or landscape for in-feed, a still with a strong headline for Gmail and Discover. One horizontal video stretched across all of them looks like what it is.
- It is a set, not a single ad. Several angles on the same problem, rotated, so the people you reach see the idea more than once without seeing the same ad until they hate it.
- It asks for attention, not a demo. Most of the people who see it are not ready to talk to sales. An ad that demands a meeting filters for the handful who are, and wastes the rest.
This is the clearest overlap between our Google work and our creative production. Our Demand Gen Google Ads campaigns are fed from our own shoot days, which is the main reason we treat it as a real channel rather than an afterthought. If you cannot produce for it, you are better off not running it than running stock footage with your logo on the end.
Choose the Conversion Before the Bid
Here is the decision that sinks most B2B Demand Gen Google Ads campaigns, and it is made before anyone looks at a bid strategy. What are you telling Google a success looks like?
Google’s bidding works from conversions, and it needs enough of them to learn. A B2B company that books six demos a month cannot feed a conversion-based strategy with demos alone. Google’s own Demand Gen guidance recognises this: where full conversions are scarce, it suggests optimising toward lighter events such as a site visit. And for a target CPA strategy, Google recommends a daily budget of at least 10 times that target.
The trap is in the lighter event. Tell the system that a page view is a success and it will find you page views: the cheapest, least interested visitors on the network, at a cost per conversion that looks wonderful in the report. The campaign is succeeding at the thing you asked for and failing at the thing you needed.
A better ladder for B2B:
- Start on reach or clicks while the account has no Demand Gen history, and judge the first weeks on who you reached and what they did next, not on cost per lead.
- Define a quality engagement event that a tyre-kicker rarely triggers: a pricing page view after a case study, a video watched to the end, a return visit within 14 days. Make it the conversion the campaign optimises toward.
- Import offline conversions from the CRM, so that qualified opportunities and closed deals flow back into Google Ads against the click or view that started them. This is part of the conversion tracking we set up before a dollar is spent on any account, and for B2B it is what turns Demand Gen from a guess into a system.
- Move to value-based bidding only once those imported conversions are arriving in enough volume to learn from.
Two more settings worth deciding on purpose. Demand Gen offers channel controls, so you can prioritise Google’s own channels or the Display Network rather than letting the campaign drift into whichever inventory is cheapest. And brand safety: check where the ads actually ran in the first fortnight, and exclude what does not belong next to your name.
Prove It Without a Last Click
Demand Gen Google Ads will almost always lose a last-click comparison with Search, and it should. Its job is to make somebody search for you next quarter. When they do, the conversion is credited to the Search ad they clicked, and the campaign that created the search shows nothing.
Judge it on that basis and it gets switched off every time. So agree in advance what it will be judged on, and report it separately from Search. The measures we use for the creation side of demand generation were chosen because none of them depends on a click:
- Branded search volume over time. The most direct evidence that more people know your name.
- Direct traffic to key pages. People who typed you in rather than clicking anything.
- Self-reported attribution. A free-text “How did you hear about us?” field. Buyers will mention the YouTube video that no pixel connected to the deal.
- Sales acceptance rate and win rate by source. Buyers who already understand the problem arrive better qualified.
Google adds its own tools on top. Its Demand Gen documentation points to brand lift, search lift and conversion lift studies to judge whether the spend is paying off, and supports data-driven attribution. Lift studies are the closest thing to a controlled experiment you will get from the platform, and Brand Lift is available on an allowlist basis, so ask about access before you need it.
Sources: Google Ads Help, “About Demand Gen campaigns” and “Use Lookalike segments to grow your audience”. The timing is the one published on our demand generation service page.
And the reporting rules we hold to on every account apply here with extra force: the platform named and the attribution window stated on every view, platform-reported value labelled as such, and never summed with another platform’s number as though they were the same thing. Demand Gen also counts view-through conversions, and can count a lot of them. Read those as a signal of reach, not as revenue. If you need a hand building the reporting, our analytics and attribution work starts exactly there.
A First 90 Days That Survives Review
Demand Gen Google Ads campaigns rarely fail on the first day. They fail at the first quarterly review, when somebody puts them next to Search and asks why the cost per lead is four times higher. A launch plan that expects that review is the one that survives it.
- Before launch: verify conversion tracking, build the offline conversion import, and write down what the campaign will be judged on and when. Take a baseline of branded search and direct traffic.
- Weeks 1 to 2: launch with a narrow lookalike from your best seed, three to five creative angles cut for each placement, and channel controls set on purpose. Read the placement report at the end of week two.
- Weeks 3 to 6: drop the creative that nobody watches, make more of what holds attention, and exclude placements that do not belong next to your name. Leave the bid strategy alone; changing it every week resets the learning.
- Weeks 7 to 12: test a second seed or a wider reach level against the first. Start reading branded search and self-reported attribution against the baseline.
- Day 90: report creation and capture separately. Expect branded search and inbound quality to move before pipeline does, and say so in the report before anybody asks.
That last line is the one to agree with finance before launch, not during the review. The creation half of demand generation compounds over two to four quarters. A 90-day report should show it moving in the right direction, not paying back in full.


