WHAT IS DEMAND GEN GOOGLE ADS? NOT DISPLAY WITH A NEW NAME

The question I get most often right now is some version of "what is demand gen google ads, and is it just the old Display campaign rebranded?" I understand why people ask it that way. The surface list looks familiar, the Display Network is now folded inside it, and the screenshots resemble what they ran last year. But after migrating dozens of accounts off Display and Video Action Campaigns, I can tell you the rename framing is exactly what sets people up to fail. Demand Gen does not serve ads the way Display did, and grading it on Display's scorecard means pausing your winners and scaling your losers. This is the explanation I give clients before we touch a single migration button, built on what I have watched happen in live accounts rather than marketing copy.

What demand gen in google ads actually is

Stripped to its core, Demand Gen is one campaign type that serves visual ads across YouTube, Shorts, Discover, Gmail, and now the Google Display Network, with Google AI choosing which asset reaches which person on which surface. Per Google's official documentation on Demand Gen campaigns, these campaigns are designed to capture engagement and action across Google's most visual, browsable surfaces. That last word, action, is the part everyone skims past.

The defining trait is not the surfaces, it is the intent the campaign is engineered around. Demand Gen exists to manufacture demand among people who are not searching for you yet, by pairing audience signals with strong creative and letting the system find receptive moments. It sits at the upper and middle funnel, but unlike pure awareness buying it is built to drive conversions, not just impressions. The mental model that works is "social prospecting inside Google", not "banners across the web".

Why it is not Display with a new name

Old Display was a placement-and-bid machine. You chose sites and topics, managed CPCs, excluded the junk placements that ate budget, and judged the whole thing on cheap clicks and viewable impressions. Every lever was about where the ad showed and how much you paid. Demand Gen takes nearly all of those levers off the table. There is no placement bidding to tune in the old sense, the campaign-level exclusions you relied on largely move to account-level content suitability, and the bidding is conversion-led and AI-run rather than CPC-managed.

The confusing part is that Google has now routed Display Network inventory through Demand Gen. So you can still reach the same two-million-site network, which makes it tempting to conclude nothing changed. But that inventory is now served under Demand Gen's audience-first logic, not the legacy placement model. The same impression is decided by a different question. This is why the comparison most worth internalising is how demand gen and display differ in practice, not how they overlap. Treating the two as interchangeable is the assumption that quietly wrecks performance reads after a migration.

What actually changes in how Google serves the ads

Here is the shift that matters, and it is the heart of why the rename framing fails. On Display the serving engine answered a "where" question: which placement, which site, which auction, which bid wins this impression. On Demand Gen the engine answers a "who" question: which audience signal best matches this creative, and which of Google's surfaces is this specific person most receptive on right now. The optimisation target moved from the slot to the human.

The practical consequence is that a single asset stops behaving like a single ad. The same creative can serve as a full-screen vertical video on Shorts, a feed card on Discover, and a promo in Gmail, and the AI assembles the combination per user rather than per placement. One image effectively becomes several ads across the funnel, which is why placement-level optimisation, the discipline that defined Display management, no longer has anything to grab onto. Your job shifts from curating placements to feeding the system signal: varied creative, clean conversion data, and well-built audience signals. Because that serving model is conversion-led, the bidding strategies available in Demand Gen do far more of the heavy lifting than any manual bid you set on Display. You are not steering the ad to a place anymore, you give the model raw material and a target, and it decides placement on its own.

Why your old Display KPIs break the moment you migrate

This is where I watch good marketers panic at numbers that are completely normal. The KPIs you trusted on Display, CPC, CTR, viewable CPM, were designed to reward cheap clicks and broad cheap reach, because that is what Display optimised for. None of them measure what Demand Gen is built to produce. When you carry that dashboard over unchanged, the campaign looks broken on day three when in fact it is behaving exactly as designed.

Take CTR. On Demand Gen a low CTR routinely sits next to strong conversion volume, because the click stopped being the unit of value. The path is longer and multi-surface: someone meets you on Shorts, comes back through Discover a day later, and converts from a Gmail promo at the end of the week. Last-click CTR captures almost none of that. CPC tells a similar lie, the cost per click can rise while cost per conversion falls, and if CPC is your headline metric you will "fix" a campaign that was never broken. The honest read only appears when you measure conversions, CPA or ROAS, and assisted impact across the whole journey, the same instinct behind sound conversion tracking setup, because the campaign can only optimise toward signal it can actually see.

What I have learned migrating dozens of accounts

The single most useful thing I can tell you about migration is to use the migration tool rather than rebuilding from scratch. When you migrate, Google ports across up to 42 days of performance history, which compresses the learning period to roughly one to two days and avoids a cold start. Rebuild manually and you throw that history away, the new campaign starts blind, and you eat a week or two of noisy data while it learns who responds. In the accounts I have moved, migrated campaigns stabilise far faster and waste less learning spend, so the tool is almost always the right call.

The second lesson is patience, enforced as a rule rather than a feeling. Google's guidance is to wait around 50 conversions before making meaningful changes, and I hold clients to that hard, because the most common self-inflicted wound after a migration is reading day-two CPA, declaring failure, and dismantling the campaign before it ever exits learning. Budget continuity matters too: any spend earlier in the migration day does not carry into the new campaign, so the first 24 hours can look lumpy with temporary under- or over-delivery that means nothing. Ignore the first day and start reading once learning has actually run.

When Demand Gen is the right tool, and when it is not

Demand Gen earns its place when you have strong visual creative and a real prospecting goal: reaching qualified people who do not know you yet and nudging them toward action. If your assets are weak or text-only, the campaign starves, because the entire serving model leans on imagery and video the AI can route across surfaces. It also wants conversion data to optimise against, so a brand with no tracking and no conversion volume will struggle to give the system anything to learn from.

It is the wrong tool when you are trying to do something else. To harvest existing high-intent search demand, that is Search's job. To get maximal coverage across every Google channel with the machine steering everything, that comparison lives in how demand gen compares to performance max, and the honest answer is they solve different problems. Demand Gen gives you more control over creative testing, audience targeting, and which surfaces you serve on; Performance Max trades that control for breadth. Picking on "which is newer" rather than "which matches my objective" is how budgets get misallocated. Demand Gen's job is specifically to create and capture demand on visual surfaces.

The scorecard that matches what the system optimises for

So if Display's metrics are the wrong exam, what is the right one? Lead with conversions and conversion value, then CPA or ROAS measured against your target, and read all of it across the full multi-surface path rather than per click. The numbers I actually watch are conversion volume against spend, the trend in CPA once learning has cleared, and assisted conversions, because Demand Gen frequently proves its worth in the assist column before it shows up in last-click. Conversion lift, where you can run it, is the cleanest evidence of incremental demand the campaign genuinely created.

Then use the channel and asset reporting as your optimisation surface, since that is where real control now lives: it shows which surface and which creative are carrying the campaign, and that is what you iterate on instead of placements. Keep CTR and CPC on the dashboard if you like, but demote them to diagnostic colour, never targets in their own right. The whole argument collapses into one line: grade Demand Gen on the outcomes it is engineered to produce, not on the clicks Display taught you to count. The campaign changed the question Google asks before it serves, so the answer your reporting gives has to change with it. Do that, and the campaign that looked broken on Display's dashboard turns out to be quietly doing exactly what you hired it for.