WHERE DO GOOGLE DEMAND GEN ADS SHOW? THE REAL SPEND-BY-SURFACE PICTURE

Every time someone asks me where do google demand gen ads show, they want a tidy list of surfaces. The list is the easy part. What actually matters, and what nobody puts in the setup screen, is how unevenly your budget gets distributed across those surfaces once the campaign goes live. Google tells you the inventory spans YouTube, Discover, Gmail and more, which is true, but it does not tell you that one surface will quietly eat most of your impressions while the placement that converts best sits in a corner with a fraction of the spend. This article is about that gap, read by inference from real distribution data. If you are new to the format, start with the broader overview of how Demand Gen campaigns work and come back here for the surface-level reality.

The full map: every surface demand gen ads show on

Let me lay out the complete inventory, because the answer to where do google demand gen ads show is broader than most people assume. Your ads serve across YouTube in its many forms, the Discover feed, Gmail, and the Google Display Network. Within YouTube alone that means in-stream (pre-roll and mid-roll inside videos), in-feed placements, the Home feed, Watch Next recommendations, YouTube Search results, and Shorts, the vertical full-screen format that now drives an enormous share of attention.

Outside YouTube, Discover is the personalised content feed on the Google app and the Android home screen, where ads blend in with the articles a user is scrolling. Gmail places your ads in the Social and Promotions tabs. The Google Display Network extends reach across third-party sites and apps. According to Google's official documentation on Demand Gen campaigns, the combined reach runs to roughly three billion monthly active users, with Shorts alone averaging over 50 billion daily views. That scale is real. What it hides is the distribution of demand gen placements google ads actually serves, which is where the practical decisions get made.

Real demand gen placements google ads spend distribution by surface

Here is the pattern I see in account after account once the campaign has run long enough to read. YouTube surfaces, taken together, dominate. In a typical mixed-creative Demand Gen campaign without channel restrictions, YouTube tends to absorb somewhere between 60 and 75 percent of impressions, and Shorts is usually the single largest slice inside that. Discover comes next, often in the 15 to 25 percent range, and Gmail trails as a thin band, frequently under 10 percent. The Google Display Network inventory, where enabled, fills whatever the algorithm decides it needs to.

That distribution is not a reflection of where your audience is most valuable. It reflects where impressions are cheapest and most abundant. Shorts has effectively infinite, low-cost inventory, so when you hand Google an open budget and a Maximize Clicks or broad conversion goal, it pours volume into the surface that absorbs it fastest. The numbers feel efficient because the CPMs are low, and that low cost is exactly what lures advertisers into reading the wrong story. The same instinct you need when reading any automated system applies here: the difference between Demand Gen and Performance Max is partly about placement transparency, and Demand Gen at least lets you segment and see the split rather than serving everything into one black box.

The Shorts trap: cheap impressions are not cheap conversions

The single most common mistake I watch advertisers make with Demand Gen is reading the Shorts spend share as a success signal. Shorts is cheap, everywhere, and ravenous for inventory, so it will happily take the lion's share of your budget and report a flattering CPM. The trap is treating that low cost as efficiency. A low cost per thousand impressions on a surface where almost nobody is in a buying mindset is not efficiency, it is volume looking for a purpose.

When I segment by surface and put spend share next to conversion share, the gap is often stark. I have seen Shorts consume well over half the budget while delivering a minority of the conversions, and a small Discover or Gmail slice punch far above its spend weight. That is the inversion nobody expects: the surface burning the budget is frequently not the one earning it back. The system optimises toward the cheapest path to your stated goal, so if your goal signal is weak, the cheapest path is just the highest-volume surface, not the most profitable one.

None of this means Shorts is bad inventory. For the right vertical creative and an awareness-led objective, it is excellent. It means you cannot accept the default distribution as a verdict on value. The cheap surface always wins the spend contest by default. Whether it wins the conversion contest is a separate question you must ask deliberately, because Google will not volunteer the answer.

Why the converting placement is almost never the one you expect

Once you read these campaigns by surface, a counter-intuitive truth settles in: the placement that converts is rarely the one with the most spend attached. The high-volume, low-cost surfaces win impressions precisely because they are low-intent, and low intent is what makes them cheap. The surfaces that take a smaller share, Discover with its browsing-but-curious users, Gmail's Promotions tab with a captive if distracted audience, often carry users a step closer to acting.

I think of it as an intent gradient across the surfaces. Someone half-watching a Short between two videos is furthest from a purchase. Someone who taps an ad in their Discover feed has shown a flicker of active interest. Someone who opens a Gmail Promotions ad has, in a small way, leaned in. The spend flows toward the bottom of that gradient because it is cheapest there, but the conversions cluster toward the top. That is why the converting placement surprises people: they expect value to follow volume, and in Demand Gen it routinely runs the other way.

This is why surface-level reporting matters more here than in almost any other campaign type. You cannot infer value from where the spend went, you have to measure conversions where they actually landed. The accounts that get Demand Gen right treat the spend report and the conversion report as two different maps, and pay close attention to where the two disagree.

Channel controls: steering where demand gen placements serve

Since March 2025, Google has given Demand Gen genuine channel controls, and they change how you manage the surface question. You can now choose where ads appear across YouTube, Discover, Gmail and the Google Display Network, and you can restrict a vertical creator video to YouTube Shorts only when that is the right native experience. This is a real improvement over the early days, when the distribution was almost entirely Google's call. Picking the right format for each surface also means matching the correct Demand Gen asset specifications to the placements you serve on, because a creative built for one ratio gets cropped into nonsense on another.

The catch is that channel controls operate at the channel level, not the individual placement level. You can turn Gmail on or off, but you cannot fine-tune the precise mix of Shorts versus Home feed inside YouTube. So I use them as a blunt instrument with intent. At launch I usually leave surfaces broadly open, because I want the unrestricted spend data first, it tells me what the algorithm thinks and where the cheap inventory is. Then, once the campaign has cleared its learning period and I can read the spend-versus-conversion split, I use the controls to prune the surfaces burning budget without returning it. Turning a channel off on day one, before you have data, is just guessing with a switch.

Reading the distribution by inference, before you trust it

The discipline that separates a profitable Demand Gen campaign from a wasteful one is patience with the data. Surface-level numbers are noisy early, because the campaign has not yet decided where it belongs and the conversion volume per surface is too thin to read with confidence. Judging a placement in the first few days is the most expensive impatience I know, because you will often pause the exact surface that was about to become your best converter once it had enough signal. So I infer in stages: let the campaign run unrestricted long enough to clear the learning period, then segment by network and lay spend share beside conversion share, surface by surface. Any surface where conversion share badly trails spend share is a candidate for restriction; any surface where it punches above its spend is a candidate for more budget. This is the same loop behind running Demand Gen campaigns well overall: observe, infer, adjust, and let the next round of data confirm the call.

Tagging helps the inference enormously. When I set landing pages up so Shorts traffic reads separately from Gmail and Discover traffic in analytics, the on-site behaviour tells me things the Google Ads interface alone never will, bounce, dwell, and downstream action all differ sharply by surface. That secondary read is frequently what confirms the suspicion the spend report first raised: the loud surface was not the valuable one.

Turning the surface picture into a pruning decision

Knowing where demand gen ads show is only useful if it changes what you do next. I turn the spend-versus-conversion map into a rhythm rather than a one-off audit. Every couple of weeks I revisit the surface split and make one of three calls per surface: leave it open because it earns its share, restrict it because it bleeds budget without converting, or lean into it because it over-delivers on a thin slice of spend. That last case, the under-funded over-performer, is the most valuable one to catch, and it is invisible unless you read conversion share rather than impression share.

I am cautious about cutting surfaces too hard, though. Demand Gen's algorithm uses cross-surface signal, and aggressively shutting channels down can starve it of the data it uses to find your audience elsewhere. So I prune with a scalpel: restrict the persistent budget-wasters, keep borderline surfaces open a little longer, and re-test the ones I cut after a creative refresh, because a new vertical asset can make a previously weak surface viable again.

So the honest answer to where do google demand gen ads show is: across more surfaces than the setup screen suggests, in a distribution you do not control by default, with spend pooling on the cheapest inventory and conversions hiding somewhere else. Read the two maps separately, respect the learning period, prune by conversion share, and accept that the placement quietly earning its keep is almost never the one shouting loudest in the spend column.