Google Ads Demand Gen Campaigns: What They Actually Are, and Why Display Ads Are Being Folded Into Them

If you manage a Google Ads account, you’ve probably noticed something new sitting in the campaign creation flow: a “Demand Gen” option that keeps getting pushed higher up the list, and a prompt on older Display campaigns nudging you to migrate them. This isn’t a cosmetic UI change. Google has started folding the Google Display Network into Demand Gen, and depending on when you read this, that migration is either optional, encouraged, or already done for you. Understanding what Demand Gen actually is, how it differs from the campaign types you already run, and what the Display merger changes is now a real account-management task, not background reading.

This post covers what Demand Gen campaigns are, why Google is merging Display into them, how to decide between Demand Gen, Performance Max, and a narrower Display-only setup, and the specific budget, audience, and creative decisions that determine whether a Demand Gen campaign performs or just burns spend on impressions nobody acts on.

What a Demand Gen Campaign Actually Is

Demand Gen is Google’s campaign type for reaching people who aren’t actively searching for what you sell. It runs across YouTube (in-stream, in-feed, and Shorts), Discover, Gmail, and — as of the 2026 migration — the Google Display Network. The unifying idea is that these are all visually-driven, feed-based, or video surfaces where someone is scrolling or watching rather than typing a query with intent attached. That makes Demand Gen a top-of-funnel and mid-funnel tool: it’s built to create interest and consideration, not to catch someone who already decided to buy and is comparing options on Search.

This is the core distinction that trips people up when they first touch Demand Gen: it is not a Search replacement, and judging it by Search-campaign metrics like cost-per-click will make it look like a bad decision even when it’s doing exactly what it’s supposed to do. A Demand Gen campaign showing a video ad to someone scrolling YouTube Shorts is planting a seed. If you compare its cost-per-click, or even its cost-per-conversion, directly against a Search campaign catching someone who typed your exact product name, Demand Gen will almost always look worse, because it’s solving a different problem.

Structurally, a Demand Gen campaign looks more like a Performance Max campaign than a classic Search campaign. You set a campaign-level goal and budget, choose a bidding strategy, define one or more audience signals, and upload a pool of creative assets — images, videos, headlines, descriptions, a logo — that Google’s system assembles and tests across the eligible placements. You don’t pick individual placements or write one ad per ad group the way you would in a legacy Display campaign. The system decides where and to whom each asset combination gets shown, within the audience and budget constraints you set. That’s a meaningfully different skill set from writing manual Display ad groups, and it’s part of why so many advertisers who were comfortable with old-style Display are now finding Demand Gen unfamiliar even though the underlying inventory overlaps.

The Change Nobody Asked For: Display Ads Are Moving Into Demand Gen

In May 2026, Google announced that the Google Display Network is being folded into Demand Gen rather than staying a separate campaign type. Starting in June 2026, advertisers with eligible Display campaigns got access to an in-account migration tool to move existing campaigns voluntarily. The stated plan is that new campaigns will eventually only be creatable within Demand Gen, and any remaining eligible standalone Display campaigns get migrated automatically later.

Two things are worth being precise about here, because the announcement has been misread in both directions. First, the Display Network itself is not being shut down and your ads aren’t disappearing from Display inventory. What’s changing is the campaign wrapper: GDN placements now live inside Demand Gen campaigns alongside YouTube, Discover, and Gmail inventory, instead of sitting in their own dedicated campaign type. Second, you can still run something that behaves like a Display-only campaign — Demand Gen supports narrowing inventory to just the Display Network if that’s genuinely what you want — so this isn’t forcing every Display advertiser into a full YouTube-and-Gmail mix against their will. It’s a consolidation of the campaign type, not an elimination of the inventory.

Google’s own reporting on this migration cites an average 9.5% increase in ROI for advertisers who add GDN inventory into a Demand Gen campaign rather than running it standalone. Treat that as a directional number from the company that built and benefits from the merger, not a guarantee for your account — but it lines up with the general logic of Demand Gen: giving one bidding and creative system access to more inventory and audience overlap tends to help it find efficiency it couldn’t when Display was walled off with its own budget and targeting.

The practical question for most advertisers isn’t whether to accept this migration — it’s largely happening on Google’s timeline either way — but when to do it deliberately rather than let it happen automatically. If you have a standalone Display campaign with a long performance history, a specific placement exclusion list, or frequency caps you’ve tuned carefully, migrating on your own schedule lets you rebuild those controls inside the new structure and watch performance closely for the first couple of weeks. Waiting for an automatic migration means those settings might not carry over cleanly, and you find out something broke after the fact instead of before.

A Short Checklist Before You Migrate a Display Campaign

If you’re migrating deliberately rather than waiting for it to happen to you, a few things are worth checking before you click through the migration tool, not after:

• Export your current placement exclusion list and managed placement list — confirm they carry over, and reapply them manually inside the new Demand Gen campaign if they don’t.
• Note your existing frequency cap settings; Demand Gen’s frequency controls live in a different part of the campaign settings than legacy Display, and it’s easy to migrate and simply forget to set one.
• Pull a baseline of your last 30 days of Display performance — cost, conversions, cost per conversion — so you have something concrete to compare against once the migrated campaign has had two weeks to settle.
• Check whether your existing creative assets meet Demand Gen’s aspect-ratio requirements; static banner sets built for classic Display often only cover one or two of the ratios Demand Gen expects.
• Decide in advance whether you want the migrated campaign to stay Display-only or expand into the full YouTube, Discover, and Gmail mix, and set that scope explicitly rather than accepting whatever the migration tool defaults to.

None of this is complicated, but skipping it is exactly how a campaign that was quietly profitable for two years turns into one nobody can explain the performance of six weeks after a migration nobody planned for. The five minutes it takes to write these down before you click “migrate” is cheap insurance against a much longer troubleshooting session later, when the campaign looks different and nobody remembers exactly what changed or when.

Demand Gen vs. Performance Max vs. Plain Display: How to Actually Decide

With three campaign types now capable of touching Display inventory, the decision isn’t which one is “best” — it’s which job you’re hiring each one to do.

Performance Max is built to maximize conversions across every Google surface simultaneously, including Search, Shopping, YouTube, Display, Discover, Gmail, and Maps. It’s the right choice when conversions are your primary KPI and you’re comfortable handing Google’s automation control over which channel gets credit and budget on any given day. The tradeoff, as covered in how to audit a Performance Max campaign when you can’t see inside it, is that you lose visibility into which channel is actually driving the results you’re paying for — a PMax campaign can be quietly living off Search and Shopping traffic that would have converted anyway, while nominally also running on Display and YouTube.

Demand Gen is narrower by design. It doesn’t touch Search or Shopping inventory at all, which means it can’t cannibalize demand that was already going to convert on its own. It gives you more manual control over audience definition than PMax does — you can build and layer custom segments, not just feed the algorithm a conversion goal and hope it finds the right people — and it’s built around the assumption that you’re trying to create interest rather than capture it. If your goal is genuinely building awareness or warming up an audience before they ever search for your brand, Demand Gen is the more honest tool, because it isn’t going to blend its numbers with Search performance that was going to happen regardless.

A narrowed, Display-only Demand Gen setup is the closest thing left to the old standalone Display campaign, and it still makes sense in specific cases: retargeting a warm audience with static banner creative where video isn’t practical, running in a market or vertical where YouTube and Discover inventory underperforms for reasons specific to your audience, or simply wanting a smaller, easier-to-audit test before committing budget to the full Demand Gen mix.

A useful way to frame the decision: if your honest answer to “what is this campaign for” is “get in front of people before they’re looking,” pick Demand Gen. If it’s “convert as many of the people who are already in-market as possible, wherever they are,” Performance Max is doing that job, even if it also touches Display inventory along the way. Running both without a clear split in what each is meant to accomplish is how budgets end up competing against each other for the same audience instead of covering different parts of the funnel. This split matters slightly differently depending on the business — an ecommerce account chasing repeat purchase volume weighs this differently than a B2B account trying to build a pipeline of leads who won’t convert for months, and it’s worth deciding deliberately rather than copying whatever split a template account uses.

Budget and Bidding: What “Enough Budget” Actually Means

Demand Gen campaigns are conversion-optimized by algorithm from day one if you choose target CPA or target ROAS bidding, and Google’s own guidance is that the algorithm needs enough daily spend to gather signal quickly, or it stalls in a slow, expensive learning period. The commonly cited rule of thumb is a daily budget of at least 15 times your target CPA. If your target cost per lead is $20, that points to a daily budget in the neighborhood of $300 before the campaign has a realistic shot at exiting the learning phase in a reasonable window.

For advertisers without an established target CPA yet, Google’s general starting-point guidance for a single Demand Gen campaign is somewhere between $75 and $150 per day, with a recommendation to leave the campaign alone — no bid changes, no audience edits, no pausing underperforming assets — for at least two weeks before judging results or making structural changes. That patience requirement is the part advertisers most often ignore, and it’s usually what causes a campaign to get killed prematurely. An algorithmic bidding system that’s still exploring which audience-and-creative combinations convert will look inefficient in week one almost by definition; that’s what exploration looks like before the model has enough data to lean into what’s working.

This is the same logic that applies to choosing between target CPA, target ROAS, and Maximize Conversions in any smart-bidding context: the algorithm needs a minimum volume of conversion signal to calibrate against, and starving it of budget or yanking the strategy before it’s had time to learn tends to produce worse results than either committing properly or not running it at all. If your total Demand Gen budget can’t sustain the 15x-target-CPA threshold with room to also survive a two-week learning window, the honest move is either to raise the budget, raise the target CPA to something the budget can actually support, or hold off on Demand Gen until the numbers work.

One structural point worth flagging: Demand Gen supports both target CPA and target ROAS bidding, along with Maximize Conversions and Maximize Conversion Value for accounts that don’t want to set a hard target yet. Starting on Maximize Conversions without a target, then layering in a target once you have enough conversion history to set one intelligently, is often a more realistic path than guessing at a target CPA on day one and hoping it’s close enough not to choke delivery.

Audiences: Where You Still Get a Say

Unlike Performance Max, where audience signals are more of a suggestion the algorithm can freely override, Demand Gen gives you real, definable audience targeting: custom segments built from keywords, URLs, and apps, remarketing lists built from your own site and app data, Customer Match lists uploaded from your CRM, and Google’s own affinity and in-market audience categories. You can layer several of these into one campaign or split them across different Demand Gen campaigns to control budget allocation and compare performance between audience types directly — something Performance Max makes much harder to do cleanly.

One change worth building into your plan now: starting March 15, 2026, Lookalike segments in Demand Gen are transitioning from strict audience-matching to AI-powered “audience signals.” Practically, this means a Lookalike list stops being a hard boundary the system won’t cross and becomes more of a directional hint the algorithm uses alongside its own modeling — similar to how audience signals already function in Performance Max. If you’ve been relying on tightly-scoped Lookalike lists to keep a campaign inside a specific customer profile, expect that control to loosen, and plan to monitor who’s actually converting more closely after the transition rather than assuming the audience will stay as narrow as it used to be.

If you have a Customer Match list built from real purchase or lead data, that’s the strongest signal you can feed into a Demand Gen campaign, and it’s worth building deliberately rather than as an afterthought — the same discipline that applies to building a Customer Match list that actually works for Search and Shopping campaigns applies here too. A clean, reasonably fresh first-party list gives the algorithm something concrete to model lookalike behavior against, which matters even more now that pure Lookalike targeting is loosening into a signal rather than a rule.

Creative: What Actually Gets Rewarded

Demand Gen is an asset-pool format, not a one-ad-per-placement format, and the size and quality of that pool has a direct, measurable effect on how the campaign is scored and how much reach it gets. Google’s Ad Strength rating for Demand Gen rewards providing a comprehensive set of assets across every supported ratio rather than just the minimum required to launch.

The image specs break into four aspect ratios: 1.91:1 landscape at 1200×628, 1:1 square at 1200×1200, 4:5 vertical at 960×1200, and 9:16 vertical at 1080×1920. Horizontal images, square images, and a logo are required to launch; the vertical formats are technically optional but are the ones most worth adding, since vertical placements on Shorts and in-feed Discover are a meaningful share of Demand Gen inventory and a campaign with no vertical assets simply won’t compete well for that space. Images are capped at 5MB, logos at 150KB, and you can upload up to 20 image assets per ad — file types are JPG or PNG.

Video assets should cover landscape (16:9), square (1:1), and vertical (9:16) as well, for the same reason: providing all three prevents Google from force-cropping a single video into ratios it wasn’t shot for, which tends to look worse and perform worse than a native-fit asset. Minimum video duration is 10 seconds, but Google’s own recommendation is 15 seconds or longer for feed placements, and a tighter 6-to-15-second range specifically for Shorts inventory, where attention spans and format conventions are different.

Text elements have hard limits: headlines up to 40 characters, with at least one headline of 30 characters or fewer required so it fits cleanly in the tightest placements; descriptions up to 90 characters; and a business name up to 25 characters. None of this is exotic compared to Responsive Search Ads, but the discipline of writing genuinely different headline variations — rather than four small rewordings of the same sentence — matters more here because the system is testing combinations across a much wider range of visual contexts, from a YouTube in-stream skip screen to a Gmail promotions tab thumbnail.

The practical takeaway from Google’s own February 2026 guidance is that advertisers who implemented at least three of four specific practices — strong audience signal usage, properly sized budget and bid strategy, “Excellent” Ad Strength through a comprehensive asset set, and solid first-party data connection — saw an average of over 40% more conversions than advertisers who skipped most of them. That’s Google’s reported figure from its own advertiser base, not an independently audited study, so it’s worth treating as a strong directional signal rather than a promise, but the underlying logic holds regardless of the exact number: a thin asset pool on a system built to test combinations at scale is working against its own strengths.

Data Strength: The Boring Lever That Matters Most

Of the four practices above, data strength is the one advertisers most often shortchange, because it doesn’t feel like a campaign-level decision — it feels like a tracking-setup task that happened once, months or years ago, and hasn’t been revisited since. That’s exactly the problem. Demand Gen, like every automated bidding system, is only as good as the conversion signal it’s optimizing against, and a lot of accounts are quietly optimizing against broken or incomplete data without realizing it.

Google’s current push here is sitewide tagging through its tag gateway for advertisers, paired with connecting offline conversion sources through Data Manager, so the system has both strong on-site signal and, where relevant, closed-loop data on what happened after a lead left the website. This is the same underlying issue covered in what Enhanced Conversions actually fix, and why they’re often silently broken: a huge share of accounts have conversion tracking that looks fine on the surface — the conversion count isn’t zero, the campaign isn’t flagged — but is quietly undercounting or misattributing in ways that never get caught because nobody goes looking. Feeding an automated bidding system incomplete signal doesn’t make it cautious; it makes it confidently wrong, chasing whatever partial data it has as if that were the whole picture.

Before scaling a Demand Gen budget, it’s worth treating conversion tracking as a thing to actively verify rather than assume is fine: check that consent-mode and tag-gateway setup isn’t silently dropping a chunk of events, confirm that any offline conversions you care about — a sale that closes days after a lead form, for instance — are actually flowing back into Google Ads, and make sure Enhanced Conversions are matching at a healthy rate rather than falling back to cookie-based tracking that’s increasingly unreliable. None of this is specific to Demand Gen, but Demand Gen’s reliance on tCPA and tROAS bidding from the start makes bad data more expensive here than it is in a campaign type you’re managing more manually.

Where This Fits Alongside the Rest of an Automated Account

Demand Gen is one more entry in a growing list of Google Ads surfaces that hand meaningful control to an algorithm: Performance Max on the conversion side, and on the Search side, tools like AI Max for Search campaigns, which expands keyword matching and creative generation using AI while still leaving you some controls to constrain it. The pattern across all of these is consistent: Google is steadily reducing the number of manual levers available in exchange for systems that need less day-to-day tuning but more upfront setup discipline — the right budget, the right signals, the right asset pool, the right data — to actually perform.

That tradeoff is fine as long as you go into it clear-eyed about what you’re giving up. You’re not giving up control entirely — Demand Gen still lets you pick audiences, set a target CPA, and curate your creative pool — but you are giving up the ability to hand-pick individual placements, write one ad per audience segment, and see exactly which combination of asset and audience drove which result. That’s a real cost, not just a UI simplification, and it’s worth acknowledging rather than pretending automation is a pure upgrade with no visibility tradeoff attached. It’s also the reason a growing number of accounts are looking at tools built specifically to sit on top of that automation and explain what it’s doing — Growera, for instance, exists because reading raw performance data out of an increasingly automated account and turning it into an actual decision is a different skill than running the campaigns themselves.

Why Watching an Automated Campaign Type More Closely, Not Less, Actually Matters

There’s a natural instinct to treat automated campaign types as “set and forget” because the interface has fewer dials to turn day to day. That instinct is backwards. A Search campaign with manual bids and exact-match keywords will drift slowly if you ignore it for a month — a few keywords get expensive, a couple of ad groups underperform, nothing catastrophic. A Demand Gen or Performance Max campaign running on target CPA can drift much faster and much less visibly, because the algorithm is making dozens of small reallocation decisions a day based on signals you’re not watching in real time. If those signals degrade — a tracking issue, a landing page problem, a shift in who the audience list is actually reaching after the Lookalike-to-signals change — the campaign doesn’t necessarily throw an error. It just quietly gets less efficient while still technically “working,” and the first place that shows up is usually cost-per-result creeping upward over a week or two, not a dramatic single-day spike anyone would notice by accident.

This is really the same question raised in how often you should actually check your Google Ads account: the more decisions you hand to an algorithm, the more that daily or near-daily check becomes about catching drift early rather than making tactical changes yourself. It’s part of why Growera’s approach is built around continuous daily monitoring of the account rather than a periodic manual review — flagging when a Demand Gen campaign’s cost-per-conversion moves outside its normal range, when Ad Strength drops because an asset got disapproved, or when a tracking issue starts quietly distorting the signal an automated bidding strategy is optimizing against, catches problems while they’re still small and cheap to fix instead of after a budget has been spent chasing a broken signal for two weeks.

The Bottom Line

Demand Gen isn’t a rebrand of Display, and treating it like one — either by ignoring it because “Display already exists” or by expecting it to behave exactly like the campaign type it’s absorbing — will lead to the wrong decisions. It’s a genuinely different tool built for reaching people before they’re searching, it now includes the Display Network as one of several inventory sources rather than as its own walled-off campaign type, and it rewards the same disciplines that matter everywhere else in a modern Google Ads account: enough budget to let algorithmic bidding actually learn, real first-party data feeding the signal, a genuinely comprehensive creative asset pool rather than the bare minimum, and audience targeting that’s deliberately built rather than left on defaults.

If you’re currently running a standalone Display campaign, don’t wait for the automatic migration to find out how your settings translate — move it deliberately, rebuild the controls you cared about, and watch it closely for the first couple of weeks. If you’re deciding whether Demand Gen belongs in your account at all, the honest test is whether your goal is creating interest or capturing it that already exists; that answer, more than any feature comparison, tells you whether Demand Gen, Performance Max, or a narrower Display-only setup is the right fit for the budget you actually have. Get that split right, feed it real data, and give it the runway to learn, and Demand Gen earns its place in the account. Skip any one of those and it just becomes an expensive way to find out the algorithm needed more than you gave it.