Google expanded Demand Gen with new audience signals, creative formats, and cross-channel reporting. Here’s what it means for your ad strategy.
Demand Gen in 2026: Why Google Is Pushing It Harder
Demand Gen is no longer the quiet campaign type advertisers ignored. Two years after it replaced Discovery campaigns, Google has spent the back half of 2025 and the opening weeks of 2026 layering on new audience controls, expanded creative formats, and reporting that finally lets you see across the channels it touches. The direction is unmistakable: Google wants the budget that has been flowing to Meta, TikTok, and YouTube buys placed outside Google Ads, and Demand Gen is the vehicle it is building to win it back.
That matters because Demand Gen is not a search product wearing a new badge. It reaches people across YouTube, including Shorts, the Discover feed, and Gmail, on a single visual-first canvas, and it leans on lookalike-style segments rather than keywords. If you have only ever run Search and Performance Max, this is the part of the account that behaves least like everything else you know, and the recent updates widen that gap rather than narrowing it.
This piece walks through what actually changed, what is documented versus what is still platform direction, and how we think about Demand Gen when deciding whether it belongs in a given account at all. The honest answer for many small advertisers is still not yet, but the threshold for that yes has moved, and it is worth understanding why.
From Discovery to Demand Gen: What the Migration Actually Changed
It helps to remember where this came from. Discovery campaigns ran image and carousel ads across the Discover feed, Gmail, and the YouTube home feed, optimized for conversions, and gave advertisers very little control or visibility. When Google force-migrated every Discovery campaign into Demand Gen, it was not a rename. The new format added video as a first-class asset type, brought the full reach of YouTube including in-stream and Shorts inventory, and introduced audience tools borrowed from the upper-funnel video world.
The practical consequence is that Demand Gen sits in a genuinely different place in the funnel than Discovery did, and certainly than Search does. Search captures people who are already raising their hand. Demand Gen creates the hand-raise. It puts a visual message in front of audiences who were not looking for you, on surfaces people scroll for entertainment and distraction rather than answers. That is closer to how a media planner thinks about paid social than how a performance buyer thinks about keywords.
The 2026 updates double down on this identity. Google has been expanding the creative surface area, refining the audience signals that feed it, and connecting the reporting so that the cross-channel nature of the campaign is something you can measure rather than guess at. None of that makes Demand Gen a bottom-funnel machine. It makes it a more controllable, more measurable version of the demand-creation tool it was always meant to be.
New Audience Signals: Lookalikes, First-Party Lists, and How to Seed Them
The headline change advertisers feel first is on the audience side. Demand Gen builds lookalike segments from your own first-party lists, and the quality of those segments is only as good as the seed you give them. Upload a list of your best customers, your highest-value leads, or your repeat purchasers, and Google models audiences that resemble them across its surfaces. Upload a thin or generic list and you get a thin, generic reach.
This is where the recent emphasis on audience controls pays off. You can pair lookalike segments with your standard audiences, custom segments built from search and browsing behavior, and your remarketing pools, then let Smart Bidding decide where to spend within that frame. The platform’s direction is clearly toward giving the algorithm a rich first-party signal and letting it expand outward, rather than asking you to hand-build narrow targets. That trades control for reach, which is the right trade for a demand-creation channel and the wrong one for a capture channel.
The practical guidance has not changed even as the controls have grown. Feed Demand Gen real customer-match lists, and keep them fresh with a standing sync rather than a one-time upload, because lookalikes decay as your customer base evolves. Segment your seeds by value where you can, so the model resembles your good customers and not merely your customers. And resist the urge to over-narrow. The whole point of this campaign type is to find well-matched people who have not found you yet, and a target so tight it only reaches your existing audience is a remarketing campaign wearing the wrong label.
Visual-First Creative: The Asset Requirements That Make or Break It
Demand Gen lives or dies on creative in a way Search never does. There are no headlines doing the heavy lifting against a query. There is an image or a video competing for attention in a feed full of content people actively chose to look at, which means the bar is the bar set by organic posts and influencer clips, not by other text ads.
Google’s asset requirements reflect that. The format wants multiple high-quality images across the standard aspect ratios, landscape, square, and portrait, so it can render natively on every surface from a Gmail promotion tab to a vertical Shorts slot. It supports video, and the platform’s clear direction is to reward advertisers who bring it, because short vertical video is where the cheap, abundant inventory sits and where Shorts reach is growing fastest. You also supply logos and several headline and description variations, and Google assembles and tests combinations across placements.
The trap is treating these like Search assets, generated quickly and left alone. A Demand Gen campaign starved of good creative will underperform no matter how clean the targeting, because the targeting only decides who sees the ad, and the creative decides whether anything happens next. If you have any video capability at all, lead with short vertical video. If you do not, invest in genuinely strong still imagery designed for a feed rather than repurposed banner art, and refresh it on a cadence, because creative fatigue is real and arrives faster on entertainment surfaces. This is the channel where having creative muscle, or a partner who does, separates the accounts that get value from the ones that quietly waste budget.
Where Demand Gen Fits Next to Performance Max and Search
The question we hear most is how Demand Gen relates to Performance Max, since both reach across multiple Google surfaces and both lean on automation. The cleanest way to hold the distinction is by intent and control. Search captures existing demand against keywords. Performance Max chases conversions across every surface Google has, including Search and Shopping inventory, optimizing toward your conversion goal with very little channel-level steering. Demand Gen deliberately stays out of Search and Shopping and focuses on the visual, social-style surfaces, with creative and audience controls Performance Max does not expose.
That makes them complements, not substitutes. Performance Max is conversion-hungry and will harvest whatever is cheapest to convert, which often means it leans on bottom-funnel and brand demand. Demand Gen is built to create new demand among cold but well-matched audiences and feed the remarketing pools that Search and Performance Max later convert. Run them together and Demand Gen does the introduction while the others close.
The sequencing we follow is unchanged by the recent updates: get Search efficient first, because it captures the demand that already exists and is the cheapest place to start; build a remarketing layer so warmed audiences are not lost; then add Demand Gen as the demand-creation tier once that foundation is profitable. Adding it first, before Search is dialed in, is the classic mistake. You end up paying to introduce yourself to people while leaving money on the table from people already looking for you. Demand Gen earns its place on top of a working account, not in place of one.
Lead Gen Versus E-Commerce: Who Should Actually Test It
Demand Gen is not equally suited to every business, and the recent updates have not changed that calculus so much as raised the ceiling for the businesses it already fits. E-commerce and visual-product brands are the natural home. A product that photographs and films well, an offer that reads in a single scroll, and a price point that supports an impulse-adjacent decision are exactly what these surfaces reward. If you have a catalog and creative, Demand Gen is one of the more credible ways to find new buyers on Google outside of Shopping.
Lead generation is more conditional, and the conditions are where most accounts get it wrong. A lead-gen Demand Gen campaign measured on last-click cost per lead against your Search campaigns will almost always look worse, because it is reaching people earlier, before they would have searched. Judged on its real job, introducing your offer to qualified cold audiences and building the pool that converts later, it can be a reasonable source of attention. But it needs a longer measurement window, a willingness to credit assisted conversions, and an upstream conversion goal so the algorithm is optimizing toward something the channel can actually deliver.
Our rule of thumb is honest about the floor. A local service business on two or three thousand dollars a month, with no video and a long phone-driven sales cycle, can skip Demand Gen without guilt and put that budget into Search and a remarketing layer instead. A visual brand, a business with creative to spare, or a lead-gen advertiser with enough volume to measure mid-funnel impact and a real plan to credit it, has a genuine case to test. The updates make that test more measurable. They do not lower the volume and creative thresholds that decide whether the test is worth running.
Offline Conversion Import: Steering Demand Gen Toward Quality, Not Quantity
The single most underused lever for Demand Gen in lead gen is the same one that transforms the rest of a Google Ads account: offline conversion import. When Smart Bidding only sees raw form fills, it optimizes for the cheapest form fills, and on a top-of-funnel demand-creation channel reaching cold audiences, the cheapest form fills are often the lowest-quality leads in the account. You are paying to teach the algorithm to find more of your worst prospects.
The fix is to push stage-two events back into Google Ads from your CRM, qualified leads and closed deals rather than initial submissions, so the bidding optimizes toward outcomes that represent revenue. This matters more on Demand Gen than almost anywhere else precisely because the audience is colder and the raw-lead-quality gap is wider. Feed it real qualification data and the campaign tilts toward the cold prospects who actually become customers; feed it form fills alone and it drifts toward volume that looks good in the interface and bad in your bank account.
The broader measurement stack supports this. Enhanced conversions improve the match rate so more of these events are observed in the first place, consent mode keeps the picture honest where privacy rules apply, and customer-match lists do double duty as both the seed for lookalikes and the audience signal that points the campaign at the right people. Treat offline conversion import as the prerequisite for running Demand Gen in lead gen seriously, not an optimization you get to later. Without it, you are asking a demand-creation channel to police its own lead quality with no information about which leads were any good.
How to Structure a Demand Gen Test and What to Watch
If your account clears the bar, structure the test to give the channel a fair chance rather than a last-click ambush. Start with a clear demand-creation goal and a conversion objective the channel can realistically hit, often a key page view, an add to cart, or a qualified-lead event imported from your CRM, rather than the final sale you would judge Search on. Seed your audiences with real first-party customer lists so the lookalikes have something genuine to model, and layer in your standards and remarketing pools so the campaign has range without drifting into irrelevance.
On creative, bring more than you think you need and bring video if you can. Supply images across all the aspect ratios so the format renders natively everywhere it serves, lead with short vertical video for the Shorts and in-stream inventory, and plan a refresh cadence from day one because creative fatigue arrives quickly on entertainment surfaces. Give the campaign room to learn, both budget and time, since a starved demand-creation campaign never accumulates enough signal to optimize.
Then watch the right things. Use the cross-channel reporting to see how spend distributes across YouTube, Shorts, Discover, and Gmail, and check that it is not piling into one thin-performing placement. Watch assisted conversions and view-through behavior, not just last-click, because last-click will systematically undersell a top-of-funnel channel. Track whether your remarketing and Search volume lift while Demand Gen runs, which is often where the real return shows up. And measure on blended cost per acquired customer over a sensible window rather than the in-platform CPA, because the platform report flatters the platform. Demand Gen rewards advertisers who judge it by the job it is doing, and punishes the ones who hold it to a standard it was never built to meet. Used that way, it is a genuinely useful third tier on top of a healthy account, which is exactly how we deploy it for the clients at SearchPod who are ready for it.
Want help implementing this?
Get a free proposal for your industry updates setup. We’ll show you exactly where the opportunities are.
Get Free ProposalRelated Articles