Google has long treated Demand Gen as the home for visual ad buying outside Search and Performance Max. Now one of its product managers has put a harder number on the entry point: $100 per day, described in the October 1 Ads Decoded episode as the “bare minimum best practice” for a Demand Gen campaign.
The number that caught marketers’ attention
Alejandro Osio, product manager for Demand Gen and YouTube performance ads, used the figure almost in passing. His point was about testing volume, not pricing. A campaign holding tens of thousands of images and videos, he said, cannot give every asset a fair run; the system’s goal is to deliver performance against your objectives. Feed-driven retail campaigns were excluded because product feeds let Google’s systems work at far greater scale.
The awkward part is the comparison. Since April 1, 2026 the Google Ads API has enforced a $5 daily minimum for Demand Gen campaigns. That makes Osio’s guidance 20 times the technical floor. Other Google-linked references sit elsewhere: a Fospha programme run with Google across 25 retail ecommerce brands in late 2025 pointed to roughly $130 a day, while the 2025 move of video action buyers onto Demand Gen advised daily budgets above 20 times expected CPA.
Creative coverage first, refresh second
Osio’s framework is simple: coverage before optimization. Every Demand Gen ad should carry images and videos across horizontal, vertical and square ratios, with an ad strength rating of at least Good and preferably Excellent. That rating measures asset variety, not predicted performance. Optmyzr’s 2024 analysis of more than 1 million ads found no strong correlation between Ad Strength and outcomes: ads rated Average often beat Excellent-rated ads on cost per acquisition.
On creative fatigue, Osio declined to set a fixed interval. His rule: watch asset reporting, keep assets that still perform well, and replace assets only when they have declined from their fresh baseline and the ad is at capacity. A Demand Gen ad accepts up to 20 images, so capacity is not small.
Where the small-budget tension sits
This is the practical takeaway for growth teams. Google keeps building tools — Asset Studio, video enhancements, generated images — that lower the cost of producing creative variety. But Osio’s own logic says evaluation still needs budget and time. More assets mean more time and budget per asset before the system reaches a clear conclusion. An advertiser running near the $5 API minimum and uploading dozens of generated variants is likely under-exploring them.
That matters because Demand Gen is absorbing more of Google’s visual inventory. Video action campaigns have moved in, standalone Display is folding in with completion expected in 2027, and Maps arrived in May.
For teams planning Demand Gen, Osio’s checklist in brief:
- Budget: set at least $100/day per campaign before layering large volumes of non-feed creative.
- Coverage: run horizontal, vertical and square assets; push ad strength to Good or Excellent.
- Refresh: skip planned rotation. Replace only assets that show real decline against their own history.
- Data: connect tag gateway, CRM via Data Manager and enable enhanced conversions before scaling spend.
Gemini Omni arrives in Google Ads
The episode also highlighted the first appearance of Gemini Omni inside Google Ads via Asset Studio. Osio described it as “the first time Gemini Omni has made it into” Google Ads, using an advertiser’s existing creative and URL as inputs. The flow is prompt-based and storyboard-driven, with scene-by-scene editing. Osio used it the day before recording to help his cousin launch her first Demand Gen campaigns.
For smaller teams, the use case is clear: businesses with good photos but no video production, or B2B companies weak on visual assets, can generate vertical and horizontal versions in one workflow. The catch is that production is not measurement — output quality follows input quality, and you still need the budget to test what gets produced.
Why it matters for our audience
Google Ads is asking marketers to buy Demand Gen as the default visual layer, while its own published budget signals do not converge. The API says $5, the PM says $100, and bid-strategy rules of thumb point to 15–20 times target CPA. The safest planning position: treat $100 as the real floor for any campaign where you expect the system to learn, and use smaller budgets only for tightly constrained feed-based or brand-centric activity.
The bigger insight is about evaluation, not creative. Asset Studio can produce volume cheaply, but volume without testing budget just adds noise. Osio’s advice is to favour new asset quality over extreme quantities, keep proven assets live while adding new ones, and let asset reporting — not a calendar — decide when creative is tired.
Source: PPC Land
