The Main Google Display Ad Formats (and When Each One Wins)
If you’re running Display, you’ll spend most of your time choosing between three practical formats: responsive display ads (asset-based), uploaded image ads (static or animated), and video-based placements (either by adding video assets to responsive display ads or by running Video campaigns that can extend to video partners on the Display Network). The “best” format isn’t about preference—it’s about what you need control over (design consistency), what you need scale for (inventory coverage), and how quickly you want to test and iterate.
1) Responsive Display Ads (RDAs): the default, high-coverage format
Responsive display ads are built from assets you provide—headlines, descriptions, images, logos, and optionally videos. The system then assembles these assets into many combinations and sizes to fit available placements across the web, continuously optimizing toward performance. In real account management, this is the format I use most when the goal is broad reach, efficient testing, and fast creative iteration without hand-building dozens of banner sizes.
Practically, you’ll supply up to five short headlines (30 characters each), one long headline (90 characters), and up to five descriptions (90 characters), plus a business name (25 characters). For images, you’ll want both landscape and square coverage: a 1.91:1 landscape (recommended 1200×628, minimum 600×314) and a 1:1 square (recommended 600×600, minimum 300×300). Logos are optional but strongly recommended because they improve brand clarity across “native-like” layouts; you can provide a 1:1 logo (recommended 1200×1200, minimum 128×128) and a 4:1 logo (recommended 1200×300, minimum 512×128).
Where RDAs quietly outperform most “pretty banners” is in their ability to adapt. If your assets are strong, the same campaign can serve everything from standard rectangles to more native-feeling placements without you doing extra production work. It also reduces the risk of leaving impressions on the table simply because you didn’t upload the right size.
2) Uploaded Image Ads: the controlled, “pixel-perfect” option
Uploaded image ads are what most people think of as classic display banners: you design the exact creative and upload it. This format is ideal when brand teams require strict layout control, when you’re matching other channels’ creative, or when your testing requires a fixed design (for example, isolating message changes without layout variation).
You can upload JPG, PNG, or GIF files up to 600 KB. If you’re using animated GIFs, keep the animation within strict limits: the animation must be 30 seconds or shorter, it may loop but must stop after 30 seconds, and it must be slower than 5 frames per second. Those constraints matter because many “it won’t upload” or “limited serving” issues come down to animation compliance, not targeting or bidding.
From a sizing standpoint, the “main” sizes you’ll see again and again in real campaigns include 300×250, 728×90, 160×600, 300×600, 320×50, 320×100, and 970×250—plus several additional supported sizes. If you’re only going to invest in a few, start with 300×250 and 320×100 for volume, then add 728×90 and 160×600 for desktop reach, and 300×600 and 970×250 for high-impact inventory where it’s available.
3) HTML5 and AMPHTML Uploaded Ads: interactive creatives (with extra rules)
If you need richer motion, interactive elements, or more app-like behavior than a GIF can offer, uploaded HTML5 ads are the advanced format. These are uploaded as a ZIP that contains your HTML (and optionally CSS/JS and image assets), with a total file size limit of 600 KB. The creative must declare the ad size (via an ad size meta tag) and use standard HTML structure, and the ZIP must reference assets locally using relative paths (external references are restricted).
For teams producing HTML5 banners at scale, the operational benefit is that you can build one interactive concept and export multiple sizes—then traffic them as uploaded creatives rather than hoping responsive assembly preserves your intended motion design.
If HTML5 isn’t available for your account or workflow, AMPHTML uploaded ads are another option designed for faster loading (especially on mobile). In practice, I treat AMPHTML as a specialist tool: great when you have the production capability and need speed, but unnecessary for most advertisers who will get further faster with RDAs plus a small set of top-performing uploaded image sizes.
Where “Video Display Ads” Fit (Without Confusing Display vs. Video Campaigns)
“Video” in a Display conversation usually means one of two things, and it’s important to separate them because setup and reporting differ.
Option A: Add video assets to responsive display ads
Responsive display ads can use optional video assets in common ratios (horizontal 16:9, square 1:1, and vertical 2:3). When you provide videos, you’re essentially giving the system more ways to render your ad, which can increase engagement on placements that favor motion. If you don’t provide videos, the platform may still be able to build motion-style variations depending on your settings, but advertisers who care about brand safety and message control typically supply their own edited cuts rather than relying on automated generation.
Option B: Run Video campaigns and extend reach to video partners on the Display Network
If your goal is true video delivery (in-stream behavior, view-based bidding options, stronger video-centric reporting), you’ll usually run a Video campaign. Many Video campaigns can also serve on “video partners on the Display Network,” which are websites and apps outside of the core video platform environment. This is the cleanest path when you want scalable video reach beyond a single video property while still using video-first formats such as skippable in-stream, non-skippable in-stream, bumper, and in-feed video.
The key operational detail is that this is controlled in campaign network settings, and performance often shifts when video partners are included. In my experience, you should treat video partners like a separate “distribution layer”: test it deliberately, watch placement quality, and evaluate incremental reach versus changes in view rate and conversion efficiency.
How to Choose the Right Format (and Avoid the Most Common Performance Traps)
If you want maximum reach and fast learning: start with responsive display ads
For most advertisers, responsive display ads are the best baseline because they scale into more placements with less creative friction. The biggest performance trap here is weak asset variety. If you only upload one image and one headline, you’re limiting the system’s ability to find winning combinations, and you’ll often see “samey” output across placements. Strong RDAs look effortless, but they’re built on deliberate asset diversity: multiple angles, multiple value props, and multiple image compositions that still feel on-brand.
If you need strict brand control or a specific layout: add uploaded images (selectively)
Uploaded image ads tend to shine when brand consistency is non-negotiable, when you’re running seasonal creative that must look exactly like the rest of your campaign, or when your landing page experience depends on a particular visual promise. The biggest trap is wasting production cycles on dozens of low-volume sizes. Instead, focus on a tight set of proven sizes, validate performance, and only expand when the incremental impressions justify the design cost.
If your offer benefits from motion or explanation: use video intentionally
Video is most effective when the product needs demonstration, when trust needs to be built quickly, or when you’re moving users from awareness into consideration. The trap is assuming video automatically improves performance on Display. It can—when the cut is built for the placement and the audience—but it can also dilute results if you distribute broadly without monitoring where the video is actually serving and how users are engaging with it.
Critical pre-launch checklist (the items that prevent 80% of avoidable issues)
- For RDAs: Provide both landscape (1.91:1) and square (1:1) images, include at least one clean logo, and write multiple distinct headlines/descriptions that don’t repeat the same claim.
- For uploaded images: Keep files under 600 KB, use top-volume sizes first (especially 300×250 and 320×100), and ensure animated GIFs comply with the 30-second stop and frame-rate limits.
- For HTML5/AMPHTML: Confirm ZIP structure and size declarations are correct, keep total ZIP size within limits, and avoid prohibited external resource calls.
- For video distribution beyond core video inventory: Treat “video partners on the Display Network” as a controlled test, and review placement quality before scaling budgets.
Let AI handle
the Google Ads grunt work
Let AI handle
the Google Ads grunt work
If you’re weighing Responsive Display Ads versus uploaded image banners, HTML5/AMPHTML creatives, or adding video assets, it helps to have a clear view of what each format is best at, what specs matter most, and which common traps (like too few RDA assets, non-compliant GIF rules, or messy HTML5 ZIP structures) can quietly limit performance. Blobr fits naturally into that workflow by plugging into your Google Ads account and continuously checking what’s running, what’s wasting budget, and what to improve next; its specialized AI agents can also help refresh and align creative assets, like using the Headlines Enhancer or Ad Copy Rewriter to generate on-spec variations that better match your top search terms and landing pages.
The Main Google Display Ad Formats (and When Each One Wins)
If you’re running Display, you’ll spend most of your time choosing between three practical formats: responsive display ads (asset-based), uploaded image ads (static or animated), and video-based placements (either by adding video assets to responsive display ads or by running Video campaigns that can extend to video partners on the Display Network). The “best” format isn’t about preference—it’s about what you need control over (design consistency), what you need scale for (inventory coverage), and how quickly you want to test and iterate.
1) Responsive Display Ads (RDAs): the default, high-coverage format
Responsive display ads are built from assets you provide—headlines, descriptions, images, logos, and optionally videos. The system then assembles these assets into many combinations and sizes to fit available placements across the web, continuously optimizing toward performance. In real account management, this is the format I use most when the goal is broad reach, efficient testing, and fast creative iteration without hand-building dozens of banner sizes.
Practically, you’ll supply up to five short headlines (30 characters each), one long headline (90 characters), and up to five descriptions (90 characters), plus a business name (25 characters). For images, you’ll want both landscape and square coverage: a 1.91:1 landscape (recommended 1200×628, minimum 600×314) and a 1:1 square (recommended 600×600, minimum 300×300). Logos are optional but strongly recommended because they improve brand clarity across “native-like” layouts; you can provide a 1:1 logo (recommended 1200×1200, minimum 128×128) and a 4:1 logo (recommended 1200×300, minimum 512×128).
Where RDAs quietly outperform most “pretty banners” is in their ability to adapt. If your assets are strong, the same campaign can serve everything from standard rectangles to more native-feeling placements without you doing extra production work. It also reduces the risk of leaving impressions on the table simply because you didn’t upload the right size.
2) Uploaded Image Ads: the controlled, “pixel-perfect” option
Uploaded image ads are what most people think of as classic display banners: you design the exact creative and upload it. This format is ideal when brand teams require strict layout control, when you’re matching other channels’ creative, or when your testing requires a fixed design (for example, isolating message changes without layout variation).
You can upload JPG, PNG, or GIF files up to 600 KB. If you’re using animated GIFs, keep the animation within strict limits: the animation must be 30 seconds or shorter, it may loop but must stop after 30 seconds, and it must be slower than 5 frames per second. Those constraints matter because many “it won’t upload” or “limited serving” issues come down to animation compliance, not targeting or bidding.
From a sizing standpoint, the “main” sizes you’ll see again and again in real campaigns include 300×250, 728×90, 160×600, 300×600, 320×50, 320×100, and 970×250—plus several additional supported sizes. If you’re only going to invest in a few, start with 300×250 and 320×100 for volume, then add 728×90 and 160×600 for desktop reach, and 300×600 and 970×250 for high-impact inventory where it’s available.
3) HTML5 and AMPHTML Uploaded Ads: interactive creatives (with extra rules)
If you need richer motion, interactive elements, or more app-like behavior than a GIF can offer, uploaded HTML5 ads are the advanced format. These are uploaded as a ZIP that contains your HTML (and optionally CSS/JS and image assets), with a total file size limit of 600 KB. The creative must declare the ad size (via an ad size meta tag) and use standard HTML structure, and the ZIP must reference assets locally using relative paths (external references are restricted).
For teams producing HTML5 banners at scale, the operational benefit is that you can build one interactive concept and export multiple sizes—then traffic them as uploaded creatives rather than hoping responsive assembly preserves your intended motion design.
If HTML5 isn’t available for your account or workflow, AMPHTML uploaded ads are another option designed for faster loading (especially on mobile). In practice, I treat AMPHTML as a specialist tool: great when you have the production capability and need speed, but unnecessary for most advertisers who will get further faster with RDAs plus a small set of top-performing uploaded image sizes.
Where “Video Display Ads” Fit (Without Confusing Display vs. Video Campaigns)
“Video” in a Display conversation usually means one of two things, and it’s important to separate them because setup and reporting differ.
Option A: Add video assets to responsive display ads
Responsive display ads can use optional video assets in common ratios (horizontal 16:9, square 1:1, and vertical 2:3). When you provide videos, you’re essentially giving the system more ways to render your ad, which can increase engagement on placements that favor motion. If you don’t provide videos, the platform may still be able to build motion-style variations depending on your settings, but advertisers who care about brand safety and message control typically supply their own edited cuts rather than relying on automated generation.
Option B: Run Video campaigns and extend reach to video partners on the Display Network
If your goal is true video delivery (in-stream behavior, view-based bidding options, stronger video-centric reporting), you’ll usually run a Video campaign. Many Video campaigns can also serve on “video partners on the Display Network,” which are websites and apps outside of the core video platform environment. This is the cleanest path when you want scalable video reach beyond a single video property while still using video-first formats such as skippable in-stream, non-skippable in-stream, bumper, and in-feed video.
The key operational detail is that this is controlled in campaign network settings, and performance often shifts when video partners are included. In my experience, you should treat video partners like a separate “distribution layer”: test it deliberately, watch placement quality, and evaluate incremental reach versus changes in view rate and conversion efficiency.
How to Choose the Right Format (and Avoid the Most Common Performance Traps)
If you want maximum reach and fast learning: start with responsive display ads
For most advertisers, responsive display ads are the best baseline because they scale into more placements with less creative friction. The biggest performance trap here is weak asset variety. If you only upload one image and one headline, you’re limiting the system’s ability to find winning combinations, and you’ll often see “samey” output across placements. Strong RDAs look effortless, but they’re built on deliberate asset diversity: multiple angles, multiple value props, and multiple image compositions that still feel on-brand.
If you need strict brand control or a specific layout: add uploaded images (selectively)
Uploaded image ads tend to shine when brand consistency is non-negotiable, when you’re running seasonal creative that must look exactly like the rest of your campaign, or when your landing page experience depends on a particular visual promise. The biggest trap is wasting production cycles on dozens of low-volume sizes. Instead, focus on a tight set of proven sizes, validate performance, and only expand when the incremental impressions justify the design cost.
If your offer benefits from motion or explanation: use video intentionally
Video is most effective when the product needs demonstration, when trust needs to be built quickly, or when you’re moving users from awareness into consideration. The trap is assuming video automatically improves performance on Display. It can—when the cut is built for the placement and the audience—but it can also dilute results if you distribute broadly without monitoring where the video is actually serving and how users are engaging with it.
Critical pre-launch checklist (the items that prevent 80% of avoidable issues)
- For RDAs: Provide both landscape (1.91:1) and square (1:1) images, include at least one clean logo, and write multiple distinct headlines/descriptions that don’t repeat the same claim.
- For uploaded images: Keep files under 600 KB, use top-volume sizes first (especially 300×250 and 320×100), and ensure animated GIFs comply with the 30-second stop and frame-rate limits.
- For HTML5/AMPHTML: Confirm ZIP structure and size declarations are correct, keep total ZIP size within limits, and avoid prohibited external resource calls.
- For video distribution beyond core video inventory: Treat “video partners on the Display Network” as a controlled test, and review placement quality before scaling budgets.
