The standard sizes every ad network accepts

Eight dimensions form the IAB-recognized core of digital display. They sit inside Google Display Network's accepted-sizes list, get traded across the major SSPs, and survive both desktop and mobile rendering.

Every one of them tops out at the same 150 KB file weight that Google Ads enforces on uploaded display creatives, which is the single hardest production constraint you'll work against on every size you build.

a. Desktop sizes

The 728x90 Leaderboard runs at the top of articles and forums, generally above the fold, which is where it picks up the broad reach it’s known for. It’s the format that publishers default to for header rotations and the one ad networks lean on hardest for non-targeted reach campaigns.

• 160x600 (Wide Skyscraper) lives in the sidebar. It’s narrower than a left-rail nav and tall enough to track the reader as they scroll, which is why it tends to outperform the legacy 120x600 it largely replaced. 336x280 (Large Rectangle) is the in-content workhorse on long-form sites; it’s slightly bigger than its 300x250 cousin and most sites that accept one accept the other.
  
  
  

• 300x600 (Half Page) takes the largest non-premium footprint on the page. It sits sidebar or anchored to the main column and gives you usable creative real estate for storytelling: a hero shot, a headline, a CTA, breathing room. The size shows up in luxury, automotive, and B2B campaigns more than any other category because brands willing to pay for it want the canvas.
  
  
  

• 970x250 (Billboard) is the premium desktop format. It runs as a takeover above the masthead on publishers that support it, and the inventory is shallow on purpose. Premium placements aren’t sold by the same auction logic as the rest. The IAB included it in the New Ad Portfolio (2017) precisely because publishers wanted a high-impact above-the-fold unit that could compete with native and sponsored placements.

b. Mobile sizes

Mobile accounts for roughly 63% of US display impressions, which is enough that mobile dimensions should never be afterthoughts in a production set.

• The 320x50 Mobile Leaderboard is the baseline. Every ad network accepts it, every app SDK serves it, and it anchors the bottom of nearly every mobile web page that runs display.
  
  
  

• 320x100 (Large Mobile Banner) is the upgrade: twice the height, same width, more room for an actual readable headline and a tap target that doesn’t feel cramped.

Both formats squeeze your entire creative into 16,000 to 32,000 pixels of canvas while the file weight ceiling stays at 150 KB.

That makes mobile the tightest production environment in the standard set, which is exactly why production order matters more here than anywhere else: a mobile creative that works at 320x50 will scale up to 320x100 cleanly, while the reverse rarely does.

c. The universal size

300x250 (Medium Rectangle) is the one size every designer should know without thinking. It runs in-content on desktop articles, in the sidebar of news sites, inside the feed of mobile apps, and at every position in between. No other format has the cross-device reach: 300x250 is what makes a single creative work on a laptop, a tablet, and a phone without rebuilds.

It also has the deepest inventory. Publishers that don’t support anything else still support 300x250, which is why nearly every campaign starts with this size and almost every retargeting flight relies on it. The format is non-disruptive enough to sit inside editorial flow without breaking the reading experience, and most readers have learned to tune it out or click it depending on relevance.

336x280 (Large Rectangle) is the upsize variant of the same slot. Sites that run 300x250 in-content typically accept 336x280 too, giving you a 12% larger canvas for the same placement when the publisher allows substitution.

How each size actually performs

Most size guides tell you what's popular. Viewability data tells you what actually works. The two answers don't always agree, which is why size selection benefits from looking at the performance number publishers and advertisers actually trade on.

1. Viewability by size

The IAB and MRC define a display ad as "viewable" when at least 50% of its pixels are on screen for at least one continuous second. Google Active View measures this across every impression the Display Network serves, and the rate varies sharply by size.

Per Google Active View data compiled in searchlab.nl's 2026 programmatic display benchmarks:

The pattern isn’t coincidence: larger sizes that occupy more of the viewport at typical scroll positions stay viewable longer. The 300x600 outperforms the 728x90 by 26 percentage points despite the leaderboard’s broader reach, which is the entire reason agencies have shifted spend toward half-pages over the last few years. For designers, the practical read is that a creative built for a high-viewability size has a structural advantage before the design itself does any work.

Viewability connects to outcomes that designers don’t usually see on the back end. Per Campaign Monitor and GroupM data published in 2026, creative that holds above 70% viewability delivers 2.4x higher brand lift than creative below that threshold. The size you pick puts you on one side of that line or the other before you’ve started.

2. Inventory and reach

Performance data is one half of the story. The other half is whether the size will get served at all. Programmatic captures 91% of US display spend, which means standard sizes don’t just outperform: they outprint. 300x250 and 728x90 dominate impression volume not because they’re better creative containers but because publishers universally accept them, which feeds the open auction with constant supply. Designers who treat inventory depth as a production input rather than a media-buying detail end up producing the right set of creatives the first time.

That supply matters at production time. Two creatives at a high-inventory size return more impressions than five at fringe formats, simply because publishers will buy what they have slots for. The 300x250 is the canonical example: it sits inside every networked publisher’s accepted-sizes set, which is why a single creative at this dimension reaches further than any other you’ll produce.

Animation pulls in the same direction. Standard display banners run at 0.046% average CTR. Rich media and HTML5 banners run at 0.18% average CTR, four times the 0.046% static display average, per the IAB benchmark series. The size you pick determines reach; whether you animate it determines whether anyone clicks.

Fixed pixel sizes VS responsive display ads

Build fixed pixel sizes first. Build responsive assets when the buy calls for them. The eight standard sizes still run the majority of digital display, and that’s the right starting point for nearly every campaign you’ll produce.

Responsive display ads are Google’s automated format. You upload assets (multiple images at different aspect ratios, several headlines, a few descriptions, a logo), and Google’s system assembles ad variations to fit whatever placement comes up across the Display Network and Performance Max.

There’s no single named deliverable. The designer’s job is producing an asset library the algorithm can recombine, with Google’s spec calling for at least:

• one landscape image at 1.91:1

• one square at 1:1

• and a logo at 1:1

• plus optional vertical at 4:5

The IAB tried to formalize a shift to flexible sizing back in 2017 with the New Ad Portfolio, which defined ad units by aspect ratio rather than by pixel dimensions. Adoption has been partial. Publishers still sell named sizes because programmatic auctions still bid on named sizes, and the buy side still requests them because reporting, creative QA, and trafficking workflows are built around fixed dimensions.

Responsive display ads layer on top of that ecosystem rather than replacing it. The practical effect is that designers produce both:

1. fixed pixel sets for the majority of placements

2. responsive asset libraries for the Google-automated portion of the spend

For designers, the practical read: produce fixed sizes for almost every campaign and treat responsive specs as an additional deliverable when the media plan includes Performance Max or pure programmatic reach. The asset prep is different enough that retrofitting fixed creatives into responsive sets after the fact tends to compromise both. A 300x250 cropped to 1:1 loses headline space, and a 1.91:1 asset stretched into a 728x90 loses center weighting.

Which sizes to build first?

Four sizes give you maximum coverage with minimum production. Build 300x250, 728x90, 320x50, and 300x600 in that order, and you have one universal in-content unit, one desktop reach unit, one mobile baseline, and one high-viewability storytelling unit.

That set runs on more inventory and delivers more impressions per creative produced than any other four sizes you could pick.

The number is four rather than three because a three-size set leaves either mobile or storytelling on the cutting-room floor. It's not five because the fifth size you'd add (336x280) is functionally interchangeable with 300x250 on most placements, so the marginal coverage you gain doesn't justify the marginal production effort.

Four is the smallest set that covers every meaningful axis: universal placement, desktop reach, mobile, and large-format viewability.

Build mobile early, not late. Mobile carries 63% of US display impressions, and the production constraints at 320x50 are tighter than anything you'll face on desktop. A creative that works in 16,000 pixels of canvas with a 150 KB ceiling will scale up to larger formats without losing its core idea. The reverse rarely holds: a desktop creative cropped to mobile loses its center weighting, its hierarchy, and often the headline that made it work in the first place.

Once the starter set is built, scale to the full eight by adding 336x280, 160x600, 320x100, and 970x250 in roughly that order. Smart Resize handles the canvas math: import the master design and Bannersnack generates dimensionally-correct adaptations across the standard set, leaving you to refine layout and hierarchy rather than rebuild from scratch. The same workflow extends from the eight standards into responsive asset sets when the buy calls for them.

Start with the universal 300x250

Sizes you can skip

Some sizes are still in the IAB list but barely show up in programmatic inventory anymore.

• Skip 468x60 (Full Banner) and 120x600 (Skyscraper) on most production sets. Both were standard a decade ago and remain technically supported, but publishers have moved their inventory to 728x90 and 160x600 respectively, and the formats that look similar in the old IAB tables aren’t competing for the same impressions in 2026.
  
  
  

• A 468x60 placement that gets sold today is usually a remnant slot at a low CPM, not a campaign-quality placement worth a dedicated creative.
  
  
  

• The 250x250 (Square) and 200x200 (Small Square) sit in the same category. They’re occasionally requested for specific layouts or app placements, but the inventory is thin enough that producing them on speculation rarely pays back the design time. Build them when a campaign specifically calls for them; don’t add them to a default production set.
  
  
  

• The 970x90 (Large Leaderboard) sits in a slightly different category. It’s technically supported and occasionally requested, but the 970x250 (Billboard) has largely replaced it for premium top-of-page placements that want a bigger canvas than the standard 728x90.

A quick decision rule: if a size isn’t in the eight-size standard set and the campaign doesn’t name a specific placement that requires it, leave it off the production list. The exception is when a publisher’s direct-buy spec requires a particular size, which is a real reason to produce it. Generic “the more sizes the better” requests usually aren’t.

Production specs that affect every size

A 970x250 hits the 150 KB ceiling faster than a 320x50. The same creative idea has roughly 16 times more canvas area at the billboard size, which means roughly 16 times more image data to compress before the file even includes typefaces or animation. Plan compression as a per-size budget, not an article-end optimization.

Animation behavior tracks the same logic. The 300x600 has the screen real estate for multi-frame storytelling: opening shot, product reveal, CTA. The 320x50 doesn't. A four-frame loop that works on a half-page reads as visual noise on a mobile leaderboard. Match animation complexity to format, and the rich-media CTR uplift (4x static, per the IAB benchmark series) follows naturally.

Center-safe areas matter when creatives serve responsively or run on networks that crop for unusual placements. Keep the logo, CTA, and headline inside the inner 80% of the canvas on every size, and the same asset survives unexpected aspect-ratio adjustments without losing its hierarchy.

HTML5 export remains the universal acceptance format across Google Ads, Display & Video 360, and the major SSPs. Static GIF and JPG fallbacks ship alongside it for placements that don't support HTML5.

Conclusion

Eight sizes cover most of the inventory worth covering, and four of them (300x250, 728x90, 320x50, and 300x600) return more impressions per production hour than any other combination you could pick. The data behind that hierarchy comes from viewability and inventory depth, not opinion.

Build that starter set first, scale to the full eight with Smart Resize when the campaign grows, and the production effort tracks the impressions you actually receive rather than the sizes someone listed in a brief. That sequencing, production-first then media-second, is what makes the size set work for the people who have to build it.

P.S. You’ll spend more time optimizing file weight than picking sizes. The 150 KB ceiling is the real constraint. Master that, and the rest is just canvas math.