Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Google Ads & PPC › Display Campaigns

Responsive Display Ads

How responsive display ads assemble images, logos and text into every size, the asset requirements, creative rules that hold up, and how to test.
Guidebeginner10 min readUpdated

How a responsive display ad is assembled

Google documents the responsive display ad as the default ad type for Display campaigns. The advertiser uploads a pool of assets of each kind, and for every impression the system selects a combination and lays it out to fit the slot: a wide banner takes a landscape image and a headline; a tall slot stacks a square image above text; a native placement renders the assets in the publisher's own typography and colours; a slot that cannot show images at all serves a text-only version. The ad shown on one page is therefore a different object from the ad shown on the next, and the preview tool shows a sample of possibilities rather than a fixed design.

Three consequences follow. First, any asset may appear with any other, so each asset has to make sense in isolation and in every pairing. Second, the system chooses which assets to show more often based on performance, so the pool is also a running experiment. Third, Google may apply enhancements such as auto-generated video from the images or small changes to the rendering, under settings the advertiser can switch off; the current list of enhancements and the controls over them are documented in the Google Ads help pages and change over time. The tutorial on reviewing AI enhancements in your ad accounts turns that into a recorded decision per campaign. The format exists to maximise Reach: fixed-size image ads can only serve where a slot of exactly that size exists, while a responsive ad is eligible nearly everywhere.

Asset requirements and ratios

Google specifies the asset types, aspect ratios, pixel sizes and maximum counts on its help page for responsive display ads, and those specifications change, so the page is the reference and this table is the shape of it rather than a substitute.

AssetShapeRole in the ad
Landscape imageWide rectangleBanners and most horizontal slots; required
Square imageSquareVertical and small slots, native cards; required
Portrait imageTall rectangleSome mobile and native slots; optional
Landscape logoWide rectangleNative and wide layouts; optional but recommended
Square logoSquareSmall slots and native cards; recommended
HeadlineShort textThe first line; may appear alone
Long headlineLonger textUsed in larger layouts in place of the short headline
DescriptionTextSupporting line; may be omitted in small slots
Business nameTextShown as the advertiser identity
VideoOptional file or linkWhere the slot supports it; otherwise ignored
Call to actionChosen from a listButton text in layouts that show one

Images can be uploaded, taken from the landing page, or chosen from a stock library inside the interface; Google notes that the system may crop an image to fit a ratio, which is the reason to supply native square and landscape versions rather than relying on cropping. Text assets have character limits that the interface enforces as you type. The counts of each asset type allowed per ad are generous enough that the constraint in practice is having enough good assets, not the limit.

Writing headlines and descriptions for display

Display copy is read by someone who was looking at something else. The headline has to do the work of the whole ad, because in many layouts it is the only text that appears, and it has to work beside any of the images. Write each headline as a complete thought that names the offer or the benefit, not as half a sentence that expects the description to finish it. Avoid headlines that depend on each other, since only one appears at a time. The long headline is used in larger layouts and can carry a fuller claim; it is worth writing as its own line rather than as a longer version of a short headline.

Descriptions appear when there is room and are omitted when there is not, so nothing essential should live only in a description. They are the place for the proof, the condition or the detail: delivery terms, the range, what distinguishes the product. The business name is displayed as the brand, so it should be the name people recognise rather than a legal entity. The same principles apply to responsive search ads, which the responsive search ads guide covers in the search context, with the difference that search copy answers a query while display copy has to earn attention first. Click-through rate (CTR) on display is low by nature, so copy is better judged by whether it earns a glance and a memory than by clicks alone.

Images and logos that survive every size

The image is what people see; the text is often what they do not. Images that work in responsive display ads share a few properties. The subject sits in the centre, because the edges are what cropping removes. There is no text baked into the image, both because Google's guidance limits text overlay on images and because baked-in text becomes unreadable when the image is shrunk to a small square. The image reads at thumbnail size: one subject, strong contrast, no clutter. And the same message is available in both landscape and square versions produced from the source rather than by automatic cropping.

Logos follow the same logic: a square mark for small slots and a landscape lockup for wide ones, each on a plain background so that native rendering can place them cleanly. Logos with fine detail or thin type disappear at the sizes native placements use. Image preparation overlaps with the discipline the image SEO guide describes for websites: correct dimensions, sensible file sizes and a subject that is obvious at a glance are good for a landing page and good for an ad.

Google's image policies apply to every asset, including restrictions on misleading, sexual or shocking imagery and on image quality. Rejected assets are reported in the interface with the policy named, and an ad with a rejected required asset will not serve.

Native rendering vs image ads vs uploaded ads

Responsive display ads render in three broad ways, and the advertiser controls only which assets go in. Image layouts place the image and text in a conventional banner or square. Native layouts restyle the assets to match the publisher's page: the fonts, colours and arrangement come from the site, and the ad can look like an editorial card. Native placements are a large share of the inventory the format reaches that fixed-size ads cannot. Text layouts drop the image entirely where the slot does not support one.

Uploaded image ads are the alternative: finished designs in the fixed sizes Google supports, served exactly as designed and only where a slot of that size exists. They suit brand campaigns where the layout is the message, promotions with a specific design, and any case where a designer needs the final word. The two can run together in one ad group, and in practice most accounts do both: responsive ads for reach and native coverage, uploaded images for the sizes that matter most. Animated HTML ads are a further option with their own eligibility requirements, which Google restricts to accounts that meet its criteria.

Testing creative: asset performance, variations

The asset report inside a responsive display ad shows how each asset has performed relative to the others in the same ad, with a rating that Google documents as Low, Good or Best once there is enough data, and a "Learning" state before that. It is the primary testing tool for the format: assets rated Low are replaced, assets rated Best are kept and used as the model for new ones. The ratings are relative within the ad, so an ad full of weak assets will still have a Best.

Structured tests sit above that. Running two responsive ads in the same ad group with different image concepts, or different headline angles, is a workable A/B Test as long as both ads are given time and the comparison is made on the same measure. Google's ad rotation settings decide how evenly the ads are shown; an optimised rotation will favour the early leader, which is fine for performance and poor for a clean test. Judge display creative on the measures that fit the objective: click-through for a traffic campaign, conversions and view-through conversions for response, and reach at an acceptable cost per thousand impressions (CPM) for awareness, which the CPM calculator turns into an impression budget. Creative tests only mean something when targeting is held still; the targeting guide covers that side and this guide stays with the ad.

Illustrative example — not a real result

A language school rebuilds its display assets

A language school's responsive display ad has one landscape photo of a classroom, a logo with the full legal name in thin type, and five headlines that are all variants of "Learn Spanish online". After six weeks the asset report rates the logo and three headlines Low; the ad has delivered 820,000 impressions at £1.90 per thousand and a 0.21% click-through rate. The team replaces the assets: four images (a student on a video call, a tutor, a certificate, a city scene) each in native square and landscape versions, a square icon logo plus a landscape lockup, headlines that each make a distinct claim ("Live classes, six students max", "Speak from the first lesson", "Tutors from Madrid and Mexico City"), and descriptions that carry price and schedule. A second ad with the same text and a different image set runs alongside. After a month the rebuilt ad's click-through rate is 0.38% at a similar cost per thousand, the tutor image is rated Best, and the city-scene image is Low and replaced.

Creative by objective (awareness, remarketing, promotion)

The same format serves different jobs, and the assets should change with the job.

  • Awareness. The image carries the brand and a single idea; the headline names the category and the brand; the logo is prominent. Clicks are not the goal, so copy that asks for action is wasted space. Judge on reach and on how the campaign changes brand search and direct traffic.
  • Remarketing. The viewer already knows the brand, so the assets should move them on: the product they viewed, the reason to return, an answer to the objection that stopped them. Several distinct ads for different list segments beat one ad for everyone. Change the assets on a schedule, because the same person will see them many times.
  • Promotion. The offer is the headline, the condition is the description, the end date is visible, and the image shows the product rather than a mood. Uploaded image ads often earn their place here because the design can make the offer unmissable.

Whatever the objective, every asset should be checked in the preview at the smallest and largest sizes and in a native layout before the ad goes live. The display hub sets out where creative sits alongside targeting and measurement, and the Google Display Network expertise page describes the author's working process for building and refreshing asset sets. The display campaign launch tutorial places the ad build within the full launch sequence, after targeting, exclusions and the frequency cap.

Common mistakes

  • One image, one shape. Everything else is cropped by the system, badly.
  • Text baked into images. Unreadable at small sizes and limited by policy.
  • Headlines that only work in sequence. Each appears alone; each must stand alone.
  • Putting the essential claim in the description. It is dropped in small slots.
  • A logo nobody can read. Fine type and detail vanish in native placements.
  • Never opening the asset report. Low-rated assets stay in rotation for months.
  • Testing creative while changing targeting. Two variables, no answer.
  • The same assets for prospecting and remarketing. Different viewer, different job, different ad.
  • Relying on auto-generated enhancements without checking them. The system may build video or re-crop images from the assets; review what it produces and switch off what does not fit the brand.
  • Never refreshing the pool. Assets that were Best a year ago are often invisible now; replace on a schedule, not only when the report complains.

Frequently asked questions

Should I still upload image ads if I use responsive display ads?

Usually both. Responsive display ads reach far more inventory because they fit any slot, including native placements that fixed-size images cannot fill. Uploaded image ads give full control over how the ad looks, which matters for brand-led campaigns and for designs that must not be re-laid out. Running both in the same ad group lets the system serve whichever is eligible and performs, and the reports show which that is.

Why does my responsive display ad look different from the preview?

The preview shows a sample of possible layouts, not a guarantee. The ad is assembled per slot from the asset pool, so a headline may appear without a description, an image may be cropped to a square, and a native placement may restyle fonts and colours to match the publisher's page. Google documents this as the point of the format; the fix is assets that work in every combination rather than a layout that only works in one.

How many assets should I provide?

Enough variety for the system to assemble distinct combinations: several landscape and square images that each carry the message on their own, both logo shapes, headlines that read as complete thoughts, and descriptions that add something the headline did not say. Google's help page lists the current maximum per asset type, and getting close to those maximums with genuinely different assets is better than filling the slots with near-duplicates.

Terms used on this page

Tools to try

Where this page is used

Also related

Sources

  1. About responsive display ads — Google Ads Help
  2. Google Ads Help — Google Ads Help

Related