What Are Responsive Ads? Submission Specifications and How to Create Effective Creatives
Responsive ads are the standard Web advertising format in which Google's or Yahoo!'s machine learning automatically adjusts the size, layout, and format of the ad by combining the headlines, descriptions, images, and videos an advertiser uploads as assets, so each placement and device receives the most suitable combination. Since Google Ads' expanded text ads could no longer be created after June 2022, both search ads and display ads have entered an era in which 'responsive ads are the default.' However, once operators start running them, many hit walls such as 'How many of each asset should I submit?', 'What are the character-count and image-size rules?', and 'How do I design creatives so machine learning can work effectively?' This article provides a practical, systematic explanation that beginners and intermediate operators can use right away: the fundamentals of responsive ads (RSA vs. RDA), submission specifications for Google Ads and Yahoo! Ads, how to create effective creatives, operational improvement points, and how to use MMM to accurately measure responsive ad performance under 2026's cookie restrictions.
Responsive ads are an ad type in which the ad's size, layout, and display format are automatically adjusted to fit each ad slot and device. Advertisers simply register multiple assets—headlines, descriptions, images, logos, and videos—and the platform's machine learning generates the optimal combination for each ad slot, serving them as text ads, image ads, native ads, and other formats. Unlike traditional banner ads, there is no need to produce individual images for each size; a single responsive ad can cover more than 50 ad-slot sizes and formats, which is its biggest advantage.
Responsive ads fall into two main categories. The first is Responsive Search Ads (RSA), text-format listing ads shown on Google Search and Yahoo! Search result pages. Advertisers submit multiple headlines and descriptions, and the platform automatically displays the combination that best matches each user's search query, device, and behavioral signals. The second is Responsive Display Ads (RDA), image/video/text hybrid ads served across the Google Display Network and Yahoo! Ads partner sites and apps. When you upload images, logos, videos, headlines, and descriptions, creatives auto-generated to fit each placement are displayed.
RSAs are designed for 'bottom-of-funnel harvesting' of users whose needs are already surfaced through search keywords, while RDAs target 'upper-funnel audiences' who might be interested in your product or service based on interests and behavioral history. Combining the two enables ad operations that cover the entire funnel from awareness to conversion.
Traditional banner ads (upload-type display ads) require individual creatives for each size—300×250, 728×90, 160×600, and so on—and any placement without a matching size is simply skipped. Responsive display ads, by contrast, only require you to register images, a logo, and text; machine learning handles the size and layout automatically, so ads can run across a broad set of placements and typically generate more impressions and clicks. That said, if you need to craft a design that strictly controls brand worldview, traditional banner ads let you deliver your exact intended message more faithfully. In practice, a hybrid approach is standard: 'use RDA to expand reach and accumulate learning data while supplementing the core banners with traditional banner ads.'
The biggest advantage is that you no longer need to produce multiple banners for every size. For responsive display ads, you just register base images (landscape and square), a logo, and text (headlines and descriptions), and machine learning auto-generates creatives that fit any placement. This dramatically cuts creative production time, outsourced design cost, and revision cycles, freeing up bandwidth for examining messaging angles, analyzing data, and other operational improvement work.
Traditional banner ads skipped placements whose sizes didn't match, but responsive ads can handle almost any size and format, minimizing opportunity loss and maximizing reach. Smartphones in particular have many screen-size and aspect-ratio variations, so the flexibility of responsive ads pays off significantly. Responsive search ads also benefit: submitting multiple headlines and descriptions broadens the match to user search queries, increases auction participation, and helps grow impressions and clicks.
With responsive ads, the platform's machine learning serves and learns from every submitted asset, preferentially surfacing combinations with strong performance. Because it can test far more patterns at high speed than a human could by hand, it raises the performance floor regardless of operator skill. Global brands such as HUGO BOSS have reported significant improvements in CTR and ROAS by combining responsive search ads with assets and Smart Bidding.
With responsive ads, you can see coarse asset-level performance labels (Best/Good/Low/Learning), but granular data such as 'how much did the specific Headline A × Description B combination contribute to CV?' is hard to obtain. Since all patterns are aggregated as a single set, rigorous A/B testing in the traditional sense is difficult, and the resolution of 'why are results what they are?' drops—something to keep in mind.
Because assets are combined in no particular order, it's an absolute requirement that 'any headline and description can stand next to each other without feeling off.' For example, pairing 'Industry No. 1' with another headline whose claim contradicts it can produce awkward ads depending on the combination. In industries with strict brand-guideline requirements, you can enforce some control with pinning features (Google's headline pinning).
When submitting multiple headlines and descriptions, packing in too many nearly identical phrases or repeating the same keywords may trigger asset-review disapproval. Rather than lining up headlines that mean essentially the same thing with slight ending variations, diversify the messaging angles themselves—benefits, specific numbers, offers, credibility, CTAs—which improves both review pass rates and machine-learning accuracy.
Google Ads' responsive search ads allow up to 15 headlines and up to 4 descriptions per ad. Character limits are 30 half-width characters for headlines (equivalent to 15 full-width characters), 90 half-width characters for descriptions (equivalent to 45 full-width characters), and 15 half-width characters for display URL paths (equivalent to 7 full-width characters). For full-width languages such as Japanese, note that one full-width character counts as two half-width characters. When the ad serves, up to three headlines can appear in traditional headline positions (above the ad or above the body), and up to two descriptions can appear.
Google Ads recommends at least one RSA per ad group with an Ad Strength of Good or Excellent. According to Google's own data, advertisers who improved Ad Strength from Poor to Excellent saw an average 12% increase in conversions, making headline and description strength a critical, performance-linked metric.
Yahoo! Ads' responsive search ads also allow up to 15 headlines and 4 descriptions, with shared character-counting rules ('full-width and half-width kana: 2 characters; half-width alphanumeric and symbols: 1 character'). Character limits also match Google Ads—30 half-width characters for headlines, 90 for descriptions, 15 for paths—so headlines and descriptions created for Google Ads can be reused in Yahoo! Ads almost as-is.
However, the allowed symbol sets differ slightly between Google and Yahoo!, and a symbol that works on one may be disapproved on the other. Yahoo! Ads also enforces its own policies (restrictions on superlatives, pharmaceutical-law-related wording, etc.), so when running on both platforms, always check the platform-specific review policies as well.
Google Ads' responsive display ads accept up to 5 short headlines (30 half-width characters or fewer), 1 long headline (90 half-width characters or fewer), up to 5 descriptions (90 half-width characters or fewer), and 1 business name (25 half-width characters or fewer). Short headlines appear in smaller ad spaces while long headlines appear in spaces with more room, so upload with balance in mind, considering which headline will appear where. Business name and logo support brand recognition, so always set them.
You can upload landscape images (aspect ratio 1.91:1, recommended 1200×628 px) and square images (aspect ratio 1:1, recommended 1200×1200 px), with at least one of each and up to 15 each. Logo images come in landscape (4:1, recommended 1200×300 px) and square (1:1, recommended 1200×1200 px). Maximum file size is 5120 KB (5 MB) per image, and supported formats are JPEG, PNG, and static GIF. If you include text in the image, keep it under 20% of the total area and place important information within the central 80%. Designing for up to 5% cropping on the left and right helps you prepare for auto-cropping safely.
Video assets can link up to 5 videos uploaded to YouTube; if videos are present, they may be served in place of images depending on performance. Supported aspect ratios are 16:9 (landscape), 1:1 (square), and 9:16 (vertical).
Yahoo! Ads (YDA) responsive image ads accept aspect ratio 1.91:1 (recommended 1200×628 px, minimum 600×314 px) and 1:1 (recommended 1200×1200 px, minimum 600×600 px). Supported formats are JPEG and PNG (animated GIF is not supported), and the per-image file-size ceiling is 3 MB (3000 KB)—significantly stricter than Google's. Auto-cropping of up to 15% is assumed on all four sides, so placing important information in the center of the image is essential.
Unlike Google Ads, Yahoo! Ads does not have a 20%-or-less in-image text restriction, but it strictly enforces restrictions on superlatives (No. 1, World's First, Best, etc.) and pharmaceutical-law-related wording. The review window is also about 3 business days—longer than Google's—and ads can flip to disapproved during post-review, so build schedule buffer when submitting.
When reusing the same images on both platforms, the efficient move is to build to Yahoo! Ads' specifications. Yahoo!'s file-size limit (3 MB), image sizes, and aspect ratios mostly fit within Google's specs, so the rule of thumb 'if it works in Yahoo!, it works in Google' reduces production effort. However, since Google recommends keeping in-image text at 20% or less, avoiding large in-image text is safe for both platforms.
The first thing to do with responsive search ads is fill headlines up to the 15-asset cap. The more you submit, the broader the range of combinations machine learning can test, and the more likely ads highly relevant to user search queries will be shown. Diversify headlines across angles such as (1) keyword-containing messages, (2) benefits (user upsides), (3) specific numbers and track record, (4) offers (free, limited-time, discount), (5) credibility (awards, client count), (6) calls to action, and (7) differentiation points (unique features, support). Listing similar expressions triggers disapproval and reduces learning efficiency, so craft headlines that differ in the messaging angle itself.
Up to 3 headlines (in any order) and up to 2 descriptions are auto-combined at serve time, so each asset must work as a standalone unit that makes sense paired with any other. Avoid situations like 'Request materials here' and 'Try free now' sitting side-by-side awkwardly in the same ad. And for headlines you absolutely need at the top of the ad (brand name or legally required copy), use pinning (Pin to Headline position 1 / position 2) to guarantee placement.
Combining sitelink, callout, structured snippet, price, promotion, phone-number, image, and lead-form assets expands the ad's real estate, improves visibility, and meaningfully lifts CTR. Google has publicly stated that advertisers who add image assets to responsive search ads tend to see better performance, and layering visual elements onto text-centric RSAs remains an effective best practice in 2026.
In responsive display ads, descriptions are frequently truncated in smaller ad slots, leaving only the short headline visible. For that reason, the 5 short headlines should contain catchy, self-contained copy that 'conveys your message with this one line alone.' If your design relies on descriptions to complete the message, intent won't come through once they're dropped—so prepare strong copy that can win on short headlines by themselves.
Google Ads RDA treats landscape images (1.91:1, 1200×628 px), square images (1:1, 1200×1200 px), and logos (landscape 4:1 and square 1:1) as required or recommended assets. If you provide only one type, the ad serves only on placements that fit that size, so always provide all three—landscape, square, and logo—to cover every placement. Start with 1–2 variations per asset type and expand winning patterns through operation.
Responsive ads auto-crop images to fit placements. Google Ads may crop up to 5% on the left and right, while Yahoo! Ads may crop up to 15% on all four sides, so place important text, logos, faces, and products within the central 80% of the image and fill the edges with background color or neutral elements. Yahoo! Ads crops significantly top and bottom too, so keep generous vertical margins.
Within a single ad (or ad group), keep creative and ad copy consistent in tone. Because machine learning auto-combines assets, if brand color, tone & manner, and messaging direction are all over the place, awkward combinations will surface. If you want to test different messaging angles, split them across separate ads (Angle A ad, Angle B ad), run them in parallel, and isolate the A/B test to a single variable. That makes it much easier to judge what drove results. Change only one of color scheme (red vs. blue vs. green), angle (price-focused vs. feature-focused), or CTA copy (free trial vs. request materials) at a time—that's the standard approach.
In the Google Ads UI, each RSA's Ad Strength is rated on a 4-step scale: Poor, Average, Good, Excellent. Advice like 'Add more headlines,' 'Include keywords in headlines,' and 'Add more unique headlines' appears, so follow it to add and improve headlines and descriptions and aim for Excellent. Google has stated that improving Ad Strength is directly correlated with conversion gains, making it one of the first operational improvements to tackle.
Running the same assets for long stretches creates user ad fatigue (banner blindness) and gradually erodes CTR and CVR. Google recommends refreshing responsive display ad creatives every few weeks to one month. In the admin 'Asset list,' periodically swap out assets marked 'Low' and continually introduce new messaging and imagery. When adding new assets, keep the high-performing existing ones and swap 1–2 at a time so you refresh without disrupting learning.
No matter how strong your responsive ad is, if the landing page (LP) that users click through to is inconsistent with the ad, they'll bounce in seconds and CVR will crater. Is the benefit you promised in the headline clearly stated in the LP's first view? Is the imagery consistent in tone? Is the CTA in an obvious spot? Design ad → LP as a single continuous path. Especially with RDA, which targets upper-funnel audiences, you need an opening line on the LP that convinces them 'why this ad is being shown to you right now.'
Responsive ads pair extremely well with Smart Bidding (tCPA, tROAS, Maximize Conversions, and so on), letting you offload 'which asset combination to serve to which user at what bid' to machine learning as a package. When measurement foundations such as accurate conversion tracking, Enhanced Conversions, Conversions API, and first-party data integration are in place, AI learning precision rises, and the combination of responsive ads × Smart Bidding can deliver significant performance gains.
Responsive ad operations are not 'set it and forget it'—they assume ongoing PDCA on a weekly to monthly cadence. Systematize a Plan (hypothesize angle and image patterns) → Do (add/submit assets) → Check (asset evaluation, CTR/CVR) → Act (swap low-performers, introduce new angles) cycle and build up winning patterns month over month. Also review search terms reports (for RSA) or placement reports (for RDA), and maintain negative keywords and excluded placements in parallel.
Accounts that stop at 'I'll submit 3 for now' don't give machine learning enough choices to unlock responsive ads' real potential. 15 headlines, 4 descriptions, and multiple images for both landscape and square—filling assets close to their caps is the non-negotiable prerequisite.
Trying to pad the count by changing only endings or phrasing creates near-duplicate headlines, triggering disapproval and reducing learning efficiency. Diversify messaging angles themselves—benefits, numbers, offers, credibility, CTAs—and aim for a state where all 15 headlines each attack a different angle.
There's roughly a 2-week learning period after submission or bid-strategy changes, and stacking heavy asset additions/deletions or bid changes during that window resets learning and destabilizes performance. Keep changes small during the learning period and hold off on serious improvements until data accumulates.
Responsive display ads drive awareness and interest among upper-funnel audiences, so measured by last-click CPA alone they always lose to search ads. But without RDA, users may never trigger the branded or comparison-stage searches in the first place, and killing RDA on short-term CPA alone can shrink search CV and the whole pipeline. Measurement design that captures indirect effects is essential.
A topic you cannot avoid in 2026 responsive ad operations is measurement-accuracy decline under cookie restrictions and the visibility of indirect effects. iOS's ATT, Android's Privacy Sandbox, and browser third-party cookie restrictions are rolling out progressively, making view-through CV and cross-device measurement harder than before. Enhanced Conversions, Conversions API (server-side measurement), and first-party data integration are must-have responses, but even so, pure last-click CPA evaluation cannot fairly assess the indirect contribution of awareness and interest channels such as responsive display ads.
Marketing Mix Modeling (MMM) is an effective answer to this challenge. MMM uses time-series data on each channel's ad investment and outcomes (conversions, sales, branded search volume) to statistically estimate each channel's contribution, making it possible to visualize the true contribution of responsive ads—including indirect effects—without relying on user-level tracking. With a cloud-native MMM platform like NeX-Ray, you can compare responsive search ads, responsive display ads, social ads, video ads, and even offline initiatives side by side and quantitatively answer 'how should we allocate budget across channels to maximize company-wide sales?'
Furthermore, MMM can estimate the response curve of sales against changes in responsive ad budget, supporting scientific budget-optimization decisions such as 'if we add ¥1 million to RDA this month, how much more sales do we earn?' or 'where is RSA's saturation point?' Combining last-click-based operational improvement with the whole-optimum view from MMM is the standard approach for sustainably growing responsive ad performance from 2026 onward.
Responsive ads are the standard format on Google Ads and Yahoo! Ads, comprising two types: Responsive Search Ads (RSA) and Responsive Display Ads (RDA). RSAs take up to 15 headlines and 4 descriptions, automatically displaying the best combination for each search query, while RDAs take images, logos, videos, and text and serve across ad slots of every size. Submission specs on both Google and Yahoo! use 30 half-width characters for headlines, 90 for descriptions, and aspect ratios 1.91:1 and 1:1 for images. Note that Yahoo! has stricter file-size limits (3 MB) and wider auto-cropping (up to 15% on all four sides).
Effective responsive ad creatives boil down to six points: (1) fill assets to the upper limit, (2) diversify angles and avoid near-duplicates, (3) design assets that make sense in any combination, (4) use strong copy that can stand on short headlines alone, (5) center-biased layouts that anticipate cropping, and (6) maintain message consistency between ad and LP. On the operations side, the basic playbook is raising Ad Strength to Excellent, refreshing creatives every few weeks to one month, pairing with Smart Bidding, and running PDCA weekly to monthly.
That said, in 2026's cookie-regulated era, last-click-based evaluation alone cannot properly assess the contribution of indirect-effect channels like responsive display ads, making whole-optimum investment decisions difficult. Using MMM like NeX-Ray to visualize contributions and response curves across responsive search ads, responsive display ads, social, video, and offline lets you run both wheels—last-click optimization and MMM-based whole optimization—and sustain responsive ad performance even in the cookie-less era. Use this article as a reference to review your responsive ad operations end-to-end, from submission specifications through creative design, and plan your next move.
A comprehensive 2026 guide to performance advertising (un'yougata koukoku): basic definition, auction-based bidding mech...
A 2026 practitioner's guide to running Google and Yahoo! search ads — from setup, KPIs, and keyword grouping to seven pr...
A 2026 practitioner's guide to Google Ads campaign types. Covers all 8 main types (Search, Display, Video, Shopping, App...
