Banner Ad Design: Best Practices, Sizes and Testing

What banner design can influence

Effective banner design can determine whether a display impression communicates a clear offer or disappears into the page. But design is only one input into display performance. A banner works alongside the offer itself, the audience targeting, the quality of the inventory it runs on, and the landing page it points to. A well-designed ad cannot rescue an uncompetitive offer, poor targeting, or a landing page that fails to fulfill the promise made in the creative.

Be precise about what design influences and what it does not. Design strongly shapes whether a viewable ad is understood quickly, whether the brand is recognizable, whether the next action is easy to find, and whether the creative feels consistent with its destination. It has far less influence over viewability, which depends largely on placement, page layout and load speed; over who sees the ad, which depends on targeting and auction dynamics; and over whether the underlying offer is compelling.

This distinction sets realistic expectations. Improving a CTA button will not fix a campaign whose real problem is inventory quality or offer strength. Good design removes creative friction so the rest of the system can work.

Key takeaways

• Design is one input, not the whole campaign. It removes creative friction; it cannot fix a weak offer, poor targeting, low-quality inventory or a mismatched landing page.
• Start from the objective. Awareness, traffic, lead generation, e-commerce and retargeting each call for a different hierarchy, CTA and set of success metrics.
• Match the message, not just the visuals. The landing page should fulfill the same offer, price and next action the ad promised.
• Design each format as its own layout, preserving the proposition, brand asset, focal point, CTA meaning and offer terms.
• Judge creative on downstream outcomes (conversion rate, CPA and revenue), not on CTR in isolation.

Start with the campaign objective

Before choosing colors, copy or layout, decide what the campaign is for. The objective shapes almost every downstream design decision, and it changes how you measure success.

• Brand awareness campaigns prioritize distinctive brand assets and are measured by viewable reach and frequency, with brand lift where applicable, and attention metrics applied cautiously where the method is transparent. They are not primarily judged on click-through rate.
• Traffic campaigns focus on a clear, clickable value proposition and are measured on click-through rate and post-click engagement.
• Lead generation campaigns value a credible offer and a low-friction next step, measured on qualified lead rate and cost per acquisition.
• E-commerce campaigns prioritize product clarity plus offer terms, measured on conversion rate, revenue, and return on ad spend.
• Retargeting campaigns remind and reassure a warm audience. Evaluate them on CPA and frequency, using incrementality testing where sufficient scale and platform controls allow it.

Design a banner without a stated objective, and you tend to get a creative that does several things adequately and nothing well. Name the objective first, then let it justify the hierarchy, copy, and CTA that follow.

Build a clear message hierarchy

Visual hierarchy determines the order in which a viewer processes your ad. Because display ads are often encountered briefly and in peripheral vision, the primary offer should be understandable without sustained reading.

No single hierarchy suits every ad. Choose a deliberate order based on the objective. A direct-response ad may lead with the offer or the problem it solves, so the value lands first. An established-brand campaign may lead with a distinctive brand asset, because recognition is the point. A product-led creative may lead with the product image as the entry point.

What matters is that one element clearly leads, the sequence is intentional, and the CTA does not compete with the core message. When every element bears equal visual weight, the intended sequence becomes harder to identify.

The CTA should be easy to locate once the offer is understood and visually distinct from its immediate surroundings, without eclipsing the primary message. If the button dominates before the viewer grasps the value proposition, the contrast has been misallocated. A visually distinct button is easier to identify as the next action, though distinctiveness alone does not guarantee a click, and a button that looks out of place can read as deceptive or poorly branded.

The blur test

Squint at the ad, or step back until the details blur. If the leading element and the CTA still register, the hierarchy is working. If everything merges into one undifferentiated mass, the ad lacks focus. Use it as a fast hierarchy check, then validate the layout through campaign testing. Run it on every size variation, because a hierarchy that works in a 728×90 leaderboard can fail in a 300×250 rectangle.

Match the message to the landing page

Preserve recognizable brand cues and, more importantly, maintain message match between the ad and its destination. The landing page should clearly fulfill the offer made in the creative (same offer, same price or discount, same eligibility conditions, same product, same next action) even if the layouts are not visually identical. A display ad may legitimately use a campaign accent color or condensed typography because of size constraints; continuity of expectation matters more than exact visual duplication. A mismatch between ad and landing page can raise bounce rates and weaken trust, so test message match directly rather than assuming visual sameness is enough.

Adapt layouts across banner formats

Design each format as its own layout, not a resize of another. Forcing a leaderboard composition into a skyscraper frame produces awkward spacing and a weakened message.

• 300×250 (medium rectangle): balanced space suited to a centered composition with imagery, headline, and CTA stacked vertically. Embedded inside content or in a sidebar.
• 728×90 (leaderboard): wide and short, suited to a left-to-right flow: brand, headline, CTA. Top or bottom of the page.
• 160×600 (wide skyscraper): tall and narrow, suited to vertical storytelling which builds downward. Sidebar placement.
• 300×600 (half-page): high-impact sidebar unit with room for more developed messaging.
• 320×50 (mobile leaderboard): minimal space; needs the shortest copy and a clearly tappable CTA. Top of mobile screens.

When you adapt one creative across formats, be explicit about what to preserve and what may change. Preserve the core proposition, the distinctive brand asset, the visual focal point, the CTA’s meaning, and the offer terms. Let line breaks, image crop, copy length, element order, and (for animated units) the animation sequence change to fit each frame. The illustration below shows a single “Save 30% on annual plans” offer carried across all five formats: the brand mark, screenshot, headline, meaning, and CTA stay constant while the layout, crop, and copy length flex to each frame.

Whitespace improves separation and legibility, particularly in smaller formats, and clean layouts often direct attention more efficiently than dense ones. But the right density depends on the offer, format, and audience; dense retail creatives can work well when multiple products or price cues are central. Treat generous whitespace as a strong default, not an absolute rule.

Avoid generic stock imagery that could belong to any brand. Product photography suits direct-response campaigns where the viewer needs to see what they are buying; lifestyle imagery suits brand awareness campaigns where the viewer needs to picture themselves in the story; custom illustration can differentiate where stock cannot.

Write copy and CTAs for the available space

Use the fewest words needed to communicate the offer clearly. Smaller formats demand very short copy; larger or sequential formats (300×600 units, animated frames, responsive assets, complex B2B offers) can carry added context as long as the hierarchy stays clear. Rigid word counts do not survive contact with that range of formats, so let the space and the offer set the length.

Concrete, benefit-led headlines are usually easier to understand quickly than ambiguous or purely clever phrasing. “Save 30% on annual plans” communicates value as well as specificity in one line; “Unlock your savings potential” makes the reader work for it.

For CTAs, the guiding principle is alignment between the button and the destination, not a blanket rule about which words work. Action-oriented, specific language such as “Start Free Trial” or “Get My Free Quote” is strong when it accurately describes what happens next. “Learn More” is not automatically weak: it can be the honest, appropriate choice for high-consideration purchases, information campaigns, complex B2B products, educational content, or sensitive categories where urgency would be inappropriate. Choose CTA language that truthfully describes the user’s next step: “Start Free Trial” when a trial begins immediately, “Learn More” when the landing page is educational.

Urgency can motivate someone to act now, but only when the deadline or scarcity is genuine and clearly supported on the landing page. Manufactured urgency creates misleading ads, damages trust, and attracts low-quality clicks that hurt downstream performance.

Design for accessibility and mobile

Accessibility is a core part of responsible creative, not an afterthought. Use sufficient text-to-background contrast so the copy is legible, not simply striking. Keep type legible at the rendered size, and avoid text embedded in images at resolutions that blur on high-density screens. Do not rely on color alone to convey meaning. Prefer restrained motion: for HTML5 units, respect reduced-motion settings where the serving environment supports them; for GIFs, avoid rapid flashing and unnecessary looping. Provide clear controls for interactive or expanding units. Provide accessible labels or alt text where the environment and format support them, and ensure the essential offer is not communicated only through an image that assistive technology cannot interpret. Test the actual rendered unit in context, not just the source artwork.

Mobile deserves more than “shorter copy, bigger CTA.” Keep the visual CTA legible and away from crowded edges, and test the full clickable unit on mobile to avoid layouts that encourage accidental taps. Supply high-density assets and correct cropping across aspect ratios. Account for differences between in-app and mobile-web rendering and compression. Monitor landing page load speed, as it disproportionately impacts mobile conversion. Carry the message match through to the mobile landing experience.

Accidental taps matter for measurement as well as usability: a high CTR on a 320×50 unit can partly reflect mistaps rather than genuine interest, which is another reason CTR should never be read in isolation.

Follow current technical requirements

Platform file size and format rules vary by campaign type and account eligibility, and they change over time, so confirm the current limits in the platform’s documentation before building. The values below are grouped by context because they are frequently, and incorrectly, merged into a single number.

Uploaded image ads accept only GIFs, JPGs, and PNGs, with a limit of 150 KB per file; animated GIFs share the same 150 KB ceiling. Responsive display image assets follow a different rule and can be up to 5,120 KB each. Uploaded HTML5 ads are packaged as a ZIP capped at 600 KB, and access to HTML5 uploads is gated: the account must generally be more than 90 days old, have over $9,000 in lifetime spend, and hold a clean policy record. HTML5 ads for App campaigns are a separate context with a larger 5 MB ZIP limit. Because so many contexts exist, “the file-size limit” is never a single figure: which limit applies depends on whether you are uploading a static image ad, supplying responsive assets, uploading an HTML5 ZIP for display, or building for App campaigns, DV360, or CM360.

Sources: Google uploaded display ads specifications, Google responsive display guidance, Google HTML5/Playable ads for App campaigns, and DV360 HTML5 creative guidelines.

Fixed dimensions such as 300×250 and 728×90 remain widely accepted across display inventory, but current workflows also rely on responsive and flexible units. IAB’s New Ad Portfolio moved toward flexible, aspect-ratio-based units and lightweight (LEAN) principles rather than a fixed set of pixel sizes. Build for the buying platform and publisher specification rather than assuming five fixed dimensions cover every placement.

Where you supply components for responsive display ads rather than finished designs, each asset must be strong enough to work both in isolation and combined with others. Google recommends providing horizontal, square, and vertical image ratios; supply assets that each stand on their own.

One shift worth planning around: Google has begun moving Display campaigns into Demand Gen, with eligible advertisers able to start migrating existing campaigns from June 2026 and new campaigns eventually created only within Demand Gen. The creative principles in this guide still apply, but confirm current campaign types and asset requirements in Google’s documentation before you build, since the surface you are building for is changing.

Test and diagnose performance

Systematic testing improves creative, but the quality of the learning depends on how the test is designed. Two clarifications improve on the common “just test one variable at a time” advice.

First, sequence matters. A headline, image, and CTA usually work as a system, so testing button colors while leaving a weak concept unchanged produces low-value learning. A more productive order is to test materially different creative concepts first, identify the stronger concept, then test individual elements within that direction, holding audience, placement, bidding, and measurement conditions as stable as practical, and waiting for a sufficient sample size before deciding. Single-variable testing is a legitimate, useful tool for isolating a specific effect; apply it once you have a concept worth refining, not as the only method.

Second, valid testing needs statistical discipline that the “one variable” framing tends to omit: a defined primary metric, a minimum detectable effect worth caring about, enough impressions or conversions to detect it, consistent targeting conditions, an adequate duration, and the patience to avoid picking an early “winner.” Account for conversion lag and platform learning periods before concluding.

Manage ad fatigue as an operational discipline. When the same creative runs too often, response declines as users learn to ignore it. Use frequency controls where available, monitor frequency by audience, and refresh creative when performance deteriorates with repeated exposure.

Match the primary measures to the objective rather than applying one universal ladder:

Two cautions on interpretation. Viewability is an opportunity to see, not attention: under the common MRC display-viewability standard, at least 50% of an ad’s pixels must be in view for at least one continuous second, and larger display units can carry a different pixel-percentage threshold. Viewability is mainly driven by placement, latency, and page layout, so designers cannot solve all viewability problems through creative alone. See the IAB/MRC viewability guidance for the distinction between viewability and ad effect.

A higher CTR is not automatically good, either. Extra clicks are valuable only if they maintain or improve conversion quality; clicks from low-intent or accidental taps can reduce efficiency. Evaluate creative against downstream conversion rate, CPA, and revenue rather than CTR alone. Instead of inventing a bespoke metric, use standard measures: CTR as a diagnostic, then post-click engagement (in GA4, an engaged session lasts longer than 10 seconds, includes a key event, or includes at least two page or screen views), then landing-page conversion rate, CPA, and revenue.

When performance disappoints, this table points to the area most worth inspecting:

A final note on attribution: platform reports may credit conversions after a viewable impression (view-through attribution), but that shows an association, not a proven incremental impact. Interpret view-through numbers alongside incrementality testing where possible.

How Linear approaches campaign optimization

Linear pairs proprietary cross-platform AI with dedicated PPC specialists to identify and act on campaign optimization opportunities. Automated pattern detection provides speed and breadth across accounts; specialist review grounds each decision in the specifics of the campaign (audience, bid strategy, budget, and relevance) before changes are implemented. The value is in the pairing: machine scale plus human assessment.

If you are scaling display advertising, that combination is designed to surface where creative, targeting, and landing-page alignment may be limiting your campaigns, then act on it with a person in the loop.

Banner-design checklist

• Objective named before the design begins, with metrics chosen to match it.
• One element clearly leads the hierarchy; the CTA is findable without eclipsing the offer.
• Message and offer match the landing page: same offer, price, terms, and next action.
• Each format is designed as its own layout, preserving proposition, brand asset, focal point, CTA meaning, and offer terms.
• Copy length fits the space; CTA language honestly describes the next step.
• Contrast, legibility, motion, and mobile tap targets meet accessibility needs.
• Technical specs checked against the current platform documentation for the correct upload context.
• Testing runs concepts first, then elements, with a defined metric and an adequate sample.
• Performance read against downstream outcomes, not CTR in isolation.
• Blur test passed on every size variation.

Frequently asked questions

What is a realistic CTR for banner ads?

There is no useful universal target. CTR varies by industry, objective, audience, inventory, placement, device, and format. Benchmark against comparable campaigns in your own account, then judge whether the clicks produce engaged visits, qualified leads, or revenue.

What are the standard banner ad sizes?

Widely used fixed sizes include 300×250 (medium rectangle), 728×90 (leaderboard), 160×600 (wide skyscraper), 300×600 (half-page), and 320×50 (mobile leaderboard). Current workflows also use responsive and flexible aspect-ratio units, so build to the platform and publisher specification rather than assuming a fixed set covers every placement.

What is banner blindness, and how do I reduce its impact?

Banner blindness is the tendency to ignore elements that appear to be advertising or appear in familiar ad locations. It is distinct from viewability, which measures whether an ad had the opportunity to be seen. The two need different solutions: placement and page delivery affect viewability, while relevance, clarity, and creative execution influence whether a viewable ad receives attention. Test contextually aligned placements rather than assuming that contextual similarity automatically improves results.

What makes an effective CTA?

A CTA that accurately describes the user’s next step is easy to locate after the offer is understood and is visually distinct from its surroundings without overpowering the message. Match the wording to the destination, and use urgency only when it is genuine.

How is AI changing banner design?

AI tools can help generate and evaluate creative variations and support dynamic creative optimization that tailors messages based on signals such as location, device, or situation. These capabilities work best when automated recommendations are reviewed against campaign context by a person before changes go live.

To summarize

Better banner design is not about making the CTA louder. It is about matching up the offer, hierarchy, format, placement, and landing-page experience to a clear objective, then testing whether the resulting traffic is actually valuable. Design removes creative friction; it cannot substitute for a strong offer, good targeting, or a coherent destination. Get those working together, measure against downstream outcomes rather than clicks alone, and refresh before fatigue sets in.