Attention

Banner blindness in 2026: why 99% of display ads are invisible

8 min read · Updated March 2026

The term banner blindness was coined in 1998 by Benway and Lane. Twenty-eight years later, the average click-through rate on a standard display banner is 0.05% — that is one click per two thousand impressions. The number has been falling every year since 2008. Eye-tracking studies by the Nielsen Norman Group, Infolinks, and more recently the IAB have found that banner ads receive, on average, less than 0.2 seconds of gaze time per session.

In practical terms: users have trained themselves to pattern-match anything banner-shaped and skip it entirely. The brain does this below the level of conscious attention. A 2024 Infolinks study showed that even when researchers moved a banner directly into the user's visual focus and paid them to look at it, recall 24 hours later was under 10%.

The 99% number, in context

Why do we say 99% of display ads are invisible? Three data points, combined:

Viewability: The IAB's own definition of a "viewable" ad is 50% of pixels on screen for 1 second. Current industry-average viewability is around 71% for desktop and 68% for mobile. So 29–32% of banners never even render.
Ad blockers: 42.7% of internet users globally run an ad blocker on at least one device (2025 GlobalWebIndex). In the 18–34 demographic it's 55%.
Banner blindness proper: Of the remaining "technically viewable" impressions that actually render, eye-tracking research shows only around 14% receive a fixation longer than 0.4 seconds. And of those, recall 24 hours later is under 10%.

Multiply it out: roughly 0.68 × 0.57 × 0.14 × 0.1 ≈ 0.54% of served impressions produce any measurable brand recall. You are paying for 1,000 impressions to potentially reach five people. This is why every major advertiser's net promoter score for display advertising has been negative since 2019.

What the eye-tracking actually shows

In heat maps of web pages, the banner positions — right rail, leaderboard, 300×250 in-content — show up as cold zones. The user's gaze trajectories literally route around them, like rocks in a stream. The effect is so strong that design researchers now advise never making legitimate UI elements banner-shaped, because users will ignore them by reflex.

"If it's rectangular, coloured, and in the top-right corner of a content page, the modern web user does not see it. Full stop." — Jakob Nielsen, Nielsen Norman Group

Why this got worse, not better

Between 2015 and 2025, three forces compounded:

Programmatic ad abuse. The average web page in 2025 serves 12+ ad calls. Users associate the "banner look" with pop-ups, auto-play video, and sketchy retargeting. The skip reflex generalised.
Scroll-first design. Mobile use now accounts for 63% of browsing. On a phone, any banner is above, below, or between content — the user's thumb has already moved past by the time it renders.
Generative slop. AI-generated ad creative flooded the market post-2023. Average quality dropped; average user scepticism rose.

What actually gets attention now

Attention follows content that doesn't look like an ad. The formats that measurably beat display in 2026 are:

Short-form native video (TikTok, Reels, YouTube Shorts): attention measured in multi-second holds, not milliseconds.
Newsletter sponsorships: 2–5 sentences in a trusted voice, average CTR 2–4% (40–80× display).
Interactive, physics-driven formats: scrollytelling, WebGL product viewers, and pull-to-read ad units like Rollaway Ads. Dwell times in the 6–12 second range.
Creator-embedded product placement: native mentions inside a creator's content. Highest recall but hardest to scale.

The common thread: the user chooses to engage. Banners assume they can force attention. They can't, and they couldn't for most of the last decade.

The tactical takeaway for brands

If your media plan in 2026 is still majority display, you are buying paper certificates that say "we tried to reach a human." Reallocate. Every real attention channel — creator content, sponsored newsletters, interactive formats, podcast reads — has a lower scale ceiling and a dramatically higher effective reach per dollar. The correct benchmark for a banner-replacement format is not CPM; it's cost per second of attention. Under that lens, a pull-to-read ad at $40 CPM with 8-second dwell beats a $3 CPM banner with 0.2-second dwell by a factor of ten.