Creative optimization: Why your best video ad fails as a static image

Summary

AppsFlyer’s State of Creative Optimization 2026, based on 1.2 million creatives, 140 elements, and roughly 1,400 apps, argues that video ad creative and static image ads are distinct disciplines. Video rewards ideas that need time and motion, such as testimonials, gameplay, and emotional build; static image rewards immediate offers, blunt CTAs, and single-frame clarity. The testing gap is widening: gaming apps spending $7M+ per quarter averaged 3,977 creative variations by Q1 2026, up 45%, while sub-$1M tiers flattened.

Non-gaming $7M+ apps averaged 3,477 variations, up 47%; $4M-$7M rose 29% to 2,615; $1M-$4M grew 16% to 1,670. Combined top tier averaged 3,727 variations. Below $1M, growth mostly stalled.

Spend also behaves differently by format. Gaming video’s top 2% of creatives captured 60.7% of spend by Q2 2026, up from 53.5%, while gaming image’s top 2% fell from 51.5% to 46.6%. Across verticals, the top 10% of creatives capture over 75% of spend.

Non-gaming video’s top 2% rose from 44.1% to 49.1%, while non-gaming image did not show the same split. Vertical examples reinforce the split: Action & Strategy’s video tone leader is Exciting, but image flips to Dramatic at 54.1% vs. 45.9%; Puzzle & Lifestyle’s high social proof wins up to 69% of video spend, yet one DSP’s image high share drops to 3.2% and low reaches 80%; Shopping’s video CTA is fragmented, with Download at 29.2%, but image Shop Now jumps from 6.4% to 67.8%.

Finance is the exception, with features-centric content winning both video at 40.8% and image at 43.2% on Social & Search. Actionable takeaways: close the testing gap with AI-assisted variation tools; treat video and image as separate test plans and budgets; concentrate deep analysis on video winners while keeping image testing broad; measure at the attribute level; and test both visible creative choices and psychological drivers.

Analyst Note

What's notable here isn't the video-versus-image debate itself — that framing has been settled for years — but the divergence in how spend concentrates within each format. Video's top 2% of creatives now absorb 60.7% of gaming video spend while image's top 2% share is falling, which suggests the two formats are separating in their underlying economics rather than simply rewarding different creative attributes. Video is behaving like a winner-take-most market; image like a long-tail one.

The key implication is organizational. Most accounts run a single creative testing pipeline with format as a variable inside it. If the attribute-level data holds — tone, social proof, and CTA reversing between formats across nearly every vertical except Finance — then format arguably belongs above the test plan rather than within it, with separate budgets and separate analysis time.

The competitive angle is harder to dismiss: variation volume at the $7M-plus quarterly tier grew roughly 46% year over year while sub-$1M advertisers flattened or declined. AI-assisted production is the stated driver, and it's compounding an advantage that already existed in budget. Worth watching whether mid-tier advertisers close that gap through tooling alone, or whether the binding constraint turns out to be analytical capacity rather than production capacity.