Omneky Introduces TASTE BENCH for Evaluating AI Creative Quality

Omneky launched TASTE BENCH, an AI creative-evaluation suite measuring whether image and video models can convert briefs and brand assets into ready-to-run advertising. In its initial image-ad test spanning eight models, 59 briefs, four brands and five languages, GPT Image 2.5 Sunburst posted the highest ready-to-run rate at 54.2% (32 of 59) and an average quality score of 7.38/10; model pass rates ranged from 22.0% to 54.2%. The results are based on automated blind AI-judge reviews and hard compliance checks, rather than advertising conversion or ROAS outcomes.

Analysis

The investable read-through is modestly favorable for GOOG and META, but not because a benchmark establishes monetization. A standardized quality gate could reduce the operational friction that prevents smaller advertisers from scaling generative creative, increasing ad-variant production and potentially raising campaign experimentation intensity. META is the more direct near-term beneficiary if lower creative-production costs expand performance-ad spend from long-tail advertisers; GOOG benefits primarily where automated creative is paired with Search, YouTube and Performance Max inventory.

The important limitation is that automated aesthetic and compliance scores are not evidence of incremental conversion, ROAS, or advertiser retention. The relevant 1-3 month catalyst is whether major ad platforms report higher adoption of automated asset generation, rising campaign-asset breadth, or improved conversion metrics without a deterioration in advertiser ROI; absent that, this is product-marketing noise rather than an earnings input. Over 6-18 months, model commoditization could shift value away from standalone creative-generation vendors and toward closed-loop platforms with proprietary conversion data, favoring META and GOOG—but only if they retain advertiser control rather than becoming interchangeable distribution endpoints.

Contrarian view: better first-pass creative may compress the advantage of agencies and creative-software vendors, but it can also intensify auction competition. If cheaper asset creation causes advertisers to test materially more variants, conversion efficiency gains may be competed away through higher CPMs/CPCs; platforms capture the auction uplift, while advertisers may not retain the economic benefit. This favors the ad-platform duopoly over ad-tech intermediaries, though the effect is too small and unverified to justify a standalone position today.

Key Decisions for Investors

• No standalone trade on this announcement; the claimed quality improvements lack conversion and spend data, so estimated near-term earnings sensitivity for GOOG and META is immaterial.
• Maintain a 1-3 month watch item on META: add to long exposure only if management or channel data show accelerating Advantage+ creative adoption alongside stable/improving advertiser ROAS. Falsifier: higher automated-creative usage accompanied by rising CPMs and worsening conversion efficiency.
• Prefer META over GOOG as a tactical AI-ad-creative beneficiary if evidence emerges of long-tail advertiser budget expansion, given greater exposure to visual, performance-driven social inventory. Use a relative long META / short GOOG structure only after confirmation in quarterly commentary or third-party ad-spend data; key risk is YouTube/Performance Max showing faster asset-led budget growth.
• Monitor Adobe (ADBE) and agency-services proxies for a 6-18 month second-order downside: sustained platform-native creative automation could pressure pricing for basic production workflows. Do not initiate a short without evidence of seat churn, lower net retention, or reduced creative-cloud ARPU.