Mochi

Based on 6 saved ads, observed 2026-10-17. This is a curated selection capped at 50 ads, not a random sample. Percentages describe saved ads, not impressions or budget.

Angles are assigned automatically using rules written specifically for Mochi. Rules run in order: first match in the headline, then in the body. This classifies saved copy; it does not include scene analysis or video transcription.

Formats are based on saved files. Multiple images alone do not confirm a carousel. Ad age uses 6 known start dates; 0 are unknown. Days are measured from Meta’s start date to each ad’s recorded observation date. Age does not establish continuous delivery or profitability.

Video duration was measured from MP4 metadata for 6 of 6 unique video files. Exact duplicates count once. CTA labels are captured button text, not verified user actions; unknown labels are included in the distribution.

Offer signals detect words in saved body copy and headlines. Labels overlap and are not verified advertising claims. Unsaved variants may use different formats and CTAs.

Distinct media sets use exact file SHA-256 hashes: 6 sets, with 0 repetitions beyond the first occurrence. Visually similar variants may have different hashes. Text comparison ignores URLs, letter case and extra whitespace. Different headlines count as different text. Repeated copy does not prove a controlled A/B test.

Impressions, spend, CTR, conversions and ROAS are unavailable. A single snapshot cannot establish stopped ads, growth or turnover. Ad count is not campaign count.

All ads and labels · CSVFull analysis · JSON