Original research across 10,000 Kompozy outputs. Platform-by-platform engagement, format-by-format CTR, autopilot vs manual review quality.
Every "AI content works" post online cites no data. Every "AI content is slop" post cites no data either. This post cites data from 10,000 Kompozy outputs across 500 opted-in workspaces, anonymized, to give you actual numbers.
Engagement rate per platform (likes plus comments plus shares divided by impressions). Click-through rate on paid-adjacent formats (POV Hook to Demo). Autopilot approval rate (percent of outputs shipped untouched after the Persona Brief ramp). Output quality self-report (Pipeline Queue thumbs-up rate).
TikTok: 3.8% average engagement on clipped shorts, 2.1% on avatar shorts, 1.4% on template shorts. Clipped wins because TikTok rewards authentic source material.
Instagram Reels: 2.4% engagement on clipped, 3.1% on avatar, 4.8% on before-after split. Before-after wins; the algorithm privileges transformation narratives.
LinkedIn: 4.1% engagement on long-form posts, 2.9% on single-image posts, 5.3% on carousels. Carousels dominate; they earn dwell time no other format matches.
X: 1.8% engagement on standalone posts, 3.2% on threads of 4-6 posts, 2.1% on threads of 8+ posts. The 4-6 post thread is the sweet spot.
YouTube Shorts: 2.6% engagement on clipped from the creator’s own long-form, 1.3% on clipped from external sources. YouTube rewards the clip-to-long-form back-link signal.
Clipped shorts win when the source is genuinely your own long-form. They underperform when used to re-clip competitor content.
Avatar shorts underperform clipped on TikTok but match clipped on Instagram Reels and outperform on LinkedIn video.
Template shorts (HyperFrames) underperform clipped and avatar on average but win when paid-boosted — the POV Hook to Demo template delivers 3.1% CTR on paid cold traffic vs 0.9% on prior static creatives.
Carousels outperform single images on both Instagram and LinkedIn by 2-3x.
Days 1-7 (pre-Persona-Brief-tight): 54% of outputs shipped untouched, 46% edited. Days 7-14 (brief iterated from edits): 78% shipped untouched. Day 14+ (brief stable): 91% shipped untouched. The number that matters: 91%. That is the point at which autopilot becomes safe to enable on stable sources.
Posting the same output to every platform unchanged. Engagement drops 40-60% when you skip platform-specific formatting.
Running autopilot before the Persona Brief stabilizes. Autopilot-shipped posts with a loose brief earn 0.6x the engagement of manually-reviewed posts.
Cross-posting clipped shorts from external sources (competitor podcasts, for example). Platforms detect and down-rank.
LinkedIn is more AI-tolerant than X. LinkedIn audience skips the "AI written" check; X users call it out. Carousels beat video on LinkedIn. Counterintuitive, but true across our dataset. Avatar shorts underperform on TikTok but dominate on LinkedIn video.
Data pulled from 500 opted-in Kompozy workspaces across 10,142 outputs published between February 1 and April 15, 2026. Engagement rates normalize for audience size. Outliers (top 2% accounts with 10x+ follower counts) excluded to prevent skew.
Yes, anonymized across opted-in Kompozy workspaces. Individual account data stays private.
Quarterly. Algorithm changes make older data unreliable past about 90 days.
The summary tables here are the public version. Enterprise accounts can request the anonymized raw data under DPA.
AI content performs when the Persona Brief is tight, the output is native-formatted per platform, and autopilot only runs after 14+ days of manual review. Under those conditions, engagement matches or exceeds manually-written content across every platform we measured except X (where audience vigilance against AI is highest).
Start a free trial → · ← All guides · Compare Kompozy vs other tools