Know why a post worked, not just that it did

Sociolark reads every comparison post the same way: the hook, the call to action, tone, specificity, whether the caption asks a question, what is on the cover, the format, length and posting time. Then it measures how posts with each trait perform against posts without it, on your account only.

Each trait is tested against chance, with a likely range and the number of posts behind it. Traits that could be luck are kept out of the reasons and listed separately, with the experiment that would settle them. The AI writes the explanation, but every number comes from the data.

• Compared with up to 60 posts, same format when you have enough
• Posts younger than three days are flagged as still growing
• Plain-English confidence: strong evidence, likely, or not proven yet

For Reels, Shorts and videos, a video model watches the footage. It describes the first three seconds, scores the hook, counts the cuts, notes the shots, the audio, captions and on-screen text, and where the call to action lands. You get a timeline of key moments and timestamped notes on what to change.

Where the network reports it, the analysis adds watch time, skip rate and the retention curve, so you can see exactly where people left.

Comments are sorted into praise, questions, objections, complaints and buying intent, grouped by topic, with a few anonymised quotes. Unanswered questions are pulled out so you can reply or make your next post about them.

Every analysis ends with a recipe built from your proven traits: hook, format, length, visual, call to action and timing. It adds three rewritten captions in your brand's voice and, for video accounts, three shoot-ready briefs with a hook script and shot list. When a post underperformed, you get a diagnosis and specific fixes instead.

Run a Content playbook to do the same across your whole account: what to do more of, what to stop, your best posting times and format mix, and what your top five posts have in common.