Other tools study strangers' winners. Scrollsy measures every video you post against your own baseline, the median of your own recent posts, then names the recurring pattern that drags you below it and verifies the fix against your real numbers.
Drop your @handle. Scrollsy pulls your last ~30 posts, transcribes the top and bottom performers, and computes a per-post multiplier against your own median views — then buckets those multipliers by hook type, topic, format, duration, CTA, and posting slot.
The verdict names your suppressors with computed evidence, not vibes: the numbers come from a deterministic stats engine, and the write-up is only allowed to cite them. Buckets with fewer than 3 posts are ignored instead of extrapolated.
Fig. 1 — one account, 90 days
views vs own median
Same person, same niche, same audience — 30 different results. The pattern in the losers is the answer no winner-decoder can see.
Before you spend an hour filming, paste the draft. It gets scored against your own winning and losing patterns, not a generic rubric, and comes back post it, fix first, or rework, with the offending lines quoted and rewritten in your voice.
draft · repeats your 0.4x pattern
"Have you ever wondered why your reels die?"
rewrite · your 1.9x pattern, your voice
"Your reels die at 3 seconds. Here's the autopsy."
The diagnosis ends in one change for next week: a single variable, written as do/don't instructions. Seven days later Scrollsy re-scrapes your account and rules the experiment validated, failed, or inconclusive against your baseline.
It is deliberately conservative: fewer than 3 posts, or a result inside the 0.85-1.25x noise band, is called inconclusive instead of a win. Trust in the loop is the product.
No platform exposes retention curves through an API, to anyone. Scrollsy reads the chart straight off your TikTok Studio or IG insights screenshot and folds hook-hold and average-watch numbers into your next diagnosis. It's the one signal that names a hook problem decisively, and only you can supply it.
Public creators in your niche run through the same pipeline, so the diagnosis can put your number next to the niche median and show what is breaking out. Below five accounts in a niche it says nothing rather than guess.
Paste any reel URL or handle and get the craft breakdown (hook type, structure, delivery, CTA) plus a comparison against your own voice: what to borrow, what to keep yours.
Train a voice profile from your own posts. The pre-post check and the weekly protocol rewrite lines in your phrasing, not in a generic one.
| Instrument | Free | Pass · $5 once | Starter · $15/mo | Creator · $25/mo | Pro · $40/mo |
|---|---|---|---|---|---|
| Account diagnoses / mo | 1 | 3 | 2 | 5 | 10 |
| Pre-post checks / mo | 3 | 25 | 40 | 100 | 250 |
| Winner decodes / mo | 3 | 10 | 10 | 25 | 50 |
| Protocol + verified verdicts | no | no | no | yes | yes |
The $5 pass is one-time and expires after 14 days. Monthly counters reset each calendar month.