Scrollsy
Diagnose free

Your last 30 posts know exactly why your reach is stuck.

Every post measured against your own median. The habit dragging it down, named with numbers. The fix, verified seven days later.

Diagnose my account freeSee how it works
@

@you

Instagram · 12 posts · 36 days

example

Diagnosis

@you doesn't have a content problem, it has an opening problem: WEAK openers run 0.51x your median across 3 posts, CONTRAST openers run 1.68x across 3.

high confidence · Based on 12 measurable posts over 36 days, without retention data.

Opens with a greeting or wind-up

0.48x · n=4

WEAK opener

0.51x · n=3

Opens cold, first line is the point

1.54x · n=8

CONTRAST opener

1.68x · n=3

1.0x = this account's own median (13,450 views). Buckets under 3 posts are anecdotes and are shown dimmed in the real report.

This week, one change

For the next 7 days, every video's first spoken sentence must be a specific claim or number. No greetings, no context, no 'so um'.

re-measured against the same baseline in 7 days

Instagram Reels TikTok

A public profile is enough. Nothing to connect, nothing gets posted.

A creator at a desk, phone in hand, checking a video

Every tool out there decodes somebody else's winner.

Their audience, their timing, their luck. The only honest control group for your reach is your own last thirty posts.

A loop, not a report. Diagnose, change one thing, verify.

1

Diagnose

Thirty of your own posts, transcribed and tagged. Every one measured against your median, then the recurring habit named with the numbers behind it.

The verdict

@you doesn't have a content problem, it has an opening problem: WEAK openers run 0.51x your median across 3 posts, CONTRAST openers run 1.68x across 3.

Throat-clearing openers

Posts that open with a greeting or wind-up run a median 0.48x baseline (4 posts) vs 1.54x when the first line is the point (8 posts).

Self-promo before value

Every below-baseline vlog mentions the program or link inside the first half; none of the above-baseline posts do.

example · high confidence · 12 posts

2

Change one thing

The diagnosis becomes a do-and-don't protocol for the week. One variable, so the result can mean something.

This week's protocol

Open with the claim

Do

  • +First spoken sentence is a claim or a number
  • +Cut any greeting, context or 'so um' before it
  • +Hold everything else constant for the next 4 posts

Don't

  • -“Hey guys, welcome back to the channel”
  • -“So today I want to talk about”
  • -Pitching the program before the payoff

example · one variable, held for at least 4 posts

3

Verify against real views

Your account is re-read a week later and the change is ruled on: validated, failed, or inconclusive. Then the next habit.

Seven days later

Re-measured against the same baseline

failed ≤ 0.85xinconclusivevalidated ≥ 1.25x

Fewer than 3 new posts, or a move inside the grey band, is ruled inconclusive. The verdict is never softened to flatter you.

The niche corpus

Your numbers, next to creators your size in your niche.

A median tells you what is normal for you. The corpus tells you what is normal for accounts like yours, computed the same way: public creators in your niche, run through the same math, kept as medians across accounts. Never a leaderboard of celebrities.

  • +Seeded from public creators in eight niches, sized like you.
  • +A niche shows up only once enough accounts contribute. No slice, no claim.
  • +Grows with every diagnosis. Every user makes the next reading sharper.

fitness · vs the niche median

example values
metricnicheyou
  • Shares per view0.0110.0058
  • Comments per view0.0080.0060
  • Opens cold vs greeting, gap0.9x1.06x
  • WEAK openers, median0.62x0.51x
  • Posts per week3.12.33

Rates are per view; multipliers are against each account's own median. The diagnosis quotes both sides verbatim and may cite the niche only as “creators in your niche”.

What it refuses to do.

Creator tools sell certainty. This one is built to be right when it speaks and quiet when it cannot be, which is the only way a verdict stays worth reading a month from now.

  • It will not invent a number.

    Every figure is computed in code from your posts. The model writes sentences around numbers it was handed; it never produces one.

  • It will not call two posts a pattern.

    Under three posts is an anecdote and is labelled one. Under twelve posts, the whole diagnosis is marked low confidence.

  • It will not flatter you.

    A change that moves reach by ten percent is ruled inconclusive, and the verdict says so instead of calling it a win.

  • It will not touch your account.

    Public posts only. It never logs in, never posts, never messages. There is nothing to connect and nothing to revoke.

Pricing

One free diagnosis. Then pick a cadence.

Trial Pass $5 once

Two weeks of real access. One-time, no auto-renew. Expires after 14 days.

  • 3 full account diagnoses
  • 25 pre-post checks
  • 10 winner decodes
Get the pass

Starter

$15/ month

Find the leak. Check before you post.

  • +2 diagnoses / mo (re-measure every two weeks)
  • +40 pre-post checks / mo
  • +10 winner decodes / mo
  • +Retention screenshot reading
Start with a free diagnosis

Creator

runs the loop

$25/ month

Run the loop. Compound weekly.

  • +Weekly re-diagnosis (5/mo)
  • +The protocol + verified verdicts
  • +100 pre-post checks / mo
  • +25 winner decodes / mo
  • +Retention screenshot reading
Start with a free diagnosis

Pro

$40/ month

Post daily. Check everything.

  • +10 diagnoses / mo
  • +The protocol + verified verdicts
  • +250 pre-post checks / mo
  • +50 winner decodes / mo
  • +Retention screenshot reading
Start with a free diagnosis

Prices in USD. Cancel any time from Settings; the pass never renews.

Questions

Do I have to connect my Instagram or TikTok?

No. Type your handle. Scrollsy reads the public posts anyone can see, never logs in, and never posts or messages on your behalf.

How many posts does it need?

Five public videos is the minimum to run. Twelve or more gives a medium-confidence reading; twenty or more with two clear gaps gives high. Below that the report says low confidence in plain words.

How is this different from Sandcastles or 1of10?

They find other creators' outliers so you can copy the format. Scrollsy measures your own account against itself, names the habit that is costing you reach, and re-measures a week after you change it. Study a winner exists here too, as a side tool.

What is the niche corpus?

Public creators in your niche, sized like you, run through the same math. It puts a niche median next to your number so you know whether 0.3 views per follower is normal or a problem. It only appears once enough accounts contribute.

What if the diagnosis is wrong?

It shows the numbers it used, per bucket, with the post count. Under three posts is labelled an anecdote. If you disagree, you can see exactly where, which no coaching call gives you.

Which platforms?

Instagram Reels and TikTok today. YouTube Shorts is next.

Run the free diagnosis. Read what your account has been telling you.

Diagnose my account freeNo card. No password. Public profile only.