Nerdy Doctor

Your content,
diagnosed.

Every video you have posted, scored against your own median — then the exact five to film next, with the hook written word for word.

  • One health score, out of 100, that moves month to month.
  • Autopsies of your 10 best and 10 worst posts — what worked, what died, and why.
  • Five briefs to film — and a scoreboard next month showing what each one scored.
InstagramYouTubeTikTokYour handle is all we need. No password, no access to your account.
The measurement

Every post is scored against your own median

3× · outlieryour medianoldestnewest

Outlier score = views ÷ your median. You are never compared to a creator with ten times your audience. A 4,000-follower account and a 2M-follower account get the same maths.

The visit

Three steps, once a month

  1. Paste your handle

    One field. We detect the platform. No password, no login to your account, nothing to install.

    We only read posts that are already public. Connecting the account later is optional — it unlocks watch-time and saves, which public data does not expose.

  2. We examine every post against your own median

    The whole back catalogue comes in. Each post gets an outlier score: views ÷ your median. Then we watch the winners and the flops frame by frame — the opening seconds, the on-screen title card, the caption.

    The hook is transcribed as it was spoken, in your language. The title card is copied exactly as it was typed. Nothing is paraphrased into English and lost.

  3. You get a diagnosis and the five videos to film next

    A health score out of 100, the autopsy of what worked and what died, the content pillars to move toward, and five briefs — topic, angle, and the first five seconds written word for word.

    Next month the report opens with the scoreboard: which prescriptions you filmed, and what each one scored against your median.

Then it repeats. The follow-up visit is the point: a single number trending in one direction, and evidence that last month's prescription either worked or did not.

The deliverable

What you get, every month

Four things, every month. Here they are, filled in, so you can judge the depth before you pay for it.

Sample report

A worked example, built to show the format. Not a real client, not a real account — the figures are illustrative.

1 · Account Health Score

One number, six components, all checkable

Treatable condition
-7 ptsvs last check-up
Sample Account Health Score components, summing to 57 out of 100
ComponentPtsThe receipt
Reach efficiency15/25median views ÷ followers = 4.2%
Momentum3/20recent-20 median ÷ the period before it = 0.58 — the slump
Consistency8/1070% of posts land within 0.5×–2× of the median
Engagement quality9/15median (likes + 3× comments) ÷ views = 1.3%
Outlier power8/10four posts at 3× or better in 60 days; best 4.8×
Creative fitness14/2010-item rubric; points lost on stakes, length and CTA hygiene
Total57/100Band 50–64: a working formula plus an active, fixable slump.

Most components are normalised against this account's own median, so next month's movement measures real progress rather than the size of somebody else's audience.

2 · Autopsy — the winner

The three-morning-foods protocol

4.8×
Angle
A daily habit, then a numbered correction. Something they did this morning, then the fix.
Hook · 0–5s
توقف. ثلاثة أطعمة تأكلها كل صباح توقف نزول وزنك“Stop. Three foods you eat every morning are stalling your weight loss.” Talking head, urgent gesture, burned-in caption.
Title card
٣ أطعمة توقف نزول وزنك“3 foods that stall your weight loss” — three stacked lines, yellow on white.
Topic
Everyday breakfast swaps — the audience ate these an hour before watching, so the stakes are immediate and personal.
Packaging
Title card, caption and first spoken line all make the same promise. One promise, three times.
Why it worked: a habit they performed an hour earlier, plus a numbered correction, delivered in the set that signals authority. Comments ran this account's usual rate — that is the audience asking “wait, which three?”, which is the sound of a nerve being hit.
2 · Autopsy — the flop

Let’s talk about nutrition

0.2×
Hook · 0–5s
No spoken hook.Three seconds of silent b-roll — a chopping board, a knife, a bowl — before the first word arrives. The winner has already made its promise by then.
Title card
نصائح غذائية“Nutrition tips.”
Topic
General healthy-eating advice. The post never narrows.
Diagnosis
Abstract stakes. Which advice, for whom, toward what result? No named food, no number, no deadline. An audience that answers to “three foods” and “every morning” has nothing to hold.
Why it failed: same studio, same editing, same edit rhythm as the 4.8× winner. The only difference is the promise. It died in the first three seconds, before the algorithm ever pushed it. This is a packaging disease, not a content disease — and packaging is cheap to fix.
3 · The Compass

Where the next month of content should go

Sample
Prescribed mix
  • Authority Protocol35%
  • Stakes Education20%
  • Proof-of-Self20%
  • Community Loop15%
  • Newsjack10%
Your last 30 days

The gap: 62% of the last 30 days went to one pillar, and 0% went to Community Loop — a pillar this account has never tested, on an audience that asks questions constantly. Two slots move this month.

4 · Prescription scoreboard

Next month opens by marking its own homework

Sample
  • ℞1Answer the late-dinner question on cameraCommunity LoopScored4.4×
  • ℞2The one-plate ruleAuthority ProtocolScored2.1×
  • ℞3Lemon water, on the recordAuthority ProtocolScored0.9×
  • ℞4Cook every meal on camera for a weekProof-of-SelfPosted
  • ℞5Reserved slot — the week's trending claimNewsjackPrescribed
+38%vs untreated posts

Two of the three scored prescriptions beat the account median. Cancelling now feels like stopping medication that is working — which is exactly the point.

Sample report · illustrative figures · not a client account.

Why you can act on it

The rules the report is held to

A report that fails any of these four does not ship. That is the whole difference between this and asking a chatbot to look at your feed.

Every number has receipts

Each component of your score, each pillar ranking, each verdict opens to the exact posts it was computed from. If you cannot re-derive a number from the evidence we show you, it should not be in the report.

Nothing is estimated

A missing source produces an honest gap and a clear message, never a plausible-looking number. Watch-time and saves are not public on Instagram or TikTok, so for a public account we do not report them — we say so, and we say what we used instead.

Hooks are quoted verbatim

The first five seconds are transcribed as spoken and the title card is copied as typed — Arabic stays Arabic, right to left, in a real Arabic typeface. Translation sits underneath, never on top.

Every verdict cites both

A creative observation and a metric. “Strong hook” is not a finding. “The withheld-name hook paid 4.3× when the thing withheld was a breakfast food and 0.4× when it was a supplement, because one is already on the table in front of them and the other is not” is a finding.

The same flopped video, explained twice
A generic AI tool

“Your hook creates curiosity but could be stronger. Try adding more emotion and a clear call to action to boost engagement!”

True of every video ever made. Nothing to do on Monday morning.

Nerdy Doctor

“Abstract stakes. Same studio and same editing as your 4.8× winner — the only difference is the promise. Your four worst posts all lead with an abstract noun; your five best all name a body part and a number. Re-shoot the title card, not the video.”

Falsifiable, specific to your account, and filmable this week.

What we cannot see, and say so

For a public account, platforms do not expose watch-time, retention curves, saves or shares. We infer nothing from them and we never present an inference as a measurement. If you connect the account, those numbers become real and the report says which ones are measured and which are not.

Before you sign up

The questions people actually ask

Which platforms do you cover?

Instagram Reels, TikTok and YouTube — all three, today. Short-form is compared to short-form and long-form to long-form — a Reel is never scored against a 12-minute video.

YouTube runs end to end on the public Data API. Instagram and TikTok have no open API for an account you do not own, so collection goes through an established provider. When a collection fails we tell you it failed and when we will retry. We do not quietly serve you week-old numbers.

A TikTok chart is the more complete one. TikTok publishes a follower count with every post and Instagram does not, so reach efficiency — how far past your own audience the work travels — can only be scored on TikTok and YouTube. An Instagram chart says so rather than guessing at it.

What do you need from me?

Your handle. That is the whole intake. No password, no OAuth, no browser extension, and no access to your account — we read posts that are already public, the same ones any of your followers can see.

Connecting the account is optional and comes later. It unlocks the numbers platforms keep private — watch-time, retention curves, saves and shares — which turns several inferred findings into measured ones.

Is my data private?

Your diagnosis lives inside your own workspace and is visible to you and to the people you invite. Nothing about your account is published, and we never share a report or a case study without your written permission.

Every read is scoped to your workspace on the server, security events are logged, and nothing is ever deleted silently. If you want everything removed, ask and it is removed.

My account is small. Is this for me?

Yes, and it works better than you expect, because every score is relative to your own median. A 4,000-follower account gets the same arithmetic as a 2M-follower one, and its outliers are just as real.

The one requirement is a catalogue: a median needs roughly 20 published posts of the same format before it means anything. Below that we say the sample is too thin rather than printing a confident number on top of noise.

What does it cost?
  • Diagnosis $197 once. One full report, one platform, one account.
  • Treatment Plan $397/month. A report and prescriptions every month, follow-up scoring, two platforms, pre-flight hook checks. This is the core product.
  • Full Care $797/month. All platforms, connected-account deep metrics, competitor scans, reports on demand.
  • Clinic $297 per seat per month, minimum 5. White-label reports for agencies managing a roster.

A single month of Treatment Plan costs less than one filmed video that nobody watches.

Can I cancel?

Any time, yourself, from the billing portal. No email, no retention call, no notice period.

When the paid period ends the workspace goes read-only rather than dark: every report and every document you uploaded stays viewable and exportable for 60 days. Nothing is deleted silently, ever.

Stop guessing which video to film

The answer is already inside your own feed. It takes one handle and one examination to read it back to you.

Nerdy Doctor — your content, diagnosed