Skip to content
Pre-publish diagnostics for short-form video

Find out why it will flop. Before you post it.

HitRadar scores your video against your own account’s history and returns the exact factor that’s about to kill it — the weak hook, the 11-second dead zone, the missing reason to share. With the fix, and the timestamp.

Free while in private beta. No card. One email when your invite is ready.

See a sample report

A HitRadar readout: breakout potential 86 out of 100, rated high potential, with hook strength 85, retention 82, rewatch value 79 and breakout likelihood 88 — 37% above this account’s average.
Works with TikTok · Instagram Reels · YouTube Shorts Every threshold measured against your own baseline — grounded in 22,666 randomized headline tests

You’re not short on ideas. You’re short on feedback.

Analytics tell you a video failed after it failed. By then the only lesson available is “that one didn’t work” — not which of the five things didn’t work. So the next video repeats the same structural mistake, and the one after that.

Blind publishing.

You find out at 400 views, three days too late.

Generic scores.

A 0–100 from a tool that has never seen your channel is a horoscope.

No transfer.

Nothing you learn from one video makes the next one better, systematically.

How it works
01

Connect your account

We pull your last 30–90 posts and build your baseline: your median views, your typical hold rate, your niche vector. Every number after this is measured against you.

02

Drop in a video or a script

Upload the cut, or paste the script before you shoot. HitRadar transcribes it, reads the on-screen text, maps the shot rhythm, and scores five factors.

03

Get a verdict and a fix list

PUBLISH or REVISE, the weakest factor named, and an ordered list of blocking issues with timestamps. Then publish — and the result feeds back into your baseline.

What’s inside

Four tools. Each one tells you what it can’t see.

These are the real panels, with real sample data. Pick a tool, then pick anything inside it — every score on this page explains where it came from.

Trend Radar — find proven demand

Surfaces videos performing far above their own channel’s median, then mines the transcripts and comments for the pattern underneath — so you copy the mechanism instead of the topic.

Measured against real audience behaviour. This is the accurate half of the product.

Outliers worth copying

4 /5

Last 7 days

Outliers in your niche

10×+ worth modelling 5×+ worth deconstructing below 5× watch only

The pattern underneath

@studio_nine 11.4× their median

“Three shots, no talking”

The whole video is one continuous action with the payoff withheld until the final second. No narration at all — the on-screen count does the storytelling, so it survives being watched on mute in a feed.

  • silent cold open
  • on-screen counter
  • single take

Video Analysis — the pre-flight score

One verdict from five factors: Idea, Hook, Retention, Emotion, Share Trigger. Multiplicative, so any single near-zero forces a REVISE. You get the score, the weakest link, and the reason.

Reliable at finding failure modes. Weak at predicting hits — and it says so.

Overall Score

67 /100

Needs Revision

Score Breakdown

0.75 0.82 0.54 0.70 0.60

0.67 (67/100)

Geometric mean of the 5 factors — one weak factor drags the whole score down, which is what “multiplicative” means here.

Weakest Link

Retention 54/100

Why?

Expected retention curve drops too fast after 15s. No new visual information arrives between 00:15 and 00:26, and the payoff sits behind it.

Factor Impact

Idea Hook Retention Emotion Share Trigger

Hook & Packaging — the evidence-backed layer

Hook length, preamble, archetype, title front-loading, thumbnail squint test and engagement-bait detection — every rule traceable to a published effect size.

Directional. Ranking is relative to your other variants, not to the world.

Best variant vs current

+0.15 score

22,666 tests

Rewritten hooks

CURRENT “Here are some tips for editing your videos faster” 0.72

Packaging checks

demonstration 0.87 +0.15

  • Number in the first four words
  • No preamble before the claim
  • Keyword front-loaded in the title
  • Thumbnail squint test — text is 6% of frame height
  • No engagement bait detected

Grounded in randomized headline experiments where the best variant beat the worst by a median of 1.7× CTR.

Breakout Detector — the first six hours

After you publish, HitRadar watches early completion, send rate and reply velocity against your own trailing baselines — so you hear about a breakout while you can still act on it.

Measured, not predicted. This is the tier that is genuinely predictive.

Strongest signal

+2.4 σ vs baseline

Breaking out

First six hours vs your trailing baseline

your baseline 0h 1h 2h 3h 4h 5h 6h

Early completion Completion is climbing rather than decaying, which is the shape that precedes a push. At 2.4σ this is outside anything this account has done in the trailing 30 posts.

Why it matters now

6h the window where a pinned comment, a reply pass, or a cross-post still changes the outcome.

Every threshold here is a z-score against your own trailing baselines, per format — not a global average. A video that is breaking out for you may be unremarkable for someone else, and that is the only comparison that changes what you should do next.

Your baseline. Not the internet’s average.

Every platform ranks your video against your own recent performance. So does HitRadar. Outlier score is views divided by your trailing-10 median. Breakout tiers are z-scores against your own baselines, per format. A tool that scores you against a global average is measuring something no algorithm cares about.

@channel_a · same video
50,000 views
4.2× OUTLIER
@channel_b · same video
50,000 views
0.6× — below your median

What HitRadar can’t tell you.

Most virality tools quietly imply that a high score means you’ll win. Ours doesn’t, and we’d rather lose the sale than pretend. A high pre-publish score means “no known reason to fail.” It is a floor, not a ceiling. No model that hasn’t shown your video to an audience can promise you a hit — including this one.

  • Retention analysis is structural. It’s blind to your delivery, your edit energy, and your audio.
  • Pre-publish scores predict failure modes well and hits poorly.
  • Trend mining and breakout detection are the accurate half, because they measure real behavior.

Every score in the product ships with its accuracy class and a one-line statement of its blind spot. That’s a product feature, not a disclaimer.

This is the whole output.

REPORT · anon_creator · reel_038.mp4
0.41 REVISE
idea
0.81
hook
0.34
retention
0.18
emotion
0.72
share_trigger
0.55

‘retention’ scores 0.18, below the 0.20 floor — an 11-second dead zone forces a REVISE regardless of the other factors.

BLOCK

00:19–00:30 — dead zone: no new information, no visual change for 11s. Cut to the payoff or intercut B-roll.

WARN

00:00–00:04 — hook states the topic, not the tension. Lead with the outcome, then explain.

INFO

title — front-load the keyword; the first 30 characters carry the CTR weight.

No dashboard tour required. One video in, one page out.

Who it’s for

Solo creators

One concrete, timestamped fix per video, instead of publishing on feel.

Agencies

Multi-account baselines and a repeatable pre-publish quality gate.

Brand teams

A go/no-go REVISE verdict before the edit locks and the spend commits.

Strategists

Proven-demand ideas from outlier mining, not brainstorms.

Questions

Does this actually predict whether a video goes viral?

No, and be suspicious of anything that says it does. It predicts failure reliably and success poorly. That’s still the most useful thing available before you publish.

Which platforms?

TikTok, Instagram Reels, and YouTube Shorts.

Do I need to connect an account?

For account-relative scoring, yes — that’s the part that makes the numbers mean anything. You can run a one-off analysis without it, and the report will tell you which parts are weaker as a result.

What do you do with my videos?

Analyzed for your report, stored under your account, never used to train models sold to anyone else. Delete them any time.

How much will it cost?

Not decided yet. Waitlist members get early access and the founding price when it’s set.

When does it open?

Rolling invites through private beta, in signup order, weighted to people who tell us what they’re posting.

Stop finding out at 400 views.

Join the waitlist and get a free teardown of one video when your invite lands.

Free while in private beta. No card. One email when your invite is ready.