Virality Score
A checklist and an AI read, shown separately. Not a predicted view count.
What the Virality Score is
The Virality Score is a 0-100 grade Bulk Scheduler gives a video before you publish it, made of two separate parts. The first is a deterministic checklist — fixed, inspectable rules about the file and its metadata, which score the same way every time because none of it is a model's opinion. The second is anAI read of the content itself, judging what a rule cannot: whether the opening earns attention, whether the idea survives being watched on mute, whether the payoff arrives before people leave. The two halves are shown separately rather than blended into one number, so you can tell whether you lost points on something you can fix in thirty seconds or on the idea itself. It costs one AI credit per check and is available on every paid plan.
What it is not: a view prediction
Nothing can honestly predict a view count from a video file, and we are not going to pretend otherwise. Views are decided by things the file does not contain — how your account has performed lately, what the platform is testing that week, who saw the first hundred impressions and what they did next. The proof is that the same clip posted from two different accounts gets wildly different results, and no property of the file changed between them. A tool that shows you "predicted views: 47,000" is showing you a number it invented. What a score can honestly tell you is whether a video has the properties that short-form videos which do well tend to share, and whether it is missing something cheap to fix before it goes out. That is a smaller claim, and it is the only one we make.
What the checklist actually checks
Hook length
How long before the first visual or spoken payoff arrives.
Aspect ratio and safe area
Whether captions or faces sit under the platform's own UI.
Caption presence
Most short-form is watched muted; no captions is a measurable drop.
Duration against the format
A 90-second Short and a 12-second Reel fail for opposite reasons.
Title and description quality
Whether the metadata gives the algorithm anything to match on.
Thumbnail contrast
For destinations that actually show one.
Questions about the Virality Score
What is the Virality Score?
The Virality Score is a 0-100 grade that Bulk Scheduler gives a video before you publish it, made of two separate parts. The first is a deterministic checklist: fixed, inspectable rules about the file and its metadata - how long the hook runs, whether captions are present, whether the aspect ratio is right for the destination, whether the title gives the algorithm anything to work with. The same video always scores the same on this half, because nothing about it is a model's opinion. The second part is an AI read of the content itself, which judges the things a rule cannot: whether the opening actually earns attention, whether the idea is legible without sound, whether the payoff arrives. The two are shown separately rather than blended into one mysterious number, so you can see which half you lost points on and whether it is something you can fix in thirty seconds.
Is the Virality Score a prediction of how many views I will get?
No, and it is important that it is not. Nothing can honestly predict a view count from a video file, because views are decided by things the file does not contain: how your account has performed recently, what the platform is testing that week, who happened to be shown the first hundred impressions and what they did next. Any tool that shows you a predicted view number is showing you a number it made up, and the giveaway is that the same clip posted from two accounts gets wildly different results. What a score can honestly tell you is whether a video has the properties that short-form videos which do well tend to share, and whether it is missing something cheap to fix. That is a meaningfully different claim, and it is the only one this feature makes.
What does a Virality Score check cost?
One AI credit per video or per post checked, and it is available on every paid plan including Starter. The credit is charged only when you explicitly ask for a check or a re-check - every screen in the dashboard that merely displays a score you have already paid for reads it from cache and costs nothing, so browsing your library never quietly spends credits. Scores are stored against the specific video, so re-checking is only worth a credit after you have actually changed something. For context, one credit is also what writing a title, description and tag set costs, and an AI thumbnail costs two; the full per-action table is on the credits page. If you are checking every clip in a large batch, that is one credit per clip, so it is usually worth scoring a sample first.
Why did stat snapshots have to ship before the score could?
Because the data the score needs was being destroyed every time it synced. Both of the tables holding engagement counts updated view, like and comment numbers in place, which meant each sync overwrote the previous number with the current one. That makes the single strongest signal in short-form - how fast a post gained views in its first day - impossible to compute and, worse, impossible to backfill, since nothing can recover a count that was overwritten last Tuesday. So an append-only snapshot table had to start recording history before any scoring work could begin, and it was written and shipped for weeks while nothing read it. That ordering is the difference between a score grounded in what your account actually does and one that grades a file in the abstract.
Included on every paid plan — see pricing and the per-action credit costs.
Last updated