How ViralLoop reads a post

Every post page on this site says why the post worked — its opening move, its structure, the mechanisms it leans on, and a few plain sentences a marketer could act on. This is what produces that, and what does not.

What has to be true before a post gets in

A post has to have travelled. That means real public engagement — likes and replies as the platform reported them, captured with the date — and a floor under it, so the library is not a random sample of the internet but a sample of what people actually passed on. The floor is deliberately not high: a post from a small account that did well for a small account is more instructive than a celebrity's throwaway, and the ranking (below) makes sure it can outrank one.

A post also has to be readable: the text, not just a link. Posts we hold only by identity — a URL whose content we could not fetch — are shown as the platform's own embed and marked as pending, never analysed, never remixed.

And a post has to teach something. That is the analysis's job to decide, and posts where it cannot find a structure or a mechanism are kept out of the feed.

What is measured

The public numbers, as published, with the platform named — and only the ones the platform actually publishes. This matters more than it sounds. Instagram and TikTok publish no like counts through the routes we can read; an X post publishes likes and replies but not, reliably, reposts or views. Where a number is not published we record that it is not, and we never treat a missing number as zero. A filter on "at least a thousand likes" keeps posts whose platform publishes no like count, and says so, because the alternative is a filter that silently removes two platforms.

From those numbers, two scores. The viral score is engagement, honestly measured and labelled as such. The editorial score is what the library ranks by: it blends the engagement with what the analysis found about the post's craft, so reach alone cannot win. A well-built post from a small account sits above a badly built one from a large account. That is a choice about what the library is for.

What is read

Each post is read for its structure — the beats it moves through, labelled and described — and for how it opens, which we sort into eleven kinds (the pattern articles cover each). It is read for the psychological mechanisms it leans on: curiosity gaps, social proof, specificity, vulnerability and so on — eleven of those too. And it is read for its topic and, where it has one, the industry it is about.

Then two things are written for a person rather than a database. Why it worked: a few plain sentences about what the post did that made it travel. How you would remix it: what carries over if you wrote your own post from its shape — and, always, what does not.

What is deliberately not done

Nothing is rewritten. The post's text is the post's text, verbatim, with its author's name, handle, platform and a link to the original on every surface it appears — the card, the page, the studio.

Nothing is invented about the author. Follower counts appear only when the platform published them. We do not estimate audiences.

Nothing is copied into a draft. The remix engine is built so that a draft cannot contain a source's facts, figures, names or phrasing. It works from the structural reading — the beats, the opening move, the mechanisms — and from the user's material, and where a fact is needed that the user did not supply, the draft leaves a bracket rather than making one up.

Why say all this

Because the product's whole premise is that you can learn from other people's work without taking it, and that premise only holds if the line is drawn clearly and kept. The line is: the shape of a post is knowledge, and knowledge is shareable; the words of a post are the author's, and they stay that way. Everything above is in service of that distinction.