InfluencerBooster - live demo

1. Timestamped transcript

One line per cue: M:SS, MM:SS or H:MM:SS, optionally in [brackets], then the words. Lines with no timestamp are appended to the previous cue (the skipped word count is reported below). Sample data is a transcript written for this demo, not a real creator's upload.

2. Largest drop and the retention curve

Move over the chart to read a second.

Solid line: estimated retention. Amber marker: the largest drop, drawn where its source seconds end. Click a drop in the sidebar, a dot on the chart, or a transcript line below to highlight the exact source lines.

3. Transcript, line by line

Word counts and words-per-second are computed from each cue. Highlighted rows produced the selected drop.

Measured vs modeled

Exact, counted from your text: timestamps, cue lengths, word counts, words per second, silence gaps, and the quoted words at any second.

Modeled, not measured: viewer retention. Real retention curves come from the platform's analytics for a published video; this page computes a deterministic pacing model from your transcript instead, and every number below is labelled as an estimate.

Retention model (estimated)

Each second decays as d = base * (1 + 3 * pacing deficit + 0.2 * pacing surplus), where deficit compares that second's words-per-second with the transcript average. Silence gaps decay at base * 4. Change the inputs and the curve is recomputed.

Base decay, percent per second:

Drop window, seconds:

Steady speaking rate, words per minute (used only to estimate the length of the final cue)

1. Timestamped transcript

Result refused

2. Largest drop and the retention curve

3. Transcript, line by line