RetensisDashboard

Performance

Prediction Accuracy

See how close our retention predictions land to your real audience data, and let Retensis calibrate future predictions to your channel.

LiveGrowth and above

Overview

Prediction Accuracy closes the loop between what Retensis predicts and what actually happens. Connect your YouTube channel and analyze your own published videos, and Retensis compares its predicted retention curve against your real audience retention from your YouTube Analytics, then shows how close it was, video by video. As the samples add up, it calibrates future predictions to your channel so they get sharper over time.

How it works

  1. 1

    Connect and analyze

    Connect your YouTube channel, then analyze your own published videos by pasting their YouTube link so each one has a prediction to compare against.

  2. 2

    See the accuracy

    Retensis lines up its predicted retention curve against your real audience retention and shows how close it landed, both as an overall readout and per video.

  3. 3

    Watch calibration kick in

    Once a few of your videos are measured, channel calibration activates and quietly tunes future predictions to your audience.

What you get

  • An overall accuracy readout across your measured videos
  • A per-video breakdown of how close each prediction was
  • Channel calibration status once enough of your videos are measured
  • A direct link from each video back to its full analysis

Important details

  • A video appears here when it has a Retensis prediction and something real to compare it against: either measured retention from your connected YouTube Analytics, or a retention graph you imported from your own Instagram or TikTok insights.
  • The accuracy readout is measured against your real YouTube Analytics. The retention curve itself is an AI prediction, kept distinct from your real numbers.
  • Imported graphs are listed separately, with their own figure. They are read from a screenshot you supplied rather than reported by a platform, so they are never mixed into the measured numbers and never used to tune future predictions.
  • Channel calibration activates once at least three of your measured videos are in, and it learns from measured retention only.
  • On Instagram the comparison lives on the analysis itself. A Reel analyzed from your Instagram library is linked to the published Reel for you, and its analysis shows Instagram’s real skip rate beside the one Retensis predicted. For a draft you analyzed before posting, your dashboard suggests the Reel it became and you confirm it. Instagram’s skip rate is not a retention curve, so it is not part of the accuracy readout on this page.