Method · CTR prediction
No tool can promise "this thumbnail will get 7.2%". What a good score does is rank your drafts on the signals that move click-through, before the feed does it for you.
A thumbnail's click-through rate depends on the image, the title beside it, the audience it is shown to, the competing thumbnails in the same feed, and the time of day. Only the first of those is known before upload. So an honest CTR prediction is not a number; it is a ranking of your own drafts on the image-side signals that correlate with clicks.
That is what the Pixado Score is. It measures six signals at feed size — face dominance, emotional intensity, curiosity gap, text readability, colour contrast, visual clutter — and combines them into a score you use to compare draft A against draft B. Treat it as a pre-flight check, not a forecast.
Each one correlates with click-through across large sets of thumbnails. None of them predicts your title or your audience.
Share of the frame a face occupies, and whether the expression still reads at 168 pixels.
Neutral expressions score low; surprise, tension and joy score high because they read instantly.
Whether the image poses a question the title answers, or resolves it and leaves nothing to click for.
Character height, weight and contrast measured at feed size. The most common failure on otherwise good drafts.
Separation of subject from background, and from the light or dark feed around the thumbnail.
Number of competing focal points. One is right; two is usually one too many.
Not as an exact number, because CTR also depends on the title, the audience, the competing thumbnails and the time. What a scoring model can do reliably is rank your drafts on the image-side signals that correlate with clicks — face, expression, text readability at feed size, contrast, clutter, curiosity gap. That ranking is what you need before upload.
Accurate enough to order drafts, not to forecast a percentage. Two drafts scoring 82 and 61 will, on average, differ in real CTR in that direction; two scoring 82 and 79 are a coin flip that YouTube's A/B test should settle.
It is the final word, but it needs impressions to reach significance, which costs the first hours of a video — exactly when the feed decides how hard to push it. Scoring first narrows three drafts to the two worth testing, so you spend that window on stronger candidates.
No. It reads the image only. Pair it with a title that answers the question the thumbnail poses; a strong image with a title that spoils it will underperform its score.
Higher than your last upload. Absolute thresholds mislead because niches differ — a faceless finance thumbnail cannot score on face dominance and does not need to. Use the score to beat your own previous thumbnail and to fix the single weakest signal each time.
Every plan unlocks the full studio. See the plans and volumes.