Cover Frame Autopsy: judge it at grid size
Two failures, and creators keep fixing the wrong one
Nobody opened it is a cover problem. They opened it and left is a retention problem. If your watch-through is fine but views are flat, rewriting the script will not help — the video is losing before it starts.
The sizes that actually matter
A cover is almost never seen at full size before the tap. It is seen in a profile grid cell, a feed shelf card, a search row. This kit judges your candidates at those effective sizes, which is where most covers die: designed full-screen, reviewed full-screen, shipped — then displayed at a fraction of that size where the text is unreadable and the subject unrecognisable.
It also handles the crop. Covers get re-cropped per surface, and a 9:16 cover in a 1:1 grid cell loses the top and bottom. If your text lives there, it is gone. The contact sheet shows you the crop, not just the scale.
What you get
1. cover-contact-sheet.html — one self-contained file. No external stylesheets, scripts, fonts, or images; no network requests. Every candidate rendered as a layout block at real grid, shelf, and search sizes with your exact text set proportionally, the crop overlay drawn on top, and a side-by-side ranking.
2. A legibility score per text element — character count, line count, contrast, crop safety, and redundancy. Each gets a pass, warning, or fail with the specific fix: not "make it bigger", but which words to cut and what the shortened line should be.
3. A ranked pick with the criterion that decided it. If two candidates are genuinely close it says so rather than manufacturing a winner, and tells you what to test to break the tie.
4. Redraw instructions for the runner-up — the one fixable defect that put it second.
5. A cover system — the winning decision generalised into a repeatable rule set for your account: text position, maximum character count, contrast treatment, and how much consistency to keep between covers.
Honest limits
- It does not process images. You describe your candidates in words and paste the exact on-cover text. Nothing is uploaded, no account is connected, no API key, no subscription.
- Never invents a click-through or impression figure. Missing numbers are marked with where to find them.
- Promises nothing about views, reach, click-through, or revenue.
- If your watch-through is also weak, it says so — the cover is not your bottleneck, and it points you elsewhere instead of selling you a fix you do not need.
- Declines, with the reason: buying views or engagement, bot or multi-account automation, evading platform enforcement, using someone else's footage or likeness, and covers that misrepresent the video's content.
Related
Retention Check (Free) tells you from one number whether a published video's watch-through is healthy — run it first to confirm the cover really is the bottleneck. Retention Rebuild Kit does the full second-by-second teardown when it is not. Both are on my publisher page.
