Back catalogues quietly decay. Links break, offers end, titles reference things nobody remembers. A periodic youtube metadata audit repairs the videos that are still working, which is usually a small number carrying a large share of the views.
Start where the views are
Sort the library by views in the last ninety days and work down. A handful of videos typically account for most ongoing traffic, and fixing those returns far more than a complete pass through everything in upload order.
Videos that still attract viewers years later are the ones where stale metadata costs most, because they continue to receive new people who meet a dead link or a reference to something discontinued.
What to check
- Every link in the description resolves, and still points somewhere sensible.
- No references to ended offers, old pricing or discontinued products.
- Title still describes the content in current language.
- Chapters present on longer videos that lack them.
- Captions reviewed rather than left as raw automatic output.
- Playlist assignment still correct after any restructuring.
- End screen destinations still exist and still make sense.
- Thumbnail still legible and consistent with current channel style.
Change carefully on working videos
A video performing steadily deserves cautious treatment. Change one element, note the date, and observe before doing more. Rewriting the title, thumbnail and description of a well-performing video simultaneously removes any way to tell what happened afterwards.
Corrections are different from improvements. Broken links and inaccurate claims should be fixed immediately regardless of performance, since those are factual problems. Rewriting a title for better click-through is an experiment and should be treated as one. The wider review is in the channel audit checklist, and metadata standards in the title and description guide.
Common questions
How often should this run?
Twice a year for the top videos, annually for the rest. More often is rarely worth the time.
Does changing metadata reset performance?
Editing fields does not reset a video. Replacing the file does, since that produces a new URL.
Should old videos get new thumbnails?
For those still receiving views, often yes. It is one of the cheapest improvements available.
What about videos with no views at all?
Leave them, or unlist them if they misrepresent the channel. Repairing metadata will not revive them. See contact.
Is it worth adding chapters retrospectively?
On longer videos still being watched, yes. They help viewers navigate and often improve how long people stay.
Should descriptions be standardised across the library?
Standardise the recurring block only. Forcing old videos into a current template usually strips the specific detail that made each description useful.
Who should run the metadata audit?
Whoever writes metadata now. Reviewing older work is the fastest way to see which habits have improved and which have simply persisted.