How to run a video content audit
An audit is not an analytics session. It is five decisions applied to everything you have published — leave, update, re-cut, consolidate, remove — and a rule for which one wins.
8 minute read
An audit is a decision pass, not a reading pass
The reason most content audits die halfway is that they turn into analytics sessions. You open a dashboard, get absorbed in a retention curve, learn something interesting about one video, and three hours later nothing has been decided. Separate the two jobs. Reading the numbers is its own skill and there are guides for each platform. An audit takes the numbers as given and produces one thing: a decision attached to every item you have published. Twice a year is enough for most catalogues, and once a year is fine if you publish weekly. Put it in the calendar as a fixed block with an end time, because an audit that runs until it feels finished never does.
Build the inventory first, and keep it boring
Start with a flat list, one row per published item, and resist enriching it. Six columns carry the whole exercise: URL, publish date, format, topic, one distribution number, one attention number, and an empty decision column. On YouTube the content tab plus an analytics export gets you there; for pages, Search Console does the same job. Pull a fixed window rather than all time, because comparing a video from 2022 against one from last March mostly measures age. The temptation is to add columns for everything the export offers. Do not. Every extra field is another reason to stop and look something up, and the value of this list is that you can sort it and move down it without thinking.
Five outcomes, and every row gets exactly one
Leave alone, update, re-cut, consolidate, remove. Forcing one choice per row is the mechanism that makes an audit terminate. The content marketing agency Animalz publishes a clean version of the criteria for written content that transfers well: refresh when the piece already has links and rankings but the information inside it has gone stale; consolidate when several thin pieces compete with each other for the same query and would be stronger merged into one with the old URLs redirected; prune when there are no links, no rankings, no meaningful traffic and something else you have made covers the topic better. For video the same three shapes appear as a video that still gets impressions but is out of date, several near-duplicate videos on one topic, and uploads nobody has watched in a year.
The returns are concentrated, so look there first
HubSpot published the most useful account of this on its own blog. Its team found that 76% of monthly blog views and 92% of monthly blog leads came from posts published in an earlier month, and that 46% of monthly leads came from just thirty individual posts. After systematically updating old posts, it reported lifting their monthly organic search views by an average of 106% and more than doubling the leads they generated. The selection rule is the transferable part: it split candidates into posts with traffic but weak conversion, and posts that converted well but had little search traffic, and applied a different fix to each. The video equivalent is close enough to be useful — videos with views but no subscriber conversion, and videos that convert well but barely get shown.
Three queries beat a dashboard tour
You do not need to look at everything. Three sorts surface almost every worthwhile candidate. First, items still receiving impressions with a weak click rate: the machine is still offering these to people and the packaging is losing them, which is the cheapest fix available. Second, items with strong retention or completion but small distribution: the material works and the reach did not, which is a re-cut candidate rather than a rewrite. Third, items that are now factually wrong — a price, a feature, a platform limit, a claim you would not make today. That third list is the only one where doing nothing has a real cost, so run it first. For how to read each of those numbers properly, use the platform guides rather than guessing at thresholds.
Update means changing the content, not the date
Google is unusually direct about this in its own guidance on creating helpful content. Two of its self-assessment questions ask whether you are changing the date on pages to make them look fresh when the content has not substantially changed, and whether you are adding or removing large amounts of content mainly to make the site seem fresh — and it answers the second one in brackets to say that no, it will not help. Set the bar accordingly: if you cannot name what a reader now gets that they did not get before, it was not an update. Repackaging is a separate and legitimate move. Changing a thumbnail and title on a published video alters how it is offered, not what it contains, and it is worth doing on anything still collecting impressions. Just record it as repackaging so you do not mistake it for a refresh.
Removal is the option to use least
Deleting is more expensive than it looks and buys less than people expect. YouTube's help documentation states plainly that deleting a video is permanent and cannot be undone, that the video's URL and title stop being visible or searchable in YouTube Analytics afterwards, and that data such as watch time remains in aggregate reports but is no longer attributed to that video. You destroy the record and gain nothing you could not have had by unlisting it. For web pages the arithmetic is different, mostly because the base rate is so grim: Ahrefs' analysis of roughly 14 billion pages, published in December 2023, found 96.55% get no traffic from Google at all. Even there, delete only what is actively wrong or actively competing with a better page of your own. Hide what is merely old.
Re-cutting is the highest-yield outcome and the one people skip
Of the five decisions, re-cut is the one most audits under-use, and it is usually the best trade. A long video that still gets watched is a piece of already-validated material: someone sat through it, so you know the argument lands and the delivery works. Re-recording it is expensive and re-cutting it is not. The practical version is to take the two or three moments the retention curve says people stayed for and turn each into a self-contained vertical clip with captions, then file them in your clip library with the source video linked so the audit trail survives. This is the pass FrameOS is built for: it reads the transcript of an existing upload, surfaces the strongest self-contained moments and reframes them vertically, so a second life for an old video is a review job rather than an edit.
FAQ
What is a video content audit?
A structured review of everything you have already published, ending in one decision per item: leave it, update it, re-cut it, consolidate it with something similar, or remove it. It is deliberately not an analytics session. The numbers are an input you pull once into a flat list, and the output is a set of decisions you can work through.
How often should I audit my video content?
Twice a year is plenty for most catalogues, and once a year is fine if you publish weekly or less. Auditing monthly means re-examining videos that have not had time to settle and produces churn rather than decisions. Give it a fixed calendar block with an end time, because an audit run until it feels complete never completes.
Should I delete old YouTube videos?
Rarely. YouTube states that deleting a video is permanent and cannot be undone, that its URL and title stop appearing in YouTube Analytics, and that watch time stays in aggregate reports without being attributed to the video. Unlisting or making it private removes it from search and your channel page without destroying the upload. Delete only what is actively misleading.
How do I decide between updating a video and re-cutting it?
Update when the content is wrong or out of date but the video still gets shown. Re-cut when the content is fine and the distribution is not — strong retention with small reach means the material works and the format or the platform did not. Re-cutting an existing recording into clips is far cheaper than re-recording it, which is why it is usually the better trade.
Sources
- The Blogging Tactic No One Is Talking About: Optimizing the Past · HubSpot
- Content Refresh Strategy: How to Update Old Content · Animalz
- Creating Helpful, Reliable, People-First Content · Google Search Central
- Replace or delete your video · YouTube Help
- 96.55% of Content Gets No Traffic From Google · Ahrefs
Related pages
Keep reading
More Creator Growth →The best time to post on Instagram, according to four studies
Buffer says Thursday 9am. Later says 5am. Sprout says early afternoon. They analysed billions of posts and still disagree — here is why, and what to do about it.
8 minute readThe best time to post on TikTok, according to five studies
Buffer analysed 7.1 million posts and rates Saturday the best day. Sprout Social analysed 2 billion engagements and says avoid the weekend entirely. Both are right, for different accounts.
8 minute readThe best time to post on LinkedIn, according to five studies
Buffer's 4.8 million posts point to Wednesday at 4pm and rank Tuesday among the worst days. Sprout Social's 2 billion engagements call Tuesday the best day of the week. Here is why.
8 minute readTurn one long video into a week of posts
300 credits for 3 days · no card.
Start free