Work backwards from the last person who touched it rather than forwards through the folder tree. The order that finds things fastest is people, then publishing, then drives: whoever commissioned it, whoever edited it, wherever it was published, and only then the storage everyone searches first out of habit. Decide at the start which thing you need, because a finished cut and the footage inside it live in different places, and only one of them is usually still recoverable.
Start with the people
The person who signed off on the video remembers more about it than any file name preserves: the quarter it went out, the agency, the campaign it belonged to, the name of the internal project. That is the set of words that makes every later search work. If they have left the company, the next best sources are whoever sat next to them and whoever paid the invoice, since finance records survive reorganisations better than shared drives do.
The hardest version of this hunt is the one where the person looking did not make the video and the people who did are gone. In that case, treat it as an archive problem rather than a search problem: establish roughly when it was made before trying to find it, because a date range turns an unbounded hunt into a bounded one.
The published copy is usually findable, the master often is not
Anything that went out publicly leaves a trail: the platform it was posted to, the page it was embedded in, a press release, an old newsletter. A web copy is frequently enough for what you actually need, which is often confirming the video exists or pulling a reference for a brief. What rarely survives is the master, because masters sit on production storage that gets recycled, and because nobody was assigned to keep them. Getting clear about which you need saves a week of looking in the wrong direction.
How to recover a lost marketing video
If no copy turns up in your own storage, go through the places that hold copies for their own reasons. The agency or freelancer who made it may still have the project archive. The review and approval step of the original project left exports behind. Backup snapshots of a laptop that has since been replaced sometimes survive in IT's retention system, which is worth one email. Old email threads and expired transfer links tell you at least who had the file and when, which narrows who to ask.
When only a compressed public copy comes back, be honest with the requester about what it can be used for. It is fine for internal reference, for a pitch deck, and often for social. It is not a source for a new grade, a broadcast delivery, or anything that gets projected large. Upscaling produces a new image rather than the detail that was thrown away.
Where a search layer helps, and where it does not
If the candidates are already in one indexed place, a broad pass narrows which files might be the one, which is exactly the question at this stage. In our own runs, that coarse mode returned whole videos as candidates with no position marked inside them, so you still open each one to confirm, which beats opening all of them. A retrieval layer like Vivu only covers material that was uploaded to a project, so it is useful for the second half of this hunt and no help at all for a file nobody has uploaded. That boundary is the reason the people step comes first: search works on what you have already gathered.
When to stop looking
At some point the cost of the hunt passes the cost of the alternative. If the video is old enough that the product, the branding, or the people in it have changed, recovering it may hand you something you cannot use anyway. Reshooting a short piece is often cheaper than three people spending a week in storage. The exception is anything with legal or historical value, where you keep looking because no substitute exists.
Deciding which hunt you are on
Two questions settle this. Do you need this exact video, or something that does the same job? And do you need the master, or a copy anyone can watch? A copy of something recognisable is usually available within a day, once you accept that people and publishing trails are faster routes than folders. Masters are a different matter, and a failed hunt is the best moment to introduce a delivery rule so the next one is shorter. The general-case version of this search is laid out in the routine for locating older marketing material, and the structural fix is in setting up a library that survives handoffs.
FAQ
How can I find an older advertising video?
Start from the campaign rather than the file. Anything that ran as advertising has paperwork around it: a media plan, an invoice, an approval thread, a press mention with a date. Those give you the quarter and the agency, which is what makes searching a drive or asking a former colleague productive. Public archives of advertising and the brand's own social history are also worth checking, since a copy that was published somewhere is easier to recover than one that only ever existed internally.
What if the agency that made it no longer exists?
Look for the people rather than the company. Editors and producers keep portfolios and often keep project archives, and a former staff member can usually say what happened to the storage when the company closed. If the agency was acquired, the acquirer may hold the archives. Failing all of that, the published copy is likely the only version left, so plan around it.
How far back is it realistic to go?
Finished videos that were published publicly can often be recovered from many years back, because copies spread. Source footage tends to have a much shorter life, frequently only as long as the production storage it sat on. As a rough expectation, the older the material, the more likely you are recovering a viewing copy rather than something you can edit again.
Is it worth searching the edit projects themselves?
Yes, and it is the step people skip. Edit projects reference the original media and record what was used where, so opening one tells you which footage existed even when the footage itself is gone. It also gives you the original clip names, which are the search terms most likely to match whatever survives on a drive.