Start from what you remember about the video rather than from where you think the file is. If you remember the campaign or roughly when it ran, browsing by date in the drive or the ad account will usually get you there in a few minutes. If what you remember is a moment inside the video, the shot of the warehouse floor or the line about the warranty, then filenames and folders record none of that and browsing will never surface it. Most long searches for an old marketing video are someone using the first method on a problem that is actually the second.
A marketing archive is never one archive
Finished cuts sit in a shared drive. The project files that produced them sit with whoever did the edit, which is often an agency you no longer work with. Platform variants sit inside ad accounts under names a media buyer chose for their own reasons. Anything older than the current agency contract may exist only in a handoff folder nobody has opened since the day it arrived.
Decide which copy you need before you start looking. A social cut you can repost is an easy retrieval problem. A master you can re-edit is a hard one, and the master is usually the thing that turns out to be missing. If you are chasing one particular video you know exists, do that cheap version first before you conclude the whole archive needs fixing.
Searching the storage you already have
Drive search matches filenames and, for documents, file contents. For video it matches almost nothing useful. Filtering by file type and date range narrows a drive much faster than typing keywords, and sorting by size finds masters, since a ProRes master is enormous next to the delivery files around it. This route costs nothing and works whenever past-you was disciplined about naming. It fails silently the moment the file is called final_v7_REAL.mp4.
Asking whoever was there
For a team under a few years old this is still the fastest method, and people underrate it because it does not feel like a system. The cost arrives later. Memory of an archive lives in two or three people, and when one of them leaves, the part of the archive only they could navigate is effectively gone. If you notice you are asking the same colleague every single time, that is the signal to build something.
Cataloguing it by hand
A spreadsheet with one row per video, a link, a date and a plain description of what is in it will outperform most software for a library of a couple of hundred pieces. It is boring, durable, and costs about a minute per video plus the discipline to keep adding rows. Its failure mode is drift: the sheet records what you thought mattered when you wrote the row, so it answers next year's questions only by luck.
Buying a system for it
Asset management tools give you tagging, versioning, permissions and one search box over all of it. They pay off at the point where several people need to find things without asking each other, and at that scale they are worth comparing properly. The honest cost is that most of them assume the material moves into them, and that tags get applied by a person. A tag only ever answers the question the tagger thought to anticipate.
Searching what was said
If your videos have speech in them, transcribing the archive gives you a text layer you can search. It is cheap now and it covers a real slice of what people look for, especially interviews and customer testimonials. It covers nothing that was never said out loud, which for brand video is most of the frame. Teams who have done this often find the gap the hard way: hundreds of clean, human-corrected transcripts sitting on a website, and answering a specific question still means site-scoped Google tricks. Working from the transcript is worth learning before you buy anything else.
Indexing what is inside the footage
The newer route puts a retrieval layer over material where it already lives, indexes what is visible and audible in each file, and answers a described moment with a timecode rather than a list of filenames. Vivu answers a described moment the same way, once the archive has been uploaded and indexed in a Vivu project. Nothing needs restructuring before the upload, and campaign files added to that project later are searched together with the old ones. For old marketing video that matters because the description in your head is the one handle on the file that never got lost.
When you do not need any of this
If you make a dozen videos a year, one person owns them, and they live in one drive, every option above is overhead. Spend the hour writing descriptions into a spreadsheet instead. The threshold is not really volume, it is turnover: an archive stays findable as long as the people who made it are still around and still asked.
Which side you are on comes down to one test. Think of a specific video you needed in the last month and count the steps it took to get it. If the answer involved a person's memory, ask what happens to that video the year after they leave.
FAQ
Can I get back a marketing video that was deleted from our shared drive?
Often yes, but the window is short. Google Drive, Dropbox and OneDrive all keep deleted files in a trash state for around 30 days, and business tiers usually let an admin restore beyond that from retention or vault settings. Past those windows, the surviving copies are the ones that were never in the drive: the version uploaded to an ad account, the cut sent to a partner, the export sitting on the editor's local machine, or a compressed copy in an old email thread. Those are usually delivery files rather than masters, so you can repost them but not re-edit them.
Our agency made the videos. Can we ask for the project files?
You can ask, and whether you are entitled to them depends on what the contract says about work product. Many agency agreements transfer the finished deliverable and stay silent on project files, raw footage and licensed elements, which are treated as the agency's working material. The practical move is to request the full package at the end of each engagement rather than years later, and to write it into the next contract as a named deliverable: masters, raw footage, project files and a list of any licensed music or stock with its terms.
Is it worth transcribing an archive we only search a few times a year?
Probably not as a bulk project. Transcription is cheap per hour now, but the cost that matters is the review pass, because raw machine transcripts get product names, people's names and industry terms wrong in exactly the places you would search for them. If you search the archive a handful of times a year, transcribe on demand: when someone needs a specific interview, transcribe that one. Bulk transcription earns its keep when searching the archive is part of somebody's weekly work.
How far back is it worth making an old archive findable?
Look at how old the oldest video you actually reused was. For most brand teams that number is somewhere under three years, because anything older carries a logo, a product version or a spokesperson that has since changed. Treat material older than your own reuse horizon as storage rather than library: keep it, do not spend effort indexing it, and pull from it only when someone asks for something specific.