A content marketing workflow for video has five stages: brief, capture, edit, publish, and reuse. Most teams write down the first four and improvise the fifth, which is how a team can be busy every week and still start each new brief from an empty timeline. The stage that decides whether the workflow holds is none of the five on its own. It is the handoff between them, and handoffs run on names and notes. If you fix nothing else, fix what gets written down at the moment material changes hands.
Brief: name the things you will want to find later
A video brief usually says what the video is for. The version that helps you a year later also says what will be in it: the product, the claim, the location, the people who appear and what their release covers. That paragraph costs five minutes to write and becomes the only description of the footage that exists before anyone touches it.
This is also where you decide, in advance, whether this shoot is meant to produce one asset or a set. That decision changes what the crew captures, and it is much cheaper than discovering three months later that the b-roll for six cutdowns was never shot.
Capture: on-set notes are the only free metadata you get
Whatever anyone writes on the day is free. Everything written afterwards costs someone an hour of watching. Camera reports, a slate log, the interview questions in order, a note about which take the client liked. None of this is glamorous and all of it is the raw material for every search you will run later.
The workflow rule that matters here is that footage does not enter shared storage without whatever notes exist alongside it. Material that lands in a folder with no context is not managed, it is stored.
Edit: the bottleneck is selection, not cutting
The assumption baked into most workflows is that the edit is assembly. Editors will tell you otherwise. On a shoot with a lot of coverage, the largest single block of edit time goes to reviewing everything and deciding what is worth using, before any cutting happens. A ten-minute finished piece can eat a couple of working days on that basis, and most of it is watching.
This has a direct workflow consequence. Anything that shortens selection buys more time than anything that speeds up the timeline work. A logged shoot with usable notes is the cheap version. Search across the transcripts of every take is the version people ask for when they have tried the cheap one and found it insufficient, and it is a real request: production teams setting up infrastructure for a new project routinely ask for phrase search across all their transcripts, because they have worked somewhere that had it.
Publish, then the stage nobody owns
Publishing is the best documented stage in most teams because platforms enforce it. Reuse has no enforcement, so it does not happen. The workflow ends when the video goes live, and the material goes quiet.
The fix is not another process document, since the reason reuse fails is that someone has to remember what exists. This is where indexing at ingest changes the arithmetic: if uploading each shoot's footage to a Vivu project becomes part of the publish step, every piece is indexed once and reuse stops depending on anyone's memory. Someone planning the next brief can then ask an AI assistant that has Vivu connected, Claude or ChatGPT for example, what the project already holds on a topic, and get back time ranges to review before anyone books a shoot. It does not cut or generate anything, so the edit stage stays exactly where it is. If reuse for you mostly means turning long recordings into short pieces, the mechanics of that specific conversion are worth reading before you design the stage.
The naming and tagging question
Every team has this argument and most resolve it badly, in one of two directions. Elaborate taxonomies get abandoned within a quarter because nobody fills in fourteen fields at 7pm on a delivery day. No convention at all leaves you with three files called final.
The workable middle is a short required set that a tired person can complete, plus optional depth for whoever cares. Decide it once, apply it at ingest rather than retroactively, and accept that partial compliance is the realistic target. Automated tagging can carry some of this load, though what those tools actually produce is worth checking against your own material first, because generic labels answer generic questions. If the tagging discussion is really a folder discussion, the structural version is the one to have first.
When the workflow is just overhead
A team producing one video a month does not need five stages, it needs a checklist and one person who remembers everything. Formal workflow earns its cost at the point where two productions overlap in time, or where a freelancer or agency touches material that has to survive their departure. Below that, the process is more expensive than the confusion it prevents.
How to tell if yours is working
Run one test. Pick a brief from eighteen months ago and ask someone who was not involved to find the raw footage, identify the final cut, and confirm it can still legally run. A workflow that survives that has the handoffs right. One that fails at the second step has a naming problem, and one that fails at the third has a rights problem, and both are cheaper to fix than the shoot you would otherwise book to replace what you already have.
FAQ
How many people do you need to run a video content workflow?
Two roles, which can be one person: someone accountable for what gets made, and someone accountable for what happens to the files afterwards. The second role is the one teams skip.
Below about four videos a month, one person holding both is fine and a formal process adds more overhead than it removes. Past that, the file role needs to be named explicitly in someone's responsibilities, because it is invisible work and invisible work goes to whoever is least busy, which means it stops.
Should the video workflow live in the project tool or in storage?
In the project tool for the stages, in storage for the material, with one agreed link between them. Trying to run production tracking out of a folder structure fails, and trying to keep footage inside a project tool fails harder.
The link that matters is a project code that appears in both places. It is the cheapest possible join between the thing that tracks the work and the thing that holds the output, and it is what lets someone reconstruct a project two years later.
How do we keep freelancers and agencies inside the workflow?
Put the delivery requirements in the contract rather than in an email after the fact. Specify the naming pattern, what raw material comes back, and what documentation accompanies it.
Freelancers follow whatever you asked for at the point of hiring and reasonably ignore conventions introduced at delivery. The most expensive gaps in most video archives are exactly the quarters where outside help delivered finished files and kept everything else.
Do we need to tag every raw clip?
No, and trying is how tagging projects die. Tag at the level of the shoot rather than the clip, then rely on content search for anything more granular.
Shoot-level metadata is cheap because there are maybe forty shoots a year and four thousand clips. The granular layer is the part worth automating or searching, because a person doing it by hand will either quit the task or do it inconsistently enough that you cannot trust the results.