A video feedback workflow for agencies:
one room per cut, and a record of who asked for what.

Missing one of those notes is annoying and recoverable. The costly failure is different and it happens weeks later. The brand asked for a line to be cut. The edit was recut, the grade was redone, the mix was conformed. Then the brand says it never asked for that, or that it asked for something else, and the agency has to decide whether to eat the day or argue about a conversation. Arguing from memory is how agencies lose margin and relationships in the same meeting.

The thing that prevents it is not a better meeting. It is a record: a note stored against a specific frame of a specific cut, signed with the person who wrote it, dated, and still there in six months. Everything below is built around that one idea. The pillar guide on review and approval software covers how to pick a tool at all. This page is about running the process once you have one.

Approval sits on the same footing. One film carries several versions, and approval is recorded per version, by name. The film’s overall state is one of three: reviewing, changes requested, or approved. It is a stored decision naming a cut and a person, not a thumbs up on a message.

One rule makes that state trustworthy: posting a new note on an approved cut revokes the approval. An approved film therefore always means the same thing, which is that nobody has objected since. The cost of that rule is real and worth saying. A stray note from a stakeholder who is thinking out loud will push a signed-off cut back to changes requested, and someone has to resolve it before the state returns.

Two versions can be compared side by side, as a slide wipe or stacked, which is how a creative director checks that a note was actually addressed instead of taking the editor’s word for it. The version comparison guide covers that step on its own.

No software decides whether the brand manager or the creative director is right. Anything that claims to is either routing tickets or averaging opinions, and neither is a creative decision. Reconciliation is a person, and the agency should name that person on every job before the first link goes out.

What the room changes is the input to that person’s decision. They are not reconstructing four mail threads and a phone call. They open one round, read notes in timecode order, see who wrote each one, and see which are already done. The reply they write becomes the audit trail for the choice, so when the losing stakeholder asks why their note was not actioned, the answer is in the same place as the note.

Two habits make this hold up. Never action a note that arrived outside the room; ask the person to put it in the room, or paste it in yourself and say who it came from. And never silently drop a note: mark it done with a reason, because a note that disappears is the one that gets litigated later.

It is worth checking the same row on whatever you are comparing, because the answer is often better than agencies assume. Frame.io, the tool most post-production teams already have, prices review by the member: its pricing page lists Pro at $15 per member per month plus tax and Team at $25 per member per month plus tax, with Pro capped at five members and Team at fifteen. It also states plainly that outside collaborators can be given share links for free, without signing up or logging in. So a brand-side reviewer is not the thing that costs money there either.

The row that does move is your own staff. An agency where the producer, the account lead, two editors and a designer all need to be inside the workspace is buying members, and past five of them the tier changes. Count your internal people, not your client list, before comparing anything else.

A round that stays in a browser gets retyped into a notepad, and retyping is where notes get lost. A Premiere Pro panel and a DaVinci Resolve script both exist and are downloadable. They pull the room’s notes over the network and write native timeline markers: red for an open note, green for one marked done. The editor scrubs between markers rather than between a browser tab and a timeline.

The panel tags its own markers, so a refresh updates them in place and never touches markers the editor made themselves. That matters on an agency job, because the assistant has usually already laid down their own markers for music hits and stock replacements, and losing those to a sync is worse than not syncing at all. The Premiere panel is a signed extension; the signature is the studio’s own.

An agency whose client contract requires formal approval workflow should read this section as a recommendation to buy something else. There is no enforced sequential sign-off here: nothing prevents legal from approving before brand, or brand from approving a cut the creative director has not seen. Approvals are recorded per version, by name, in whatever order they arrive.

There is no audit export either. The record lives in the room and is read in the room. If a compliance officer needs a signed trail on paper, a retention policy, or an e-signature, that is a different category of product and it is worth the money it costs. The same applies to single sign-on and directory provisioning: an agency onboarding fifty people through an identity provider should buy a system built for that, not this.

What this does, it does completely. Every note in the job carries a timecode, an author, a round and a done flag. Every approval names a cut and a person. A new note takes the approval back, so the state never lies. And the brand-side people who generate most of the risk get in with a link, which is the difference between a note written today and a note written next week when it is expensive. If access control is the part you are worried about rather than sign-off, the screener sharing guide covers revocation and expiry in detail.