How to scale video production in an agency: the edit bench is the ceiling

Draw your video process as a line of stations. One of them has a queue in front of it. That station sets your output, and nothing upstream of it matters, because work arriving faster only makes the queue longer.

In almost every agency that station is the edit bench: the one or two people who can take footage and a script and hand back a finished file. Strategy can plan forty videos a month. Account can sell forty. Forty do not get made.

This is worth saying plainly, because agencies usually attack the wrong station. They buy a better camera. They hire a strategist. They add another project tool. Output does not move, because none of those were the constraint.

The test is boring and reliable. Look at where work sits waiting. If ten briefs are ageing in a folder while nobody is idle at the bench, you have found it.

One honest complication before you spend anything. The bench is sometimes the client, not you. If finished cuts sit in review for six days, hiring an editor buys you nothing at all, and the approval loop is the thing to fix first.

Time-log a single sixty second social video from brief to delivered file. The totals surprise people who have never measured them.

Roughly half the clock is assembly and the admin around it. Cutting, syncing, choosing b-roll, timing captions, exporting, naming, uploading, filing. None of it is the reason the client hired you.

The other half splits between deciding what the video says and handling what comes back. Deciding is the expensive judgement you actually sell. Handling revisions is the part nobody scoped and nobody bills.

Log four videos across two accounts before you change anything. The shape is stable enough that four is usually plenty, and an agency that guesses at this ratio tends to automate the wrong half of it.

Two editors do not give you two editors of output. They give you what is left after meetings, briefing gaps, context switching, and the days a client goes quiet and the queue empties.

SPI Research's 2025 Professional Services Maturity Benchmark put average billable utilisation across professional services firms at 66.4 percent, below the 75 percent most firms plan around.

Apply that rate to your own bench and the arithmetic gets uncomfortable. Two full-time editors land closer to one and a third, so a plan built on two is short by a third before the quarter starts.

The cheapest capacity in any agency is the hour you have already paid for and did not use. Before hiring, close the briefing gaps and batch the work, because both raise utilisation without touching payroll.

The honest limit on this lever: utilisation above the high seventies stops being efficiency and starts being attrition, and replacing an editor costs more than the hours you squeezed out.

A task is safe to automate when it is mechanical, repeatable, and wrong in ways you can see immediately. A caption landing a beat late is visible in one viewing. A dropped frame is visible. A mis-transcribed word is visible.

A task is unsafe when it is a judgement whose error only surfaces later, inside a client relationship or a regulator's letter. That is the whole line, and it has nothing to do with how capable the software is.

There are only three ways to raise output. Add hours. Raise the utilisation of the hours you already have. Lower the work each finished video takes.

Adding hours is the expensive one and the one agencies reach for first. It also lowers margin until the new person is fully utilised, which usually takes a quarter, sometimes two.

Raising utilisation is the cheapest and the least popular, because it means fixing briefing, batching filming days, and defending calendars against people who outrank you.

Lowering the work per video is where tooling belongs, and only on the mechanical half you logged earlier. Volume matters here for a reason worth naming: Motion's analysis of 550,000+ Meta ads puts the share of creatives that carry spend at roughly 5 to 8 percent, so an account running four a month is testing on a coin flip.

One tool sits directly on that assembly step, and it is the only product this page names. Cutroom takes one spoken take of up to 3 minutes and returns a finished 9:16 MP4 in about a minute: every word transcribed, the hook written, b-roll placed on the exact words that need showing, captions timed in one of six styles, silence and filler trimmed. You direct on the transcript rather than a timeline. A batch is 100 credits and export adds 20 credits per exported minute, so a finished 30 second video runs about 110 credits. Lite is $19.99 a month for 1,250 credits, about 11 of those, nothing on the picture, cancel in one click. The limits matter to an agency and they are flat: 9:16 only, no landscape master, no subtitle sidecar file, no shared workspace and no approval flow, so client sign-off stays wherever it lives today.