From 0 of 89 to 117 of 117.
Two episodes for the same client. Every piece now has its own owner, its own date, its own checklist and its own place to put the proof.
Before, 27 August
Three episodes sat in Under Review By Client with not one checklist item ticked. The work had gone to the client and nothing on the card recorded that any of it was done.
After, 3 October
A full episode now reaches the client with every item ticked across six separate pieces, each one carrying a review video and a link to the finished file.
The work did not change and the people did not change. The unit of work did.
Every figure here is readable in the screenshots.
The before column is their live workspace on 27 August 2026. The after column is the same business running the new system on 3 and 7 October 2026.
| Measure | Before | After |
|---|---|---|
| Longest checklist one person has to open | 183 items | 52 items, and most people see 15 to 20 |
| Checklist items ticked on work in progress | 60 of 332 | Complete before the piece can move on |
| Checklist items ticked on work already sent to the client | 0 of 89 | 117 of 117 |
| Pieces of work tracked per episode | 1 card | 5 to 15 cards, one per deliverable |
| Lists carrying live work | 1 of 5 | 5 across 4 spaces |
| Views built around how people actually work | 1 shared board | 7 views and 2 dashboards |
| Kinds of work measured separately | None | 14 |
| Where proof that a job was done is kept | A comment | A required field on every piece |
Read the second and third rows together
Their checklists were not ignored because the team was careless. They were ignored because one card held the whole episode, so a reel editor had to scroll past 150 lines written for other people to find his own. Make the list match the job and people tick it.
An episode is not one job. It is nine jobs, done by four people, over three days.
A top rated agency that could not see clients drifting away.
A top rated agency doing post production and editing for podcasts, working with big creators in the US and the UK. Kaleem runs it with five full time editors, a production manager and three freelancers.
Every episode they deliver is not one job. It is a package: the main edit, a teaser, two or more reels, thumbnails, short segments, an audio version, the written metadata, the upload to the master drive, and for some clients the posting itself. Nine or ten separate pieces, made by three or four different people, inside a three day window.
They were running all of it through one ClickUp list, with one card per episode and the whole package written out as checklists inside that card.
Why he called me
Not because the board was untidy. Because clients were drifting away somewhere around month four to month seven, and he could not see it coming. His score on Upwork is where his leads come from, so a client who leaves quietly costs him far more than that client's retainer.
He also told me something more useful than any brief. He said the ideas run around in his head but he never applies them. He did not need advice. He needed the system to carry the execution so it stopped depending on him remembering.
What I was asked to do
- Rebuild how production is tracked, from footage arriving to the episode being posted
- Make the client side visible too: onboarding, health, leads and renewals
- Do it without stopping a single day of live client work
What made it harder than it looks
Twelve clients, and no two of them bought the same set of deliverables. Every client's checklists were written in a different order, so nothing could be copied from one to another. The shared part turned out to be around a third.
Six of the nine people only had guest access to a single list, so whatever replaced it had to be simple enough to learn in a day.
What their workspace actually looked like.
Their live production board on 27 August 2026. One list, one card per episode, and the templates parked in a live column beside real work.
None of this was carelessness. Every one of these habits made sense when the agency was two people and one client, and nobody had a reason to stop and rebuild while the work kept arriving.
The problem was not the checklists.
It was the unit of work. They tracked one card per episode. But an episode is not one job. It is nine jobs, done by four people, over three days.
When the unit of work is wrong, everything downstream breaks, and no amount of tidying the board will fix it:
- A card cannot be assigned to four people, so the real assignment happens in conversation and leaves no record.
- A card cannot hold four different due dates, so nothing is late until everything is late.
- You cannot measure how long a thumbnail takes if the thumbnail does not exist as a thing.
- You cannot give a reel editor a short list of what to do without showing him 150 lines meant for other people.
- A dashboard built on this data can count episodes. It can never count work.
Before
- One episode card
- 183 checklist lines
- One date, no owner
- Four people working inside it
After
- Base edit, 52 checks
- Teaser, 19 checks
- Reel, 13 checks (each)
- Audio, 9 checks
- Admin, 16 checks
- One owner and one date on each
So the fix was not a better board
It was splitting the episode into the pieces people actually do, and giving each piece one owner, one date, a short checklist of its own, and somewhere to put the proof. Everything else in this project follows from that one decision.
The second thing that was tangled
A checklist confirms something was done. An instruction explains how to do it. Their lists were trying to do both at once, which is exactly why they grew to 183 lines and why most of them went unticked. Nobody reads a reference manual while they work.
I separated them. The short confirmation list stays on the card. The how to do it moved into written guides the card links to. Lists got shorter, and the knowledge that used to live in one person's head got written down for the first time.
What I built.
Built fresh alongside their live workspace, so not one day of client work was disrupted. They moved across once, at the end, after it had been tested on a real client.
How an episode starts now
Footage arrives. The manager makes one card, applies that client's template and sets the delivery date. Every piece of work for that episode then appears underneath it, already carrying its own owner, its own date, its own short checklist and its own fields for the proof. He does not date them one by one. He does not tell anyone what to do.
Before, five flat lists
- Episodes: 49 cards, in use
- All Clients: 20 cards, abandoned
- Tasks: 90 cards, abandoned
- Monthly Tracker: 2 cards, abandoned
- All Leads: 1 card, abandoned
After, four spaces
- Delivery: Production, Episodes, Other Work, manager and owner dashboards
- Growth: Leads
- Operations: Team tasks, Task templates
- Process Library: house rules and procedures
What the 23 automations actually do
- Hold a deliverable at the footage gate until the footage is approved, with a comment saying why rather than just refusing to move.
- Chase the client when an episode has sat with them for 48 hours without feedback.
- Create the 30, 60 and 90 day view logging tasks the moment an episode is posted.
The same episode, both ways.
Nothing about the work changed. The editors do exactly what they did before. What changed is that the work now exists in pieces the system can see.
Before: one card
- Everyone works inside it
- Up to 183 checklist lines
- Assignment happens in conversation
- One due date for nine jobs
- Proof buried in the comments
- Nothing countable at the end
After: one card per piece
- One owner each, one date each
- 8 to 52 lines, written for that job
- Assignment is the card itself
- Every piece can be late on its own
- Proof is a field, not a comment
- Hours and volume readable per piece
The card on the right has already been through review. Every piece under it is ticked right through: 16 of 16, 47 of 47, 19 of 19, 13 of 13. Under the old board all of that would have been one card with one checklist, and the only number anybody saw covered the whole episode at once.
Nothing moves on somebody's word.
Before, a piece was finished when the editor said it was finished. The review video he recorded went into a comment and was gone within the week.
Seven views, one per job.
The old board showed everybody everything. Six editors were guests on one shared list and each of them had to find their own work inside other people's episodes. The new system never asks anyone to filter anything.
Numbers that did not exist in August.
Two dashboards, built for two different questions. The manager's answers what is stuck today. The owner's answers how the month is going. Keeping them apart is what stops either becoming a wall of numbers nobody reads.
| The same three cards, four days apart | 3 Oct | 7 Oct |
|---|---|---|
| Awaiting footage | 4 | 3 |
| Overdue | 38 | 37 |
| Delivered and not paid | 37 | 52 |
They move because the team is working in it. A dashboard that reads the same every week is a dashboard nobody is feeding.
That second chart is the one that pays for the project. Hours, volume and cost can now be read per deliverable, per person and per client. Which clients are cheap and which are expensive stops being a feeling and becomes a number.
What else sits on them
- Who is loaded between now and next week, so work is spread before it is handed out.
- Anything in review without its proof video attached.
- Anything missing the information the next person down the line will need.
- Hours logged per client for the month, which is what the owner's dashboard is built on.
The build is the easy half.
Most operations projects fail in the fortnight after handover, not during the build. People go back to what they know the first time the new thing is slower than the old thing.
- A brand guide for each client, holding the setup that used to live in one person's memory: assets, fonts, music, where files go, what that client wants done differently.
- House rules and standard procedures, written for editors rather than for managers.
- A separate guide for how the leads board works, so the client side does not drift back into a spreadsheet.
The choice that nearly went wrong
The obvious pilot client was the newest one. I pushed back, because their work is a different service to everything else the agency does, so a clean week would have proved almost nothing. I piloted on a typical client instead.
What the screenshots show, and what the business can now answer.
Measured today
- Work reaching the client arrives complete. One episode went out with 117 of 117 items ticked across six pieces, each with its review video and its file link attached. In August, three episodes went out with 0 of 89.
- The longest list anyone opens dropped from 183 items to 52, and most people see 15 to 20.
- Around 200 individual pieces of work are tracked at any moment, each with an owner, a date and a state. The same work used to be 49 cards with nothing visible inside them.
- Four of five lists were dead. Five lists across four spaces are now in daily use, plus two dashboards nobody had to be told to open.
- The intake that lived in one person's memory is written down for every client, so an episode can be started by somebody else.
Answerable for the first time
- How many hours an episode costs, and how that differs between clients.
- How long each kind of piece takes, so quoting stops being guesswork.
- Who is over capacity this week, before the work is handed out rather than after.
- What has been delivered and not yet paid for.
- How many rounds of revisions a client asks for, which was only ever in the owner's head.
- How a posted episode performed at 30, 60 and 90 days, against what was done differently.
What I am measuring next
At 30, 60 and 90 days from launch I am pulling the same five numbers each time: hours per episode, episodes delivered per month, revision rounds per episode, days from footage arriving to delivery, and movement in client status. Same method every time, so the comparison means something.
I am deliberately not claiming a percentage on speed or cost yet. Those numbers were never recorded before the rebuild, so there is no honest baseline to compare against. The system is now producing them for the first time, and I would rather show a real number in January than an invented one today.
What happens to all of this in a year
A system that only grows gets slow, so the archiving is designed in rather than left to somebody's judgement. Delivered episodes archive after a month. Posted ones wait four months, so their 30, 60 and 90 day view logging finishes first. The archive runs off the monthly note, which means it has an owner and a date instead of being a job nobody remembers.
Archived episodes keep their status rather than being hidden by ClickUp, so they stay in the dashboards and the history stays readable.
The same problem looks like this.
- Your team works inside one card, one row or one folder per job, and several people are in there at once.
- Your checklists have grown so long that people have stopped finishing them, and adding more has not helped.
- You find out something is late when the client tells you.
- One person holds the knowledge that keeps delivery moving, and everyone quietly works around their absence.
- You cannot say what a job costs you in hours, so you cannot say whether a client is worth keeping.
If three of those are true, the problem is probably not the tool and not the team. It is the unit of work. Changing software does not fix that on its own, which is why the same business can move from Trello to Asana to ClickUp and feel no better for it.
Where this usually starts
With an audit, which is the first paid phase rather than something I give away to win the work. I spend it in your actual workspace, with your actual team, finding the shape of the work instead of the shape of the tool. You get the findings whether or not you go on to the build.
Everything after that follows from what the audit finds, which is why I will not quote a system before I have seen one.