
How much desk time does one container really take?
One container takes about 2.2 hours of desk work: 30–45 min chasing the supplier, 45–90 min on the document pack, 20–30 min tracking, 10–15 min on payments. At 500 containers a month that is roughly 1,100 hours against a seven-person desk's 1,232, so about 90% of the team is consumed before anyone trades.
Ask a trader what one container costs to process and you get a shrug. Ask the documentation executive on the same desk and you get a number to the minute, because that is the person doing the work.
We asked them. Then we timed it, container by container, before we wrote a line of product code.
Two point two hours.
Where do the 2.2 hours go?
| Task | Time per container |
|---|---|
| Supplier follow-ups | 30–45 min |
| Export doc-pack prep and checking | 45–90 min |
| Status tracking and reporting | 20–30 min |
| Payment tracking | 10–15 min |
The real range is 2 to 2.5 hours, and the variable is the supplier. A cooperative supplier who sends a correct pack the first time takes you to the bottom of the range. One who sends a packing list whose gross weight disagrees with the invoice takes you to the top, because the correction loop touches every downstream document.
Look at the widest band on that table. Document-pack prep swings from 45 minutes to 90, a full 45-minute spread on a single line. That spread is not skill. It is rework.
Sitting on top of all four lines, and belonging to no container at all, is 1 to 2 hours a day of price collection. Someone asks every supplier what they are quoting this morning, and lays the answers out against yesterday. It is a fixed daily tax on the desk. It never shows up in a per-container estimate. It always shows up in the payroll.
What does 2.2 hours a container do to a seven-person desk?
This is the arithmetic that changes the conversation, and it is four lines long.
| Line | Value |
|---|---|
| Containers a month | 500 |
| Hours consumed at 2.2 h each | ~1,100 |
| Desk capacity: 7 people × 176 hours | 1,232 |
| Share of capacity on grunt work | ~90% |
Ninety percent.
The desk is not busy because trade is busy. The desk is busy because the coordination has no owner and lands, by default, on the people who are best at documents. Buying and selling happens in the ten percent that is left, which is one reason a good trading desk plateaus at a volume that has nothing to do with its credit line.
Nothing on that list is hard. Nothing on it is un-automatable because the work is subtle. It is un-automated because every ERP, every CTRM and every documentation tool on the market records what already happened and then hands the follow-up back to a human.
What is that time worth in money?
Take Singapore, since it is one of the two markets we sell into and its cost structure is public.
Market rates for the four roles a trade-ops desk hires, from 2025–26 listings:
| Role | Base S$/month |
|---|---|
| Trade ops / shipping manager | 7,000–8,500 |
| Trade operations executive | 4,200–5,500 |
| Export documentation executive | 3,300–4,500 |
| Shipping / ops coordinator | 2,800–3,500 |
A representative seven-person desk is S$30,000 to S$36,000 a month in base payroll. Load it with 17% employer CPF, applied to wages up to the Ordinary Wage ceiling, and the customary annual wage supplement, and you land at S$38,000 to S$44,000 a month. Call it S$460,000 to S$530,000 a year, roughly US$340,000 to US$390,000.
That is the price of the desk. Ninety percent of it is going to the four lines in the first table.
How do I measure this on my own desk?
Do not run a time-and-motion study. Nobody finishes those. Run this instead, and it takes one week.
1/ Pick twenty shipments that are live right now. Not your best twenty. The next twenty.
2/ Make four columns and only four: supplier follow-ups, document pack, tracking and reporting, payment tracking. Resist adding a fifth. Five columns is where these exercises die.
3/ Log minutes at the end of each block of work, not once at day’s end. Memory at 6pm rounds everything to fifteen minutes and hides the rework.
4/ Log the rework separately, in the same column. When a document comes back wrong, that is a second entry against the same shipment, not an extension of the first.
5/ At the end of the week, divide by twenty. Then multiply by your monthly container count. Then divide by 176 to get the answer in people.
That last number is the one to walk into a management meeting with. Hours are abstract. “This is 4.25 people” is not.
Two warnings for anyone running this. Your team will guess low in week one, by a wide margin, because the follow-ups feel like nothing. Each individual follow-up is nothing. And your document-pack numbers will be higher than you expect on exactly the shipments where the supplier is a long-standing relationship, because the packs from familiar suppliers get checked less and corrected more.
Why does everyone guess low?
Three reasons, and they compound.
Each follow-up is small. Four minutes to write a message asking for a corrected weight. Nobody logs four minutes. Nobody remembers four minutes at 6pm. The follow-up line disappears from every estimate a desk makes about itself, and it is 30 to 45 minutes a container.
The waiting is invisible. A shipment that sits three days for a reply consumed no hours, so it costs nothing on a timesheet and everything at the terminal. Time-to-answer and time-spent are different measurements and only one of them shows up in payroll.
The rework is filed under the original task. A pack corrected twice is remembered as “doing the pack”, not as three separate pieces of work. That is why the document line swings 45 minutes and why the swing is invisible to the person doing it.
There is a fourth, softer one. People undercount work they are good at. The documentation executive who has done this for four years does not experience the check as effort, so they report it as fifteen minutes when the log says forty.
What share of it can a machine take?
We publish three automation rates separately, because a single blended claim hides where the product is weak.
| Work | Share automated |
|---|---|
| Supplier chasing | 85% |
| Document-pack drafting | 65% |
| Tracking and reporting | 85% |
| Blended across the desk | ~70% |
Sixty-five percent on document drafting is the weakest line and it is the honest one. Drafting a pack against a contract and a set of masters gets you most of the way. The last stretch is judgement about a specific counterparty in a specific country, and we are not going to claim a machine does that today.
Applied to the 500-container desk, 70% of 1,100 hours is 770 hours a month. At 176 hours a person, that is 4.25 people of capacity.
Two rates live in this business and mixing them inflates the answer. The 85% figure is the cut in operations cost on the chase, which is what takes a container from 2.2 hours to about 20 minutes of human time. The 70% figure is the blended rate across all desk work, which is what you use for a whole-desk calculation. Computing freed hours from the 20-minute figure would overstate the result by about a fifth. It is a tempting mistake and a prospect with a calculator will find it.
Capacity, not headcount
Here is the framing that matters, and it is not a marketing preference.
At 70% blended, the same seven people handle 1,000 to 1,200 containers instead of 500. Or two or three of them move to buying and selling, which is the work that makes money.
Nobody forwards an internal memo that reads as “fire four people”. A trader will forward a memo that reads as “we double volume without hiring”, because that memo is about growth, and growth is what they are paid on. If your business case only works by removing people, it will not survive contact with the person who has to present it upstairs.
The gap in this post
These timings are an operational baseline, not an industry benchmark, and calling them one would be a claim you could check and break. They describe a desk running physical commodity imports with documents on the critical path. The method below is the transferable part — run it on your own desk for a week and your number beats ours by definition.
The Singapore cost basis is market listings and published CPF rates, not a payroll file. UAE and India loading we have not published, because getting end-of-service gratuity or employer PF wrong is the kind of error a prospect catches in ten seconds.
Where Docket fits
Docket takes the contract, derives the deadlines, chases the supplier up the ladder from email to WhatsApp to SMS to a voice call, drafts the pack against your masters, and tracks the box. On the 500-container desk that is 770 hours a month of desk time removed, worth about US$19,250 at a US$25 loaded hourly cost. Docket at that volume costs US$2,100 a month. Net, US$17,150 a month, or US$205,800 a year.
Money freed minus what we cost. That is the only formula on this site.
Before you look at any product, including ours, go and measure your own twenty shipments. If your number comes back well under 2.2 hours, tell me, and tell me what your desk does differently. I will publish it.
Your desk is leaving about US$200k a year on the table to manually type what a machine learns in a month. If you own that desk, the software for importers page puts the same arithmetic in your terms.
Questions this raises
Does the 2.2 hours include price collection from suppliers?
No. Daily price collection runs 1 to 2 hours a day for the whole desk, not per container, so it sits on top of the 2.2 hours. It never appears in a per-container estimate and always appears in the payroll.
Is 2.2 hours a lot or a little?
It is a little per box and a lot per month. Five hundred containers is roughly 1,100 hours. Seven people at 176 hours each provide 1,232. The number only becomes alarming when you multiply it.
How do I measure this on my own desk?
Take twenty shipments, four columns, one week. Log minutes against supplier follow-ups, document-pack work, tracking and payment tracking. Log at the end of each block, not at day's end. Twenty shipments is enough to see the shape.
Does automating this mean cutting the team?
No. The same seven people handle roughly 1,000 to 1,200 containers instead of 500, or two or three of them move to buying and selling. Traders buy growth. Nobody forwards an internal memo that reads as layoffs.
What share of this work can software take?
On our measurements, 85% of supplier chasing, 65% of document-pack drafting and 85% of tracking and reporting. Blended across all desk work that is about 70%. We publish the blended figure because it is the honest one for a whole-desk calculation.
Sources
- Docket operational baseline, import–export desk · checked 2026-08-08
- Docket engineering model — projected automation rate by task · checked 2026-08-08
- Singapore market salary listings for trade ops roles, 2025–26 (Glassdoor, JobStreet, Indeed); CPF Board employer contribution rate and Ordinary Wage ceiling · checked 2026-08-08