If you saw someone in the warehouse carry a carton from one side of the room to the other, then carry it back, you would ask what they were doing.
In the office, the same thing happens all day and nobody asks. An order is typed into the web store by the customer, read off a screen and typed into the ERP by a staff member, exported to a spreadsheet for the 3PL, and keyed into the accounting system at month end. The same forty lines, handled four times.
It happens because the systems do not talk to each other, so a person is paid to be the connection.
How it creeps in
It starts small. Ten orders a day, thirty seconds each to copy the address into the label printer. Nobody would build anything for that.
Then it is a hundred orders a day, and the thirty seconds is now most of somebody's morning. Then the stock file needs updating in two places. Then finance is matching deposits to orders by hand because the payment gateway and the ledger have never been introduced. Each step was reasonable on its own. Together they are two or three salaries spent on typing.
What it costs
The wages are the visible part. In a business doing a hundred orders a day, an hour of re-keying here and there across four people is easily a full-time salary a year, spent on work a machine does for nothing.
The mistakes cost more. Every time a person copies data, some of it arrives wrong. A transposed postcode sends a parcel to the wrong state. A stock level that was updated in one system and not the other sells an item you do not have. A payment matched to the wrong invoice takes an hour to untangle at month end. Each one is small, and there are a lot of them.
And the business cannot grow without the typing growing with it. The test is simple: if sales doubled next quarter, would the admin double? If the answer is yes, the business is not set up to scale. It is set up to work twice as hard.
What it should look like
An order placed online appears in the ERP without anyone touching it. Stock sold in a store comes off the website within minutes. The 3PL receives the order and sends back the tracking number, which reaches the customer without a person forwarding it. When money lands in the bank, it is matched to the invoice.
None of this is unusual anymore. Most of the systems a growing brand already runs can be connected, and where they cannot, AI has made the messy middle much easier: reading the emailed order, matching the product codes that do not quite line up, flagging the one line that needs a human.
You do not need more software and you do not need more admin staff. You need the systems you have to do their job.
A five-minute check
Five questions. Each yes is a place where a person is currently carrying a carton across the room and back.
- Does anyone download a spreadsheet from one system to upload it into another more than once a week?
- Does anyone check whether orders have shipped, or send tracking numbers, by hand?
- Have you sold something online in the last month that you did not physically have, because the stock figure lagged?
- Does it take more than a week after month end to know the month's result, because the team is still matching invoices?
- If the operations manager was off sick for a week, would ordering and shipping fall over?
Three or more and it is worth counting properly: which tasks, how often, how long, how many people. That count is the business case, and it is the first thing the AI Ops Audit produces. Some of what it finds should be automated away. Some of it is faster to fix by training the person doing it. The count tells you which.