An Excel Sheet at Two in the Morning
"The orders came in, but the inventory doesn't match." That was a message I got last week from a small online store owner. The time was two in the morning. There he was, laptop open in a corner of his office, eleven Excel sheets spread across the screen, trying to reconcile that day's shipments.
The same product sells simultaneously on his own store, on Smart Store, and on Coupang. When a person has to manually track which channel's inventory dropped first, sooner or later one of them ends up taking an order for a product that's already sold out. That's exactly what happened that day, he told me.
Asking the Same Question of E-commerce
In the previous post, we talked about how automation in manufacturing always gets stuck at quotes and drawings. "Where should human hands stop first?" When I ask the same question of e-commerce SMB owners, the answers cluster around almost the same places: orders, inventory, and CS. Those three.
But the right starting point differs from company to company. I've seen a couple of cases where someone picked the wrong branch point and lost six months because of it.
Three Criteria — Channel Count, Inquiry Volume, Headcount
The criteria are surprisingly simple.
If you have just one sales channel, a person can still handle orders and inventory. The moment you cross two, you quickly leave the line a single person can hold by hand. If you start seeing even one missed shipment a month, take that as the signal. That single missed shipment shakes your refund rate, review score, and repurchase rate all at once.
Inquiry volume is a separate problem. Once daily inquiries pass 30, a human's response speed can't keep up. From that point on, the owner gets pinned to the chat window instead of running the business.
And then headcount. When the team is one or two people, the priority for automation is "the area that keeps me up at night." Past five people, the priority shifts to "the area where everyone is doing things differently."
The Moment You Cross Two Channels
A frequently cited industry survey shows that about 60% of small businesses in the U.S. are already using AI or automation tools in some form in their operations. 60% sounds like a big number, but when you look closer, it's the simplest tasks — printing invoices, sending status-change notifications, syncing inventory.
That's the starting point. It's nearly impossible for a small online store to roll out a full-stack OMS the way a large enterprise does. I used to think that was the right path, but every PoC we ran in the field said otherwise. Just stringing together four things — "collect orders + sync inventory + print invoices + manage status" — is enough to make the owner's early-morning hours disappear.
What "Automating 60% of Inquiries" Actually Means
CS has a different grain. One fashion accessory brand introduced a chatbot connected to its order data and found that about 60% of simple inquiries were handled automatically. 60% may not sound like much, but at 5,000 monthly inquiries, that's 3,000 cases off human hands.
The case of a pet supply store is even more dramatic. Of all their inquiries, 49% were "When is this product coming back in stock?" Once they automated restock notifications, nearly half of all inquiries simply disappeared.
There's a trap here, though. Handing every inquiry over to a chatbot backfires. In one of our own PoCs, we initially aimed for 100% automation and quickly piled up complaints around complex cases like refunds and exchanges. In the end, we only automated the simple, repetitive cases and rebuilt the routing so that anything ambiguous escalated to a human. It's not a 100% silver bullet. But once 60–70% is handled automatically, people finally have time to focus on the remaining 30%.
How Korean SMBs Actually Start
That's the theory. In practice, the combinations small Korean online stores actually use are surprisingly simple. They attach Channel Talk on top of a store solution like Cafe24 or Makeshop, then connect inventory and order flows behind it with n8n or in-house scripts. The strength is pulling store data directly without any additional development.
Nobody installs a big system all at once. One owner automated just the inventory sync "the same week he had a single missed order." The next month he added a CS chatbot, and in the third month he moved on to automatic shipping label printing. That sequence isn't universally correct, but for him, it was the right order.
Honestly, deciding where to start automating is much harder than choosing the tool. You can learn a tool in a month, but "where in this company should things stop first" is something even the owner often can't see clearly.
Next Episode — When the Same Question Moves to HR
At 5years+, we've been running these kinds of phased automation PoCs with e-commerce SMBs across Korea and Japan. Rather than bundling every area at once, we recommend starting from the single point that drains the owner's time fastest. If you're weighing this internally, even a quick snapshot of your current state is enough for us to share our take on which area should come first.
In the next post, I plan to ask the same "where should I start?" question, but in the staffing and HR space. From resume matching to onboarding, the flow of people moves in ways that resemble — and yet differ from — the flow of orders.