You probably have a group chat that’s doing more work than it was designed for.
Maybe it’s a Teams channel where employees post leave requests. Maybe it’s a WhatsApp group where the warehouse confirms stock levels. Maybe it’s a Zalo thread where your sales team passes leads to each other by typing a name and a phone number.
Nobody decided to build a system. It just happened. Someone asked a question in the chat, someone else answered, and over time that pattern became the process. Fast, convenient, and invisible to anyone who isn’t in the thread.
The industry calls these shadow processes: informal workflows that grow inside tools that were never meant to be business systems. In most small companies, they’re everywhere.
Sources: Microsoft Work Trend Index 2025, McKinsey “The Social Economy”, University of California Irvine
How group chats become business systems
It always starts the same way. A team needs a fast answer. Email feels too slow. A formal tool feels like overkill. So someone posts in the group chat, gets a reply, and the job is done.
Then it happens again. And again. Within a few months, the chat channel has become the system of record for something important: leave approvals, vendor quotes, delivery confirmations, expense receipts.
Here are the most common ones I see in small businesses:
| Process | Where it lives | What breaks |
|---|---|---|
| Leave requests | Teams or Zalo channel | HR can't build attendance reports |
| Purchase approvals | WhatsApp group with the boss | No paper trail for accounting |
| Client handoffs | Slack channel or group DM | Context lost when someone leaves |
| Inventory checks | "How many X do we have?" in chat | Conflicting answers, no timestamp |
| Shift scheduling | Group message with availability | Manager manually rebuilds roster |
The pattern works because it’s frictionless. No login, no form, no training. That’s also why it’s hard to replace. Any new tool has to compete with the one thing chat does perfectly: zero friction.
The real cost is reconciliation
The chat itself is fine. The problem shows up downstream, when someone needs to turn those conversations into a report, a timesheet, a payroll calculation, or an audit trail.
That’s when you get the spreadsheet phase: someone (usually HR, accounting, or the operations manager) opens the chat, scrolls through weeks of messages, and manually copies information into a spreadsheet. Names, dates, amounts, approval status. One message at a time.
For a 20-person office with ~200 leave events a year, this looks like:
| Task | Who does it | Time |
|---|---|---|
| Writing a formal email per leave request | Every employee | ~50 hours/year total |
| Scrolling chat to find and list all leaves | HR | 1 to 2 hours/month |
| Building the monthly leave summary | HR | 1 to 2 hours/month |
| Updating timesheets and salary from the summary | HR | 3 to 4 hours/month |
That’s roughly 100 hours a year for one 20-person office on one process. And it scales linearly. Double the headcount, double the hours.
McKinsey estimates that knowledge workers spend 1.8 hours every day just searching for information. In a small business, a chunk of that time is people looking for things that were said in a group chat three weeks ago.
The adoption problem
The natural response is to buy an HR tool, a ticketing system, or a project management app. And for some companies, that’s the right call.
But for a 10 to 50 person company, new tools carry a hidden cost: adoption. You need people to log in, to fill out forms, to change their habits. That’s not a technology problem. It’s a behavior problem. And behavior change in a busy team is expensive.
I’ve seen this cycle multiple times:
- Company buys a tool (leave tracker, approval system, task board)
- Half the team uses it for a month
- The other half keeps posting in the chat
- Now you have two systems, neither complete
- Someone still has to reconcile both
The tool isn’t the problem. The assumption is: that people will change how they work because you gave them a better form to fill out. Most of the time, they won’t.
Zero-change automation
This is the idea behind what I call zero-change automation: instead of asking people to use a new tool, you build something that watches the tool they already use and does the filing automatically.
The conversation stays in the chat. The manager still replies “OK”. The employee still types a casual message. Nothing changes for anyone on the team.
Behind the scenes, an automation reads the conversation, understands what happened (who requested what, when, who approved), and files a clean record: an email to HR, a row in a spreadsheet, a notification to whoever needs to know.
The principle: Don't move people to a system. Move the system to where people already are.
This works because:
- No training. If someone can type a message in a group chat, they can use the system.
- No adoption curve. The team doesn’t even know the automation is there. They just notice that HR stopped asking them to re-send things.
- No split system. There’s one source of truth (the chat), and the automation creates the structured version.
- No per-seat pricing. You’re not buying software for every employee. You’re running one automation that reads a channel.
In practice: leave tracking from chat
I recently built this for a garment manufacturer in Hanoi. Their 20-person office was running all leave requests through a single Teams channel: 400+ messages a year, 95% of them leave-related, zero audit trail.
The automation runs every hour. It reads the channel, uses AI to classify each conversation (who’s asking for leave, what type, what date, who approved), and sends a clean notification email to HR plus a row in a Google Sheet.
HR went from spending 5 to 8 hours a month on leave reconciliation to opening a spreadsheet that’s already filled in. Employees went from writing formal confirmation emails to not writing them at all (the automation handles it). Total team time saved: roughly 100 hours a year.
The runtime cost is $0.02 per day.
Nobody on the team changed how they work. The chat stayed the same. The reports just started appearing on their own.
Read the full case study: H.A Fashion
Three signs you have shadow processes
You probably already know if this is you. But here are the common tells:
1. Someone on your team is “the person who knows.” They’re the one who remembers that the client asked for a revision three weeks ago, because they were in the chat when it happened. If they leave, that knowledge walks out with them.
2. Month-end involves scrolling. If anyone in accounting, HR, or operations has to go back through chat history to build a report, you have a shadow process. The chat is the system; the report is the manual reconciliation.
3. You’ve tried a tool and it didn’t stick. You bought a leave tracker, or a project board, or an approval app. Some people used it, some didn’t. The chat is still where the real work happens.
None of these are failures. They’re signs that your team found the path of least resistance. The fix isn’t to fight that instinct. It’s to build on it.
Start with one process
If this sounds relevant, here’s how I’d think about it:
Pick one process. The one where someone is already spending hours on manual reconciliation. Leave tracking, purchase approvals, and order confirmations are the most common starting points.
Don’t replace the chat. Whatever your team is doing today, keep it. The automation reads from where the conversation already happens.
Start with the report. The first thing the automation should produce is the thing someone is currently building by hand: the monthly leave summary, the expense report, the order log.
Measure the before and after. How many hours per month was the manual version taking? How many after? That’s your ROI, and it’s usually obvious within the first month.
I’m a one-person automation studio based in Hanoi. I build these systems for small businesses using n8n and AI. If you recognize any of the patterns in this post and want to talk through whether automation makes sense for your situation, send me an email. The first conversation is always free.