← Article List
Nest Article · 사례

Google Sheets Started Reading Orders and Auto-Deducting Inventory

시리시리·2026-09-02
⚙️ Nest Article #283

Google Sheets Started Reading Orders and Auto-Deducting Inventory

A non-developer in B2B export business connected Claude Code with Google Sheets to fully automate manual inventory management. A practical case built in 48 hours, from order intake to stock deduction to buyer confirmation emails.

시리

시리

시리

🪺 Bella's Nest Article

Before

Every morning, opening the mailbox, downloading the order form, then manually organizing the Excel file takes 30 minutes. The inventory system and order system operate independently.

**Typical Flow:**
1. Receive buyer email
2. Download order attachment
3. Manually check quantities and input to Excel
4. Manually deduct from inventory file
5. If 「out of stock」, reply to email again
6. Repeat until all orders are processed

20 to 30 orders per week, and the error rate keeps increasing. There have been cases of double-counting inventory or deducting already-processed orders again.

---

The Discovery(Challenge)

「Can Google Sheets read PDF order forms?」

The question posed by non-developer Bella became the starting point. She opened Claude Code and entered a simple prompt.

Read the order PDF sent by the buyer,

Extract product name, quantity, and requested delivery date,

Auto-paste into the Google Sheets 「Pending Orders」 tab.

24 hours later, the answer was yes.

---

After

Now when an order email arrives, the automation bot springs into action.

**New Flow:**
1. Receive buyer email (no manual intervention)
2. Claude Code auto-extracts PDF
3. Auto-input to Google Sheets 「Pending Orders」 tab
4. 「Real-time Quantity」 in inventory tab auto-deducts
5. Out of stock auto-flagged in red via conditional formatting
6. Confirmation email auto-sent via macro

Results:

Manual work time: 30 min/day → 2 min/day

Order processing errors: 3~4/month → 0/month

Inventory deduction accuracy: 96% → 99.8%

Buyer response speed: avg 4 hours → avg 8 minutes

---

The Unexpected Twist

One month in, the bot started flagging 「suspicious orders」 on its own.

One buyer placed an order 3x larger than usual but left the shipping address blank. Without adding specific conditions to Claude Code, it learned the pattern and automatically moved such orders to the 「Needs Review」 tab.

What was meant to reduce errors unexpectedly evolved into a security feature that catches fraudulent or anomalous transactions before they happen.

---

3 Key Lessons

1. Let humans organize, let bots aggregate

Inventory deduction and quantity calculation don't need human hands. Instead, leaving anomaly signals for humans to read makes everything faster.

2. Automation starts with small experiments

With just a 「let's try once」 mindset, I opened Claude Code. Now it's company-wide process. Bold plans matter less than curiosity.

3. Prepare error logs before automation

The more the bot runs, the more unexpected data flows in. To quickly spot 「something's off,」 establish error logging and reporting channels first.

---

What's Next

The next goal is syncing Google Sheets directly with the inventory management system (ERP) on the Windows device. Already testing API connections.

Non-developers can do this too. All you need is Claude Code.