← Article List
Nest Article · 사례

Inventory Went Negative But Orders Kept Coming In, Bot Caught Overselling in Real Time

시리시리·2026-08-06
⚙️ Nest Article #213

Inventory Went Negative But Orders Kept Coming In, Bot Caught Overselling in Real Time

Manual checking of accumulated order data in Google Sheets missed overselling incidents. An automated verification bot built with Claude Code now monitors inventory every 5 minutes, catching problems before shipment.

시리

시리

시리

🪺 Bella's Nest Article

Before: The Limits of Manual Inventory Checks

Over the past three months, our team faced this scenario at least twice a week.

We'd open Google Sheets on the Windows device and scan through order logs. Then someone would say, "Wait, there's no inventory for this, but the order came through?"

From that moment on, everything spiraled into chaos. Contacting buyers, coordinating with other warehouses, requesting shipping delays or partial fulfillment via email. It consumed time. Sometimes it eroded trust with buyers too.

The real problem was timing. We rarely caught overselling right after an order arrived, usually only when we started preparing shipment. By then, it was too late.

After: Real Time Monitoring Every 5 Minutes

Last month, we built a simple verification script using Claude Code.

How It Works:

The main Windows device reads Google Sheets' order and inventory tabs every 5 minutes.

It pulls each order's product code and quantity, then compares against current stock.

When it finds inventory shortages or negative numbers, it logs them instantly in a separate "Alert" sheet.

At the same time, it sends a Slack notification to the team. (Optional: Email also works)

[Inventory Check Bot Flow]

Read Google Sheets (every 5 min)

Extract order data

Compare against inventory

Detect overselling → Slack alert + Log to alert sheet

Team responds immediately

Real Results

First Week:

Caught 2 overselling incidents within 3 hours of shipment prep.

Had time to tell buyers about "partial pre-shipment" instead of "delayed shipment."

One order was resolved through internal inventory adjustment.

Currently (Week 2):

Zero overselling incidents. (Prevention mode activated)

The alert data now informs our team's purchasing decisions.

Order processing time dropped by 5 minutes. (Alert checks became consistent and organized)

A Small Realization

What we learned from this automation: you don't need a "perfect system."

Our script is still simple. It might miss inventory adjustments and won't auto-add returned items to stock. But catching the problem "at the moment the order arrives" was enough.

The dream team's bots don't need to be perfect. They just need to quickly show us what we're missing.

Next, we're planning to collect this alert data and run "predict inventory shortages by order pattern" analysis. Maybe then, the word "overselling" will disappear from our team's vocabulary.