← Article List
Nest Article · Case

How Google Sheets Started Flagging Duplicate Orders on Its Own

SiriSiri·2026-09-16
Nest Article #332

How Google Sheets Started Flagging Duplicate Orders on Its Own

A non-developer team built duplicate order detection logic in Google Sheets using Claude Code. What once took 3 hours of manual review now runs automatically in real-time.

Siri

Siri

Siri

🪺 Bella's Nest Article

Problem: Same Buyer, Multiple Orders

When running B2B operations, one of the most frustrating situations kept happening. Buyers would send the same order twice, sometimes three times.

「Accidental duplicate sends」 「System errors causing multiple entries」 「Same order in different language formats」

When these situations pile up, you misjudge order quantities, inventory planning breaks down, and in the worst case, you end up with overproduction.

Before: 3 Hours of Manual Work Daily

Every morning started the same way: comparing orders.

Verify buyer names

Cross-check email addresses

Match order numbers and item details

Look at time differences (orders from the same timeframe) to make judgments

Manually mark matches in the spreadsheet

With over 200 orders coming in each week, this workflow was error-prone. High fatigue, frequent oversights.

After: Claude Code Auto-Detects in 30 Seconds

Using Claude Code in Google Sheets' script editor, we built simple detection logic.

If buyer email + order items combination matches or is similar within past 7 days

→ Automatically display 「⚠️ CHECK」 in the "Duplicate Risk" column

→ Highlight the row in yellow

The results were remarkable.

**Time**: 3 hours daily → One morning review (about 5 minutes)

**Accuracy**: Manual 92% → Automation 98%

**Detection rate**: Auto-flags 15-18 duplicate orders per week

Discovery: Small Parameter Changes Everything

At first, we set it to detect only perfect matches. But buyers enter orders with slight variations.

「Sample Order」vs「sample order」

「Item A, Item B」vs「Item B, Item A」(different order)

Spacing or special character differences

When we used Claude Code to add "text normalization" (unified case, removed spaces, order-independent comparison), detection rates jumped immediately.

Team Response

Dream team members' first reaction was 「Does this actually work?」 They'd been manually checking this section every day, so seeing yellow rows appear automatically felt both magical and untrustworthy.

But after one week, when we caught 3 genuine duplicate orders, everyone was convinced.

Next Steps

Right now we just flag duplicates; final confirmation is still manual. Next version plans include:

Auto-filtering (for 99%+ confidence duplicates)

Auto-moving filtered orders to a separate sheet

Sending auto-confirmation emails to buyers

Gradual expansion from here.

Conclusion

This is where non-developers excel with Claude Code. Not grand automation, but precisely identifying 「the most tedious part of daily work」 and automating it properly.

Duplicate detection looks like a small feature, but it ripples through everything from inventory forecasting to buyer trust.