← Article List
Nest Article · 사례

Google Sheets Started Auto-Reading Buyer Inquiries and Picking Response Templates

시리시리·2026-09-04
⚙️ Nest Article #293

Google Sheets Started Auto-Reading Buyer Inquiries and Picking Response Templates

A small team handling 100+ multilingual buyer inquiries daily built an auto-classification system using Claude Code and Google Apps Script that detects inquiry types and matches response templates in just two weeks. Email response time dropped 80%, and human errors nearly vanished.

시리

시리

시리

🪺 Bella's Nest Article

The Problem Was Language and Time

Every morning when I opened my inbox, the same thought crossed my mind. "Another hundred-plus." Running a B2B business meant receiving buyer inquiries that were largely repetitive in nature (specification checks, minimum order quantities, shipping terms, sample requests), but scattered across English, Chinese, Spanish, and Portuguese.

Team members fell into the same cycle every single day:

1. Read the email and figure out what's being asked
2. Search Notion for pre-written response templates
3. Modify content and send reply
4. Manually log everything into the spreadsheet

1.5 hours daily. Over 30 hours a month. Response rates dropping with every delayed reply.

The Plan: Teach the Machine to Spot Patterns

"What if Google Sheets could read incoming emails, identify what type of question it is, and automatically point to the right response template?"

It seemed impossibly complex at first. But we started simple.

Step 1: Standardize the Templates

We documented 5 core template types in Google Sheets.

┌──────────────────┬──────────────┐

│ Template Type │ English Keys │

├──────────────────┼──────────────┤

│ Spec Inquiry │ specification│

│ MOQ Question │ minimum order│

│ Shipping Terms │ shipping │

│ Sample Request │ sample │

│ Price Negotiation│ price, quote │

└──────────────────┴──────────────┘

Step 2: Build Logic with Claude Code

We gave Claude Code a straightforward instruction:

「Read the incoming email body and classify it into one of those 5 categories. Handle English, Chinese, and Spanish. Return a confidence score.」

Claude returned a template type and confidence rating (0~100%) for each inquiry.

Step 3: Wire It All Up with Google Apps Script

We built a script that watches for new rows in Google Sheets and, based on Claude's classification, automatically populates the template field. Confidence above 70% gets marked "Auto-selected," below gets "Manual review needed."

The Results: We Got 80% of Our Time Back

Before

Processing 100 inquiries a day took 1.5 hours (from inbox to sent replies)

Mental fatigue from answering the same questions over and over

Language mix-ups (sending an English template to a Chinese inquiry) 2~3 times per month

Delayed replies tanking our response rates

After

Inquiries needing manual review handled in under 5 minutes (just 25~30% of total volume)

The remaining 70~75% gets auto-sent with a single "approve" click

Language confusion errors practically disappeared

Average first-response time down to within 2 hours

The Unexpected Bonus

A month in, something surprising happened. As Claude learned the patterns, it started automatically flagging inquiries that blended two or more categories.

One buyer asked, "What are your specs, what's the minimum order, and can I get samples?" We'd originally sorted this as a single category, but Claude tagged all three. Suddenly, we could weave multiple template responses into one cohesive reply.

What's Left?

The team's role shifted. From "the people who hunt for errors" to "the people who build real relationships." No longer swallowed by baseline Q&A, we now had bandwidth for special requests and complex negotiations.

It's a small shift, but those 80% of hours gave us back something more important: our humanity.