100 Buyer Emails Auto-Sorted by Priority in 2 Seconds: The …

Why Email Becomes the Biggest Bottleneck in Automation

Every morning when opening the Windows desktop device, the first thing to do is check emails. Between 200 and 300 English emails pile up in the inbox, yet fewer than 10 represent genuine trading opportunities.

Just determining「who is a real buyer」takes 2 hours per day.

There were patterns, though.

• Repeated inquiries usually aim to negotiate prices

• First emails with specific quantities and shipping addresses indicate serious buyers

• Messages asking duplicate questions should be deprioritized

• Response patterns vary by time zone

But manually sorting hundreds of emails daily wasn't feasible.

Before: The Manual Filtering Struggle

The old workflow:

1. Receive emails

2. Quick scan of subject lines and opening paragraphs

3. Manually record in spreadsheet (sender, product, quantity, notes)

4. Manually adjust priority column

5. Repeat the next morning

The problems:

• Frequently missed follow-up emails from the same person

• Weekend emails pile up to 150+ by Monday morning

• Personal mood affects priority judgments

• Easy to miss subtle nuances in English phrasing

After: Claude Code + Gmail API + Google Sheets

Setup (Non-developers can do this)

Step 1: Create a label in Gmail

Label name: 「Inbox_Auto」

Filter settings: Auto-classify all emails

Step 2: Claude Code script (basic framework)

Google Sheets → Gmail API connection → Read emails → Claude analysis → Scoring → Write back to sheet

I assigned Claude these tasks:

1. **Email content analysis**: Purchase intent scoring (1-10 scale)

2. **Duplicate detection**: Check if sender has previous emails

3. **Quantity extraction**: Auto-parse mentioned order volumes

4. **Response speed evaluation**: Assess seriousness based on time zone and content specificity

5. **Multi-language detection**: Handle emails mixing Korean, English, and Chinese

Results

Before (manual):

• Time: 2 hours daily

• Accuracy: ~75% (frequent oversights)

• Buyer tracking: Nearly impossible

After (Claude Code automation):

• Time: 2 seconds (for ~300 emails)

• Accuracy: ~92% (AI catches subtle expressions)

• Buyer tracking: Automatic (linked by sender)

Auto-populated Google Sheets columns

|--------|---------|----------|---------|---------|---------|--------|----------|

Unexpected Discoveries

1. Time Zone Patterns Emerged

AI automatically revealed: European buyers sending emails at 6~8 AM show higher intent, while Asian buyers sending at 10 PM~12 AM provide more specific details. Now we prioritize those time windows.

2. The Line Between "Just Asking" and "Ready to Buy"

Starting with just numbers, AI learned to distinguish:

• 「Can you send me...?」usually just browsing specs

• 「Please send quotation for...」shows serious intent

• 「How many do you have in stock?」suggests actual purchasing power

3. Response Timing Matters

Within 2 hours of receiving an email, intent scores auto-generate. Key buyers get instant alerts, and switching to this method tripled our reply rate.

Where the Team's Saved Time Goes

The 2 daily hours spent on email sorting vanished.

Before: Email sorting 2 hours + Actual sales 1 hour

After: AI sorting 0.05 hours + Actual sales 2 hours

Now that time goes to crafting detailed custom proposals or market research.

Can Non-Developers Do This

Difficulty: Medium-High

• Gmail API connection completes in 15 minutes with Claude Code support

• Scoring criteria need custom definition

• Execution remains fully automated afterward

Our approach: A Mac Mini runs 24/7, executing the script once daily. Windows device and Mac Mini work in tandem.

Bottom Line

This automation isn't complicated technology. But for「repetitive judgment tasks,」AI excels. Proven again.

Next: automating email reply composition.

100 Buyer Emails Auto-Sorted by Priority in 2 Seconds: The Unexpected Win