Before (Pre-Automation)
Every Monday morning, one team member stared at dozens of incoming emails.
「Is this buyer new?」 、 「Any past transaction history?」 、 「Is the inquiry specific enough?」 、 「Do they look financially capable?」
Every judgment had to be made manually. Among 300+ monthly inquiries, just prioritizing, selecting response templates, and spotting red flags (typos, vague requests) consumed 30 hours. Many potential buyers went unanswered.
The Small Experiment Begins
The team wondered: "What if Claude Code could read the inquiries in Google Sheets and automatically grade buyer credibility?"
Their first attempt was simple.
• Inquiry message (sender's email)
• Past transaction records (existing data in Google Sheets)
• System prompt: "Rate buyer credibility on a 1-10 scale"
After (Automated)
Now, every Monday morning when the team powers up their main Windows device, Google Sheets is already prepared.
Auto-Generated Assessments
• Credibility score (1-10)
• Buyer type classification (New / Returning / High-risk)
• Risk flags (what concerns exist)
• Recommended response template (ready to use)
• Priority level (High / Medium / Low)
The results are instant. The team now tackles high-scoring buyers first and handles low-scoring ones with auto-response templates.
The Numbers
• Manual analysis time: 30 hours/month → 4 hours/month (management + exceptions only)
• Response time: Average 2 days → Average 4 hours
• Missed inquiry rate: ~15% → 0%
• Team morale: 「Mondays don't feel as daunting anymore」
Unexpected Discoveries
1. Claude Code's 「Intuition」 Was Surprisingly Accurate
When Claude assessed, 「This buyer asked a specific MOQ, so credibility is 7/10,」 the actual conversion rate was indeed high. The non-developer team members were skeptical at first, but after three months of data, Claude's "intuition" nearly matched their experience.
2. Patterns Started Emerging
With months of automated assessments, the team spotted patterns: 「High-credibility buyers typically ask these types of questions.」 They adjusted next month's sales strategy accordingly.
3. It Gets Things Wrong Sometimes (And That's Okay)
There were cases where Claude flagged a buyer as 「high-risk」 who later placed large orders. But the team accepted this. "We don't need perfection," they said. Automation's goal was cutting 30 hours down to 4, not achieving 100% accuracy.
One Takeaway
"Automation doesn't deliver perfect answers. It gives time back to your team."
Those recovered 30 hours now went to what mattered: exploring new markets, deeper buyer conversations, team recovery. Ironically, doing less led to better results.