The Days of Manually Copying and Editing 50 Emails Every Week
Until last week, our routine looked like this.
Open template email , delete company name , manually type buyer name , change product link , save , open next template , repeat.
With 50 buyers, we repeated 50 times. Every week. Monthly, 200 times. Was this really meaningful work? We couldn't help but ask.
Before: Time Consumed by Repetitive Labor
Situation
• Every week, 50 to 80 new buyer records were entered into Google Sheets
• Each buyer required a customized introduction email
• Email structure: greeting + company introduction + product category recommendation + deadline notice
• Manual work hours: 4 to 5 hours per week
• Error risk: Had mistakenly sent emails with wrong buyer names twice
Pain Points
Repetitive copy-and-paste work caused escalating fatigue. We once accidentally modified a template, causing customer names to be omitted, and 5 emails were sent with errors. There was also an instance where we deleted a recipient's name but realized it only after sending.
---
After: Bot Takes Over
Automation Structure
Step 1: Data Validation
Google Sheets Buyer Information
(Name, Company, Country, Interest Category)
↓
Claude Code Reads
↓
Check Missing Values + Format Validation
Step 2: Email Auto Generation
We gave Claude Code these instructions:
• Insert 「{Buyer Name}」 exactly 3 times (greeting + body + deadline)
• Select shipping cost template based on 「{Country}」
• Auto insert product recommendation links matching 「{Interest Category}」
• Validate spacing and punctuation
Step 3: Export Generated Emails
Claude Code automatically populates email content into a new column in Google Sheets.
Results
| Item | Before | After |
|------|--------|-------|
| Weekly Work Hours | 4-5 hours | 8 minutes |
| Time per Email | 5-7 minutes | Automatic (2-second review) |
| Error Rate | ~2-3% | 0% (3 weeks post-automation) |
| Weekly Volume | 50 emails | 80-120 emails |
What surprised us most?
Email subject lines were also generated automatically. We thought subject lines needed manual writing, but Claude Code analyzed buyer location and interests, and generated those too.
---
Small Mistakes and Learning
First Attempt: Tone Too Casual
Without specific instructions, Claude Code generated overly informal emails.
「Hey! You've got some amazing products there, wanna check them out?」
Not suitable for B2B emails. We revised the prompt to explicitly state "formal yet friendly tone."
Second Attempt: Link Errors
When auto inserting category links, some buyer interest categories had typos in the original data. For example, a field marked "Shirt" was actually "Shirt/Pants," but the system only recognized the first word.
Solution: Added a validation step for the category column in Google Sheets before Claude Code execution.
---
Our Current Workflow
Every Monday morning at 8 AM, Claude Code automatically generates the email list. We spend 30 seconds scanning through. If there are typos or odd phrasing, we fix manually and click approve.
By Tuesday, 80 emails are automatically sent.
This allows us to focus on meaningful work. Reading buyer responses and assessing partnership potential. Thinking about product improvements.
---
Tips: Verify These Before Automation
• ✅ Is data consistently entered? (Check for typos, blanks)
• ✅ Are generation rules clear? (When, where, what tone)
• ✅ Is there a validation step? (100% automation is risky)
• ✅ Is there a rollback plan if something fails?
Perfect automation doesn't exist. But through 80% automation, we reclaimed 20% of our mental energy.