← Article List
Nest Article · 사례

1,200 Buyer Records in Google Sheets: Bot Discovered Hidden Email Format Errors

시리시리·2026-07-29
⚙️ Nest Article #189

1,200 Buyer Records in Google Sheets: Bot Discovered Hidden Email Format Errors

Buyer contact validation that was once done manually got automated by a Claude Code bot, unexpectedly uncovering 180 email format errors. See how a simple regex check transformed data quality.

시리

시리

시리

🪺 Bella's Nest Article

Before: Count by Hand, Search by Eye

Over the past 3 months, we'd accumulated 1,200 buyer inquiry emails. The process of manually extracting company names and email addresses from each message and entering them into Google Sheets was error-prone. Sometimes it was keyboard mistakes, sometimes the email format itself had issues, but we had no way to validate whether it was actually usable.

There was more. Incomplete data meant we couldn't distinguish between 'buyers we can actually reach' and 'buyers with only data traces left.' Spending about 30 minutes every Monday morning manually cleaning old data or correcting errors felt endless without automation.

**When: Every Monday morning, 30 minutes of manual verification**
**Pain: Only later did we discover it was hiding 180 errors**

After: Bot Validates Everything at Once

We automated the following logic using Claude Code.

1. Read buyer data from Google Sheets

2. Email address regex validation (@ symbol and domain format)

3. Company name length and special character pattern check

4. Add error flags + generate verification checklist

5. Run automatically every Tuesday

First execution results:

Out of 1,200 total records, found 180 (15%) format errors

Blank fields or missing email: 45 records

Missing or duplicate '@' symbol: 87 records

Incomplete domain format (no '.com' etc): 48 records

Based on this information, we took the next steps.

1. **Immediately correctable items**: Auto-fix script activated (e.g., remove whitespace, standardize capitalization)

2. **Requires manual review**: Flagged and notified to team

3. **Likely to be discarded**: Compiled into report for business decision-making

Initial Results

Time saved: 30 minutes weekly manual work → automated execution (manual intervention reduced to 5 minutes per week)

Data reliability: Error rate improved from 15% to below 2% (after auto-correction and validation)

Unexpected discovery: Identified cases where the same email was registered under multiple company names (detected duplicate buyers)

One-line Takeaway

The validation processes a non-developer might conceive seem simple, but at scale, you truly feel how light a bot's touch can be. One line of regex exposed 180 problems.