The Real Problem Hidden in 1,500 Buyer Messages on Google Sheets
One morning, when I opened Google Sheets on my main Windows device, I discovered something. Three months of accumulated buyer inquiries had piled up: 1,500 messages. At first glance, they all seemed like different questions. Some were about product pricing, some about shipping timelines, some about samples.
"I guess I need to organize these first," I thought, and gave Claude Code an instruction. "Please read through all the messages in the Google Sheet and verify if they're actually all different."
A Pattern a Non-Developer Missed, AI Caught
The result Claude Code returned was shocking.
Out of 1,500 messages, only 847 are truly different in content. The remaining 653 are repetitions of 5 core questions expressed in different ways.
"What? What do you mean?"
Here's what was happening.
• Message A: "Hi, what's the minimum order quantity?"
• Message B: "Our company wants to place a large order. Is there a minimum quantity requirement?"
• Message C: "I'd like to start with a smaller order. Can you tell me the MOQ?"
On the surface, these are completely different messages. But the AI detected that they were all asking the same question (minimum order quantity) using different phrasing.
Why This Happened
When I thought about it, the reason was obvious.
The same buyer asked multiple times using slightly different wording. Or different buyers had similar questions in similar situations but expressed them differently. Messages auto-forwarded from Amazon, Alibaba, and other marketplaces were mixed in, making duplication inevitable.
If I had done this by hand? Finding 653 duplicates would have taken at least 3 hours. Or I'd never have found them at all.
Small Discovery, Big Efficiency
After Claude Code classified the duplicates, we created just 5 template responses, each addressing one core question type.
Now when buyer messages arrive, AI categorizes them first. "This is question type 2," and suggests the matching template. Response speed tripled.
The most interesting part? I started wanting to do something "grand" through automation, but what actually moved the needle was "having AI find what I couldn't see."
A Non-Developer's Tiny Insight
Three years ago, I thought "automation is for developers." Not anymore. If a non-developer asks AI the right question, we can get pattern recognition at a level that's absolutely impossible by hand.
Working with the Dream Team bots (the Hamsters and Puppies), I learned something.
"AI doesn't point out our mistakes. It shows us what we can't see."
So, what should I ask next?