Deb Kawamoto
DebGPT Insights
Talk to DebGPT →
What are DebGPT visitors curious about between meetings?
The deeply personal ones are the most interesting.
Data updated
Live since May 2026
Total conversations
real people, real questions
Came back with a follow-up
%
asked more than one question
Deep conversations (3+ msgs)
%
someone brought their own situation
"How do you pick companies?"
1×
hypothesis: wrong
The hypothesis, and what actually happened
What I expected
Skip the small talk, get to the real stuffDesign leadership, hiring decisions, how I think about leveling. The real questions behind a first coffee.
"How do you pick companies?"My main hypothesis — the question I thought would define the conversations. It came up exactly once.
No prompt needed. Just ask.No guardrails. I assumed unlimited access would pull out the interesting stuff.
What happened
They wanted to know how the trick worked of messages. Satisfied, then gone.
The question came up. Once.When it did, it wasn't abstract — someone two months in, already hating it, asking what to do.
Most people frozeThe ones who didn't had already decided what they wanted to ask before they opened the page.
What people asked about
— conversations

Proportions from full conversation log. The top category — mechanism questions — drove volume, not depth.

What drives depth
Deep (3+ msgs)
Surface (1 msg)

MOCKUP — illustrative numbers, not wired to real per-topic depth data yet.

Key signals from the data
100%
Every deep conversation started with someone bringing their own context. Not a question about me. A personal problem they were dealing with that had nuance.
Virality ≠ depth
Reach and engagement aren't the same thing. The LinkedIn spike tripled mechanism questions. The depth rate didn't move.
All or nothing
63% sent one message and left. The ones who stayed averaged 7. There's almost no middle.
Cold start
Most people freeze with unlimited access and no starting point. The ones who didn't had already decided what to ask before they opened the page.
Deb Kawamoto  ·  chat.debkawamoto.com  ·  Data reflects full conversation log, updated periodically  ·  No raw message counts or personal conversation content is shown