I read 1,400 support tickets in one weekend.

We had 1,400 tickets in our help desk from the previous 8 months. I exported them all to a spreadsheet on a Saturday and read every subject line and first message. Took about 11 hours total.

Findings:

• 412 tickets (29.4%) were about the same thing: a password reset email that landed in spam for Gmail users on custom domains. We had a help article about it. Nobody read it. They just emailed support.

• 233 tickets (16.6%) were feature requests. Of those, 178 were for one feature we had already built and shipped 5 months earlier. Users couldn't find it.

• 97 tickets (6.9%) were from users on a plan that didn't include the feature they were asking about. They had never seen the pricing comparison.

• Only 38 tickets (2.7%) were actual bugs.

The biggest insight: 45% of our support volume was caused by discoverability problems, not product problems. We were treating symptoms. We fixed the spam issue by switching to a different sending domain and adding an in-app "check spam" banner. Ticket volume dropped 31% the next month.

When is the last time you read your own support tickets end to end, without filtering?