Early on they hosted FeedbackPanda on a small German Docker hosting startup, which caused severe outages a few months after they had paying customers.

“We had made an infrastructure choice that came back to haunt us just a few months into having paying customers”

The hosting provider suffered repeated, unrecoverable outages, including one lasting almost 24 hours.

“we experienced hour-long outages that our hosting provider struggled to recover from”

FeedbackPanda's browser-extension integrations broke whenever the Chinese online-teaching platforms it plugged into changed their interfaces, with no advance warning.

“the Chinese Online Teaching platforms that we integrated with would change their interfaces”

There was no way to predict platform changes ahead of time, so the team was permanently reacting rather than getting ahead of it.

“It was a cat-and-mouse game that we couldn’t ever really win”

Arvid never hired a customer support agent or another developer to share the operational load, even as it became unsustainable.

“My biggest mistake was not hiring anyone to help, either a customer service agent or another developer”

He kept assuming he could personally absorb the growing workload, which was not true and cost him his mental health.

“I thought I could still keep up with it even when I was extremely anxious and making mistakes because of it”