## TL;DR

Pre-launch feedback from five real users beats a hundred opinions from bystanders.
A structured test script keeps feedback comparable instead of anecdotal.
Applies to any MCP server approaching a public or team launch.

## The query

```text
how to get feedback on an MCP server pre-launch
```

## Use this when

- An MCP server is feature-complete but unreleased.
- You can recruit a handful of real prospective users.
- You want friction found before launch, not after.

## Not for

- The server is already launched (run post-launch feedback instead).
- You only want praise, not criticism (skip the exercise; it will not help).
- There is no one to test it yet (dogfood it yourself first).

## Steps

1. Recruit 5 to 8 target users: people who would actually use the server daily.
   Expected output: A tester list of real prospective users, not friends doing a favor.
2. Give them a test script: install, connect, run three representative tool calls, report friction.
   Expected output: Comparable feedback covering the same critical path.
3. Fix the top 3 friction points and tell the testers what changed.
   Expected output: A server with the worst first-run problems removed.
4. Ask two testers to re-run the script and confirm the fixes.
   Expected output: Confirmed improvement before launch, not assumed improvement.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_8nuPObWHZcg8xt5vmHdFEw