Field notes. John Arndt, Soxoa. Published .
Most small companies get plenty of feedback. It just lands in six places: a forwarded email, a text from a customer's spouse, an app-store review, a note from a sales call, a comment someone made in passing that you meant to write down. By the time you sit down to plan the next month, none of it is in front of you. So you plan from memory, and memory favors the loudest customer and the most recent complaint.
The fix is boring on purpose: one inbox, one file per issue, one ranked list.
How it works
Every piece of feedback becomes one row with the customer's exact words, who said it and where it came from. Nothing gets paraphrased. "The export is missing dates again" stays exactly that, because the wording is evidence. "Again" tells you it's a regression, and a summary would have dropped that.
Each row then gets attached to an issue. Three people describing the same broken export in three different ways become one issue with three quotes and a count of three. The issue file has the quotes, a "done when" line a non-engineer could check, and, eventually, a note on the fix that shipped.
From those files the skill rebuilds two lists: one per product, and one across everything you run, most-asked first. That list is what you read when you decide what to do next.
The rule that makes it work
The list is for reading. It doesn't decide anything.
Nothing gets built because it's on the list. An AI agent with a ranked list of complaints and access to your code will happily start fixing all of them, and some of those fixes will be wrong for your business. So the skill writes the list, reports the top three, and stops. You decide. When something does ship, the issue gets a fix note with the branch and commit and is marked shipped. Next month you can see what you actually closed.
The other rules are small:
Count people, not messages
One customer emailing three times is one ask.
Store as little personal data as possible
A first name is enough. Phone numbers and addresses that arrive inside forwarded messages get dropped.
Never edit the quotes or the generated lists by hand
If the list looks wrong, fix the grouping.
Starting by hand, automating later
The skill works with no infrastructure at all. Paste a customer email into Claude, say "log this for Acme", and it appends the row, files the issue and updates the list. That's enough for most teams for a long time.
Once the habit is in place, the same files can be fed automatically: a forwarding address per product, a phone shortcut for texts, and a nightly job that pulls app-store reviews and regroups everything. We run it that way across the products Soxoa operates. The automation is the easy part. The rule that it writes the list and not the code is what matters.
Get the skill
feedback-inbox is part of the free Soxoa Ops plugin for Claude. In Claude Code, add the marketplace with /plugin marketplace add johnarndt/soxoa-claude-plugins, then run /plugin install soxoa-ops@soxoa-plugins. You can read the skill on GitHub before installing it, or copy its folder into ~/.claude/skills/.
If you'd like help wiring it into your email, support desk or review accounts, book a call.