Periodic Table of Errors
Every error code is an element; every fix must prove its half-life.
The job
Turns error codes across many platforms into a working list where each fix is proven by the code fading from customers' words. Every error gets one agreed answer in several languages, and the saving is counted in contacts and dollars.
The moment
Error codes are laid out like a periodic table, each lit by how often customers mention it. Error code 1003-An glows across three builds of the TV app.
Beside it sit 14 different explanations from agents, none in the customer's own language. The documentation lead replaces them with one answer in several languages, sent to the AI agent and the agents' ready-made replies.
After app build 5.2.1, customer mentions of the code fall by half every 26 hours.
A week later it turns a steady grey: 1,200 contacts a month gone and $46,000 saved.
What it does
- File the fix in Jira
- Play the cure
- Record fix build and half-life
- Open element verbatims
What you see
A periodic table of error codes with a decay chamber per fix, a wall of conflicting answers that collapses into one canonical card, and a monthly knowledge-health page
What it moves
Contacts per 1,000 active users for each error code, the hours it takes for mentions to halve after each fix, and the dollars saved each month from contacts no longer needed.
Built for
- Product & digital
- QA, training & knowledge
