Check you can trust the scores

Read the agreement figures for your own workspace before you act on a score.

Use this before you tell somebody their agent scores 71, so you know what that number can carry.

Open Scoring quality in the left rail.
Its opening line says how many of your team's verdicts we matched.
Read the five cards across the top.
Read the square under them.
Each cell is one of the four things that can happen to a verdict.
Read the newest verdicts we got wrong.
Each one carries the reason the reviewer wrote.
Press Read it on one, or Change the rule.
Read By score area, weakest first.
Read the How we decide column.
A model decides the area, a stopwatch does, or no live rule reaches it.
Press Open on a row, or Read more.
Read more appears where too few verdicts sit beside ours to compare.
Press For engineers, or Download the method.
The working behind every figure on the screen.

What the five cards mean

How much we looked at	The share of finished conversations that carry a score
We matched your team	How often we landed where your own reviewers landed
What the real failure rate is	A range, corrected for our own known mistakes
Your own reviewers agreed with each other	Whether two of your people who read the same rule landed together
Same conversation scored twice	How far a score moves when nothing about the conversation changed
A blank in the table	Not measured. Never zero

Act on an area that matched a person often and has enough read by a person behind it. An area with too few says so rather than showing a number.
Same conversation scored twice is your floor for change. A move smaller than that figure is not movement.
If an area you need is unmeasured, the fix is examples. Record Agree or Disagree as you read conversations.

Nothing on this screen changes a score, a rule or a setting. Change the rule and Start reading are doors to screens that do.