Chapters · 5 of 8

About 14 min · Lesson and activity

Chapter 5 of 80 complete

Treat Confidence Carefully

Say what a confidence number does and does not mean, and keep a composite score’s parts recoverable.

See the situation

Start with the situation below. Next, we will work through an example together before you try the activity.

Maya · Decision designer

The case

Old deal scoring 22

A confidence display needs a measured meaning. Evidence sufficiency and probability of correctness are separate questions.

Displayed score: 22
Base after stage penalty: 0 (floored before email points are added)
Email points: 22
Reads as: A somewhat healthy deal
Actually means: Two emails and no meeting momentum at all

Your task

Compare the confidence display with the required evidence. Decide whether the result can be released.

Need a little guidance?

You’re done when: The result distinguishes “strong signal” from “enough evidence to make this decision.”

A confident-looking number is dangerous when required evidence is still missing.

Need a reminder? Reopen the worked example

Watch one example first

A model reports 0.9 confidence, but the required company source is absent. The output may sound certain while the decision remains unsupported under the evidence policy.

Why this follows from the evidence

Confidence has meaning only with a stated basis. Evidence sufficiency asks a different question: are the required inputs present? Renaming 0.9 “high” does not supply the missing source.

GTM Lab

Saved locally

Decision Design · 0 of 8 sections started

Saved locally to your browser.

Decision workbench

Inspect a score and its evidence coverage

Your task: Change one score component and distinguish the displayed signal from the observed result of a small prepared sample.

Email-activity points · 22

One component changes; evidence coverage does notCompare the original state with the proposed change. This preview does not change live records.

One component changes; evidence coverage does not
Item	Before	Current fixture	Why
Displayed score	22	22	This bounded example adds email points to the supplied component total.
Required evidence coverage	2 of 3 inputs	2 of 3 inputs	Changing one score component does not connect the missing meeting source.
Meaning of 8 of 10 sample outcomes	Not calibrated from one example	80% observed agreement in this prepared sample	A sample proportion is not automatically a future probability.

Displayed score: Before22; Current fixture22; WhyThis bounded example adds email points to the supplied component total.
Required evidence coverage: Before2 of 3 inputs; Current fixture2 of 3 inputs; WhyChanging one score component does not connect the missing meeting source.
Meaning of 8 of 10 sample outcomes: BeforeNot calibrated from one example; Current fixture80% observed agreement in this prepared sample; WhyA sample proportion is not automatically a future probability.

Small prepared reliability sampleOne mark per supplied example record. Filled = counted; outlined = included, not counted; dashed = excluded.

8 of 10 included records (80%)

Prepared outcome 1: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 2: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 3: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 4: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 5: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 6: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 7: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 8: Included and counted. The recorded outcome matched the predeclared target in this sample.
Prepared outcome 9: Included, not counted. The recorded outcome did not match the predeclared target.
Prepared outcome 10: Included, not counted. The recorded outcome did not match the predeclared target.

What does this sample directly show?

The case still lacks a required input. What should happen to the action?

Practice draft restores when you return.

Case evidence

Old deal scoring 22

Displayed score: 22
Base after stage penalty: 0 (floored before email points are added)
Email points: 22
Reads as: A somewhat healthy deal
Actually means: Two emails and no meeting momentum at all

In this example, answers above 0.8 route automatically. A later review finds roughly one in seven of that sample was wrong, but the score was never evaluated as a probability.

Review or edit the saved decision-design sections

The workbench above is the main practice. These optional sections let you draft the contract directly; each section gives feedback before the chapter check.

Work the case · 0 of 4 sections drafted

State what the confidence measure means

Compare these alternatives one section at a time. Check a choice to see its consequence; the draft summary below keeps the proposed changes together.

Step 1 of 4

No evaluation sample has compared confidence figures with outcomes. How should the scale be presented?

The raw number, so nothing is lost in roundingBands with stated meanings: strong, mixed, and weak signalDo not show an uncalibrated number as probability; label the measure or withhold it

This is a proposed draft until the chapter check runs. Selecting a choice alone does not complete the chapter.

2b · Proposed

Decision design draft

Decision design draft · current draftCompare the original state with the proposed change. This preview does not change live records.

Decision design draft · current draft
Item	Before	Proposed	Why
No evaluation sample has compared confidence figures with outcomes. How should the scale be presented?	No saved change in this exercise draft	Not drafted yet	Choose a proposal to see what it would add to the draft.
A confident answer built on two of five required inputs. Should it act automatically?	No saved change in this exercise draft	Not drafted yet	Choose a proposal to see what it would add to the draft.
A stalled deal with a recent email thread scores 22, because the stage penalty ground the base to zero before email points were added. What fixes the number being misread?	No saved change in this exercise draft	Not drafted yet	Choose a proposal to see what it would add to the draft.
A segment average of 58 is about to be published as a benchmark. What has to travel with it?	No saved change in this exercise draft	Not drafted yet	Choose a proposal to see what it would add to the draft.

No evaluation sample has compared confidence figures with outcomes. How should the scale be presented?: BeforeNo saved change in this exercise draft; ProposedNot drafted yet; WhyChoose a proposal to see what it would add to the draft.
A confident answer built on two of five required inputs. Should it act automatically?: BeforeNo saved change in this exercise draft; ProposedNot drafted yet; WhyChoose a proposal to see what it would add to the draft.
A stalled deal with a recent email thread scores 22, because the stage penalty ground the base to zero before email points were added. What fixes the number being misread?: BeforeNo saved change in this exercise draft; ProposedNot drafted yet; WhyChoose a proposal to see what it would add to the draft.
A segment average of 58 is about to be published as a benchmark. What has to travel with it?: BeforeNo saved change in this exercise draft; ProposedNot drafted yet; WhyChoose a proposal to see what it would add to the draft.

A choice appears here as a proposal, not as evidence that a live system changed.

Your place saves as you go.

Chapter 5 of 80 complete