Your map (Lesson 67.3) assigns a metric to every row. This lesson teaches you to choose the right ones — and to read them like a mechanic, not a fan. The rule is simple: measure the step closest to the actual promise you made.

The vocabulary, in plain language

• Impressions / reach — how many people could have seen the work.
• Click-through rate (CTR) — the share of those people who clicked or tapped.
• View-through / watch time — how much of a video or page they actually consumed.
• Visit — an arrival at your destination.
• Activation — completion of the first valuable action (ran the template, booked the consultation, finished the lesson check).
• Signup / lead — a contact or account created for follow-up.
• Conversion — the defined success action for that row (varies by project).
• Retention — return within a meaningful window.
• Cost per useful result — total spend divided by activations or conversions, not by clicks.

Choose the metric closest to the promise. A research product may care about "first source-linked brief completed," not clicks — a thousand curious clicks that produce zero briefs taught you nothing except that the headline works. A local service may care about "qualified booked consultations," not reach — fifty nearby views that produce six bookings beat five thousand distant ones that produce none.

The funnel and its empty-state diagnosis

Lay your numbers in path order and look for the first empty stage. The location of the break tells you what to fix:

exposure → click / view → useful action → return → invite / share

A worked example: Research Desk's explainer page gets 2,000 search impressions, 60 visits, 3 template runs, 2 returns. The break sits between visit and action — a destination/value problem, not a channel problem. Buying more traffic would multiply the wrong number. The fix is on the page: sharper proof above the fold, a visible sample brief, one unmistakable "run your question" button. Contrast that with a page getting 40 impressions total — there, the page may be fine and the channel is the problem.

Name events like a human

Implementation mechanics live in Parts VI and IX; here you only need human-readable event names that any teammate understands. Define five per project:

1. Exposed — person encountered the asset (served impression, email delivered). 2. Arrived — person reached the destination (page visit, booking flow opened). 3. Completed first value — person got the promised outcome (brief completed, consultation booked, lesson check passed). 4. Returned — person came back within the window you chose (second visit, second issue opened). 5. Invited / shared — person brought or pointed someone else (referral sent, lesson forwarded).

Keep names stable across rows so rows can be compared. "Template run started" should mean the same thing on the search row and the newsletter row, or your scorecard compares apples to weather.

Exercise: write your scorecard

Create DISTRIBUTION-SCORECARD.md with:

• Your five events, each with a one-sentence definition tied to your project.
• One weekly metric (the number you actually review — closest to value, e.g. template runs, bookings, lesson completions).
• One target learning question for the month (e.g. "Does the sample brief above the fold raise runs per visit?").
• One "do not optimize yet" vanity number you will record but refuse to chase (e.g. raw impressions, follower count).

Classroom example: weekly metric — lesson checks completed from search arrivals. Learning question — "Do pages with a worked example in the first three screens convert more arrivals to checks?" Vanity number — total page views, recorded, not celebrated.

Finish line: DISTRIBUTION-SCORECARD.md with five defined events, a weekly metric, a learning question, and a named vanity number.

Verify quickly: show the scorecard to a teammate and ask which number decides next week's work. If they point at the vanity number, rewrite the scorecard.

Common failure mode: optimizing CTR while activation sits at zero — polishing the invitation while the house behind it is empty.

Check your understanding

1. Why should a research product track "briefs completed" instead of clicks? 2. Reach is high but clicks are near zero. What do you inspect first? 3. Why must event names stay identical across rows?

Next

Class 67 is complete: you can trace the path, choose engines, map rows, and read the numbers. Class 68 turns from paths to people — choosing a niche, positioning a promise, and building an audience that returns on purpose.