The class built the mission, the pipeline, the evidence discipline, and the compounding library. What remains is judgment over time: which pieces actually earn trust, which support the business, which need correction. Without measurement, those decisions go by loudest opinion — the content equivalent of automating the loudest task.

Quality signals and business signals

Quality signals say the piece worked as teaching:

• Qualified attention: the right readers spending real time — completion rate, deep scrolls, return visits — not raw pageviews.
• Completion: how many finished the lesson, the checklist, the exercise.
• Save/share: readers keeping it or staking reputation on it. Saves predict reuse; shares predict reach.

Business signals say the piece worked as an asset:

• Sign-up: readers joining the list, the course, the community.
• Assisted conversion: the piece appearing on the path to a purchase, booking, or partnership — even when it was not the last click.
• Source use: other pieces, support answers, or sales conversations citing it. A piece the operation itself reuses is load-bearing.
• Update need: the honest negative signal — corrections requested, facts aging, questions revealing gaps. High update need is information, not failure; it tells you exactly where editorial attention pays next.

The discipline from Lesson 84.5 applies: pair them. One quality signal plus one business signal per piece. Quality without business builds a beloved library nobody funds; business without quality builds a funnel that burns trust. When the two disagree — a piece readers love that never converts, or a converter readers never finish — treat the tension as the finding. The first needs a stronger next step; the second needs teaching that earns the click it already gets. Either way, the pair told you something neither signal alone could.

The monthly review

Once a month, review the library — not the calendar — and answer four questions:

1. What earns trust? High completion, saves, citations inward. Protect and extend these pieces; they are the canonical core. 2. What creates useful questions? Pieces generating good backlog entries are succeeding — they teach enough to make readers curious for more. Route those questions into assignments. 3. What supports revenue? Sign-ups and assisted conversions traced honestly, without claiming last-click credit for everything. 4. What needs correction? Update needs, error reports, decayed facts. Schedule the update or retire the piece — a wrong canonical piece poisons every derivative linked to it (Lesson 85.4).

Record each decision: keep, extend, update by [date], or retire with a redirect. The record is the improvement stage of the 84.1 spine, turning measurement into a system that gets better as it grows.

Attribution honesty is the skill that makes the revenue question trustworthy. Content rarely closes a sale alone; it assists — the reader learns from three lessons, asks a question the support docs answer, then books a call. Claiming full last-click credit for every conversion inflates content's value and misallocates effort toward bottom-of-funnel pieces. Instead, record assists: which pieces the converted readers actually touched, in what order, and what job each piece did. Over several months a pattern emerges — often that unglamorous explainer lessons assist more revenue than the showcase pieces. That finding, which only honest measurement produces, is exactly the kind of decision-changing output the 84.5 scorecard demands.

Exercise: score three pieces

Pick three published pieces — ideally one you suspect is strong, one average, one neglected. Score each with one quality signal and one business signal:

# CONTENT-SCORES — [month]

## [Piece title + URL]
- Quality: [signal + value, e.g. completion 41%]
- Business: [signal + value, e.g. assisted 6 sign-ups]
- Verdict: [keep / extend / update by [date] / retire]
- Reason: [one line tying the signals to the verdict]

Finish line: three pieces scored, each with a verdict and a reason a colleague could audit.

Verify: confirm no verdict rests on vanity volume alone, and every "update" names a date and an owner. An update without either is a wish.

Common failure mode: scoring everything "keep" to avoid decisions. Force the distribution: at least one update or retire per review until the library is genuinely current. Retirement with a redirect preserves links and trust; silent rot destroys both. The second failure mode is measuring once and filing the scores — a review whose verdicts never reach the backlog changes nothing. Every update verdict becomes an assignment with an owner and a date; every retire verdict becomes a redirect the same week. Measurement that does not move work is just bookkeeping with better vocabulary.

Check your understanding

1. Why must every piece carry both a quality and a business signal? 2. How does a piece that generates good questions count as succeeding? 3. Why is update need framed as a signal rather than a failure?

Class 85 is complete: an editorial mission answered by a file-based pipeline, produced with AI inside evidence discipline, compounded through reuse, and judged monthly on trust and contribution. Class 86 carries the same operating discipline into marketing — turning demand signals into useful follow-up.