Lesson 86.3 demands every campaign carry a brief, a clock, and a learning question; Lesson 86.5 demands reporting that changes the next decision. This appendix is both documents in one — copy, fill, run the two-week test, report honestly.

CAMPAIGN-BRIEF.md (fill before spending anything)

# CAMPAIGN-BRIEF.md — [Name] — dates ___ to ___

## Objective (one) + audience (one) + promise (one)
- Objective: ___ (e.g. "20 trial starts from Models readers")
- Audience/segment: ___ + signal ___ + suppression ___ (per 86.2)
- Promise: ___ (exact words the landing page keeps — human owns per 86.4)

## Offer + channel + assets + cadence
- Offer ___ / Channel ___ / Assets ___ (approved versions linked) / Sends ___

## Primary metric + learning question
- Metric: ___ + baseline ___ + target ___
- Learning question: ___ (e.g. "does the builder case study outperform the feature list?")

## Contract + gates (per 84.3/84.4)
- AI prepares ___ / human approves ___ / must never ___ (budgets, audiences, claims)
- Stop rule: pause if ___ (cost per lead > ___ / complaints > ___)

## Signatures: owner ___ / approver ___ / date ___

Keep it to one page. If the objective needs "and," split the campaign. If the promise isn't on the landing page verbatim, fix the page before launching.

MARKETING-WEEKLY.md (fill each Friday, 20 minutes)

# MARKETING-WEEKLY.md — week of ___

## Result (numbers first)
- Sent ___ / qualified leads ___ / conversions ___ / cost ___ / CPL/CAC ___

## Plausible cause + evidence + uncertainty
- Cause: ___ / Evidence: ___ / Uncertainty: ___ (what else could explain it)

## Decision log
- Continue / adjust / stop: ___ decided by ___ because ___

## One next test
- Change ___ / predict ___ / check date ___ (exactly one, per 89.5's discipline)

Worked mini-example — ByeBuy two-week test: objective 30 workshop signups from the Applications segment; promise "ship one working automation in a weekend" — owned by the founder, mirrored on the page. Week 1: 200 sends → 11 signups, CPL high; cause hypothesis (segment too broad) with uncertainty (holiday week). Decision: narrow to readers who finished Class 84, one test, re-check Friday. Reporting that changes the send list is the whole point — vanity opens never appear on this page.

Rules that keep both documents honest

One objective, one audience, one promise. A brief with three objectives is three campaigns sharing a budget and learning nothing. Split them or pick the one that matters — the learning question decides which.

Stop rules before spend. Write the pause triggers (cost per lead, complaint rate, suppression hits) while calm. Mid-campaign optimism will otherwise rationalize every overrun. The owner who signed the brief enforces the stop — no committee needed.

Cause with uncertainty, always. "Subject line worked" without evidence or alternatives is storytelling. State the plausible cause, cite the evidence (segment split, prior baseline, control group if any), and name what else could explain it. Lesson 86.5's trio — result, cause, uncertainty — is what makes next week's test smarter instead of louder.