A plan without last period's data is a guess. The order is fixed: read the data, replicate the winners, test a small number of new hypotheses, then schedule. Do not change everything at once — a team that rebuilds from scratch every period never learns what works.
Information gathering — always read the old data first
Before writing a single line of plan, you need:
- The prior period's content audit (from
39-content-audit-global). If it does not exist, run that skill first. Do not plan around it. - The prior period's report (from
07-marketing-report-global) if available. - KPI targets for the new period — what changes versus last period.
- Known variables in the new period: upcoming campaign, launch, event, seasonality; changes in budget, headcount, or capacity.
If the user cannot supply at least an audit or a post list with metrics, stop and recommend running 39-content-audit-global first.
Principles
- Data first, plan second. Every change in the plan must trace back to a specific line of prior-period data.
- Replicate winners before chasing new ideas. A winner has already proved itself. Reproduce it with small variations before betting on an untested concept.
- The 70/20/10 split:
- 70 percent of output is proven format plus proven angle — the winner's formula, held constant.
- 20 percent is optimization of a winner — change the hook, the channel, or the length while keeping what worked.
- 10 percent is genuinely new — the test hypotheses.
- Maximum 2 new hypotheses per period. More than that and no single test accumulates enough data to conclude anything.
- Losers are not automatically deleted. Separate a loser caused by a wrong angle (retire it) from a loser caused by weak execution (fix the execution, try once more).
- Plan at the week level here. Day-level detail belongs to
01-content-calendar-global. - Account for seasonality in paid amplification. Q4 media costs run well above Q1 globally; if a proven post type is normally boosted, the same plan costs more in November than in January (see
references/benchmarks-global.md).
Workflow
1. Review the prior period
From the audit (39-content-audit-global) and report (07-marketing-report-global), fill three columns:
| Worked (KEEP) | Did not work (STOP) | Not yet tried (TEST) |
|---|---|---|
| [format / angle / channel + the numbers that prove it] | [+ the data reason] | [idea from a pattern or a new insight] |
2. Set the direction for the new period
| Period | Primary objective | Priority KPI | Attached campaign | Major change |
|---|---|---|---|---|
Exactly one primary objective — reach, lead generation, conversion, or retention. The priority KPI is the metric this period will actually be judged on.
3. Build the content mix at 70/20/10
| Group | Share | Content |
|---|---|---|
| Proven (70%) | ~70% of posts | Winner formula held constant — only the subject changes |
| Optimized (20%) | ~20% of posts | Variations on winners: different hook, channel, or length |
| Experimental (10%) | ~10% of posts | New hypotheses (maximum 2) |
Then adjust the funnel and format ratios, with a data reason for each change:
| Dimension | Prior period ratio | New period ratio | Reason (from data) |
|---|---|---|---|
| TOFU / MOFU / BOFU | |||
| Video vs static vs text | |||
| Channel split |
4. Plan the winner replication
| Original winner | Result (metric) | Replicate into | What changes | Channel / timing |
|---|---|---|---|---|
Each winner becomes 2-3 variations. Hold the hook structure and format constant; change the subject, the angle of the shot, or the person on camera.
5. Define the new hypotheses (maximum 2)
| Hypothesis | Format | Channel | Metric | Test duration | Decision rule |
|---|---|---|---|---|---|
Write each one so it can be falsified: "If we [change X], then [metric Y] will [rise or fall by Z percent]." For test design and sample sizing, hand off to 19-ab-test-setup-global.
6. Build the overview calendar and assign ownership
| Week | Theme | Priority channel | Posts | Highlights (replicated winner / hypothesis) |
|---|---|---|---|---|
| Week 1 | ||||
| Week 2 | ||||
| Week 3 | ||||
| Week 4 |
| Owner | Responsibility | Posts per week | Primary channel |
|---|---|---|---|
Day-level detail (publish time, hook, CTA, status) goes to 01-content-calendar-global. Per-piece briefs go to 36-content-brief-global.
Output structure
File name: next-content-plan-[brand]-[YYYYMMDD].md
# Next Content Plan — [Period] — [Brand]
## I. Prior period review
| Worked (keep) | Did not work (stop) | Not yet tried (test) |
|---------------|---------------------|----------------------|
| | | |
## II. Direction for the new period
| Period | Primary objective | Priority KPI | Attached campaign | Major change |
|--------|-------------------|--------------|-------------------|--------------|
| | | | | |
## III. Content mix — 70/20/10
| Group | Share | Posts | Content |
|-------|-------|-------|---------|
| Proven (70%) | | | |
| Optimized (20%) | | | |
| Experimental (10%) | | | |
| Dimension | Prior ratio | New ratio | Reason (from data) |
|-----------|-------------|-----------|--------------------|
| TOFU / MOFU / BOFU | | | |
| Video vs static vs text | | | |
| Channel split | | | |
## IV. Winner replication
| Original winner | Result | Replicate into | What changes | Channel / timing |
|-----------------|--------|----------------|--------------|------------------|
| | | | | |
## V. New hypotheses (max 2)
| Hypothesis | Format | Channel | Metric | Test duration | Decision rule |
|------------|--------|---------|--------|---------------|---------------|
| | | | | | |
## VI. Overview calendar
| Week | Theme | Priority channel | Posts | Highlights |
|------|-------|------------------|-------|------------|
| | | | | |
## VII. Ownership
| Owner | Responsibility | Posts per week | Primary channel |
|-------|----------------|----------------|-----------------|
| | | | |
Related skills
39-content-audit-global: required input — run it first. No audit means no plan.07-marketing-report-global: the prior period's report adds the full-funnel KPI view.01-content-calendar-global: takes this plan and produces the day-by-day schedule.36-content-brief-global: briefs each individual piece inside the plan.19-ab-test-setup-global: designs the tests for the 2 hypotheses.13-data-analysis-global: supporting analysis when the data needs a deeper look.
Quality checklist
- Prior period data present as the basis (audit from 39 or report from 07) — nothing planned on instinct.
- Every change in the plan traces back to a specific line of data.
- Content mix follows 70/20/10 and the shares total 100 percent.
- Each winner has a concrete replication plan (2-3 variations each).
- No more than 2 new hypotheses, each falsifiable and each with a decision rule.
- Losers separated into wrong angle (retire) versus weak execution (fix and retry once).
- Funnel and format ratio changes each carry a data reason — nothing changed arbitrarily.
- Overview calendar covers all 4 weeks with named owners.
- Seasonality accounted for where the plan depends on paid amplification.
- Handoffs stated:
01-content-calendar-globalfor the detailed schedule,19-ab-test-setup-globalfor the hypotheses.