A 30-day content machine is an operating plan, not a page filled with post ideas. It connects a tested audience problem to evidence, scripts, designs, publishing, responses, and decisions. The system must fit the hours and tools you actually have.

By the end, you will have a 30-day calendar, one finished weekly batch, a content SOP, a live dashboard, and a day-30 retrospective. None of these promise reach or income; together, they make your next test deliberate and reviewable.

// concept

Assemble the Machine From Your Course Artifacts

Do not restart from a blank document. Bring the strongest artifact from every module into one folder. If an item is missing, mark it MISSING; do not let AI invent it.

Module input	What enters the machine	Ready when
1. Audience and niche test	audience segment, problem, available hours, day-30 kill criteria	the segment is narrower than “Pakistanis” and the test fits your week
2. Ideas and swipe file	eight shortlisted ideas, observation log, reusable format patterns	each idea has evidence you can access and a format you can produce
3. Research	source links, first-party observations, redacted question bank, competitor pattern notes	claims trace to a source; patterns are not copied scripts or visuals
4. Scripts	Hook–Body–CTA drafts and language style card	every script has one purpose and sounds natural when read aloud
5. Production	Canva templates, caption set, one-page style guide	mobile readability, rights, language, and brand rules pass
6. Trends	trend log and fast-lane SOP	a trend must pass relevance, freshness, evidence, and sensitivity gates
7. Distribution	four-platform map, nativization checklist, scheduler ritual	assets have platform-specific openings, crops, captions, and links
8. Commercial path	one active path, one building path, written pitch terms	CTA matches the path without promising an outcome

Use one status vocabulary throughout the month:

// prompt — copy me4 lines

BACKLOG → BRIEFED → SOURCED → SCRIPTED → DESIGNED → APPROVED
        → SCHEDULED → PUBLISHED → REVIEWED

HOLD = evidence, consent, rights, capacity, or sensitivity gate failed

A post stays on HOLD if a claim lacks a link, an observation lacks an owner/date, a person lacks permission, a visual lacks usage rights, or a sensitive reference lacks human review.

// concept

Turn Capacity Into a 30-Day Calendar

Calculate a weekly ceiling from time you can protect, not another account’s pace.

// prompt — copy me4 lines

production minutes = weekly available minutes
                   - 30 minutes for review
                   - response buffer
safe publish slots = floor(production minutes / average minutes per finished asset)

With 360 minutes, a 30-minute review, a 60-minute reply buffer, and 45 minutes per asset, the ceiling is six: floor((360 - 30 - 60) / 45). Reduce it after a disrupted week rather than borrowing from the next one.

Build your calendar with these fields:

Field	Required entry
Date and publish window	date, time zone, and manual/scheduled flag
Audience problem	one question, friction, or desired task
Pillar and format	e.g. cost clarity · short video; process · carousel
Evidence ID	source-log row or dated first-party observation
Hook and CTA	exact opening; one action suited to the audience stage
Platform variants	crop, first line, caption, and destination
Owner/status	next owner and workflow status
Review date	dashboard-entry date

Use four weekly cycles plus two closing days. A workable 30-day frame is:

Days	Machine action
1–2	confirm niche-test conclusion, capacity ceiling, pillars, evidence rules, and style guide
3–7	research and publish Batch 1; reserve one slot for community response
8	review Week 1; record one `KEEP`, one `KILL`, and one `TRY`
9–14	produce Batch 2 using only the approved `TRY`; keep a spare slot for a verified trend
15	review Week 2; check whether capacity or language mix needs adjustment
16–21	produce Batch 3; repurpose the strongest useful idea, not automatically the highest-viewed post
22	review Week 3; audit source links, consent records, comments, and unanswered questions
23–28	produce Batch 4; repeat the surviving format and test one new variable
29	close the dashboard; do not rush an extra post to improve the month’s totals
30	run the retrospective and decide continue, revise, pause, or end the niche test

Vary one meaningful element per test—hook, format, language mix, problem, or CTA. Changing all five makes the result impossible to interpret.

// concept

Run a Weekly Batch Through One SOP

Research: choose calendar ideas, attach source IDs, and redact comments or DMs.
Brief: set the problem, promise, format, platform, evidence boundary, language ratio, and CTA.
Script: draft Hook–Body–CTA with citations beside working claims.
Design: duplicate the approved Canva master and test the export on a phone.
Gate: verify facts, rights, consent, pronunciation, sensitivity, and platform fit; failures become HOLD.
Nativize: change crop, opening, caption, and CTA; remove foreign watermarks.
Schedule: prefer native tools. For Canva Content Planner, check the current official plan page; the free fallback is a spreadsheet plus native schedulers.
Respond: reserve a window and log recurring questions without private text.
Review: add comparable metrics and a short interpretation after the preset window.

Use this prompt only after the briefs and evidence IDs exist:

// prompt — copy me16 lines

You are a batch editor, not a researcher.

Audience: [one segment]
Style guide: [paste voice, language, visual and banned-claim rules]
Capacity: [number] publish slots this week
Briefs: [paste numbered briefs with evidence IDs]

For each brief, return:
1. one Hook–Body–CTA script;
2. claim-to-evidence map using only the supplied IDs;
3. Canva template type and content-zone copy;
4. platform-specific first line and CTA;
5. HOLD if a claim, permission, or source is missing.

Never create facts, quotes, testimonials, prices, trend status, permissions,
or performance results. Keep each brief separate.

// concept

Make the Dashboard Drive Decisions

Keep raw values by platform and format. Compare a post only with similar posts from your account.

Dashboard field	Entry or formula
Asset ID	stable ID matching calendar, files, and source log
Published	date, platform, format, language mix
Delivery	reach/impressions or views, using the platform’s current label
Attention	watch time/retention for video; carousel completion where available
Useful action	saves, shares, link actions, or qualified replies—choose before publishing
Useful-action rate	`useful actions ÷ reach × 100`; use only when both fields are available
Evidence note	audience words, question themes, and known measurement limits
Decision	`KEEP`, `KILL`, or `TRY`, plus one sentence of reasoning

At review, ask what stays because it was useful and sustainable, what stops because it failed a gate or cost too much time, and which single variable comes next. “Keep everything and post more” is not a decision.

// worked_example

Worked Example

Their Week 1 brief compares two mobile-data packages without declaring one universally cheapest. Its evidence is dated screenshots from the providers’ official pages; it uses Roman Urdu voiceover, English plan labels, and a question CTA.

They give the batch prompt that brief and receive this draft-one hook:

// prompt — copy me1 line

“This is Pakistan’s cheapest data package, and it will cut your monthly bill.”

The gate fails: “cheapest” depends on usage and current offers, while “will cut” promises a result. Dated screenshots can also expire. The asset moves to HOLD.

The creator changes the source note to “checked on [date]; viewer must verify the current official offer” and changes the hook to:

// prompt — copy me2 lines

“Which package fits a 10 GB monthly routine? Let’s compare the listed price,
validity, and data split—without assuming every user has the same pattern.”

The fixed calendar row is now:

Date	Pillar/format	Evidence	Production	Distribution	Review
Sample Day 4	cost clarity / 45-sec video	two dated official-page captures	Roman Urdu voice; English plan labels; comparison graphic	native 9:16 video; LinkedIn text version removes casual CTA	compare useful questions and saves with prior explainers; no promised target

At review, the creator records actual account data or N/A, never an invented zero: KEEP the three-field comparison; KILL the superlative hook; TRY English subtitles. The decision does not require a commercial result.

// failure_cases

Failure Cases to Diagnose

7 cases to diagnose

Thirty ideas, no production slots
the calendar has a topic every day but no research, design, response, or review blocks. Recalculate the ceiling and convert excess ideas to backlog.
Orphan claims
a script cites “research” but its evidence ID opens no source or dated observation. Move the asset to HOLD until the link is repaired or the claim is removed.
Batch contamination
the AI merges evidence or CTAs between briefs. Give each brief an ID and require a separate claim map for every script.
Scheduler equals strategy
a month is loaded in advance, including sensitive news and unverified trends. Keep trend slots manual and run freshness and appropriateness checks on publication day.
Dashboard vanity pile
views from different platforms are added together while saves, retention, useful replies, and production time are ignored. Compare like with like and connect each metric to its pre-set intent.
No decision changes the queue
weekly reviews produce observations but every planned post survives. Require exactly one KEEP, one KILL, and one TRY, then edit next week’s calendar.
Month-end story replaces evidence
the retrospective claims the niche “worked” from one unusual post. Review all four batches, sustainability, recurring questions, and gate failures before deciding.

// pakistan_angle

Pakistan Angle

Build around the capacity you have during load-shedding, weak mobile data, work, and exams. Keep briefs, scripts, source notes, and compressed design previews available offline; upload heavy exports in a reliable window and keep one buffer week or reduced batch ready. USD tool subscriptions stack quickly in PKR, so pay for one bottleneck only after checking its current official price and payment support. A spreadsheet, local folders, Canva Free reference card, and native schedulers can run this capstone without pretending a paid tool is necessary.

Language and region need explicit calendar fields: Karachi English–Roman-Urdu is not a proxy for Lahore, Khyber Pakhtunkhwa, rural Punjab, or an Urdu-first audience. Test the mix with real audience responses, not chai/rishta clichés. Never paste CNICs, phone numbers, unredacted DMs, customer payment screenshots, JazzCash/easypaisa details, or private addresses into an AI tool. If a post mentions payment rails, plan availability, religion, politics, or a breaking local event, verify it on publication day and obtain the required human approval.

// hands_on

Hands-On Exercise

7 steps

Create one capstone folder with calendar, style-guide, sources, scripts, designs, distribution, responses, and dashboard subfolders.
Complete the machine table using one artifact from every module; label every missing dependency.
Calculate your weekly publish ceiling and populate all 30 days with production, publishing, response, review, buffer, and retrospective blocks.
Write the nine-step SOP in your own tool names and assign an owner, status, and gate to each stage.
Produce one complete week: sourced briefs, final scripts, designed assets, platform variants, captions, and schedule records. Do not publish practice work unless it passes consent and evidence checks.
Add every asset to the dashboard, using N/A where no metric exists. Perform one review and edit the next batch with one KEEP, one KILL, and one TRY.
On day 30—or as a pre-mortem before launch—complete the retrospective: What did we test? What evidence changed our view? What was sustainable? Which gates failed? What repeats, changes, pauses, or ends? Done means another person can trace any calendar row to its source, script, visual, publication record, response notes, metric row, and next decision.

// completion_rubric

Completion Rubric

6 checks — tick as you verify

0/6

// sources

Sources

3 official sources — check every claim yourself

// check_yourself

Check yourself

4 questions · answers and options are taken word-for-word from this course

0/4

1 / 4 · diagnose
Your work shows this failure mode: “Orphan claims.” The lesson describes it like this: “A script cites “research” but its evidence ID opens no source or dated observation.” What does the lesson tell you to do about it?