Chapter template
Every chapter in Prompt & Profit follows the same eight sections in the same order. Once you learn the rhythm, you read every chapter faster — you know where the case study is, where the prompt templates are, where the failure modes get named. The Playbook trades the novelty of structural surprise for the predictability of value, and the trade is on purpose.
The page below is what you can expect to find in any chapter, and what each section contracts to deliver.
What the structure enforces
Three things make a chapter of this Playbook different from a generic “AI for marketers” blog post.
- Structural consistency. Every chapter has the same eight sections in the same order. A reader who has read Chapter 3 knows exactly where to find the prompt templates in Chapter 10.
- Evidence density. Every pain point, every productivity claim, and every case study carries a named, verifiable source. Unsourced assertions do not ship.
- Runnable artefacts. Every prompt is a complete, copy-pasteable template with a worked example. Nothing is described abstractly when it could be shown concretely.
If a chapter is missing one of those three, it does not clear the standard.
The eight sections
| # | Section | Length | What you get |
|---|---|---|---|
| 1 | Opening hook | 2–3 sentences | A bolded statistic from a named source that frames the chapter’s stakes |
| 2 | The job to be done | 150–200 words | The specific marketing job the chapter addresses, in your working language |
| 3 | Pain points LLMs solve | 300–400 words | Four to five pain points, each with a research citation, industry benchmark, or named case |
| 4 | Productivity gains | 200–250 words | Six to eight validated data points quantifying the uplift, woven into argument |
| 5 | A real-world example | 300–400 words | One named company, specific implementation, measurable result, source URL |
| 6 | Detailed prompting techniques | 4–5 templates × 200–300 words | Complete, runnable prompts with worked examples |
| 7 | Pro tips | 4–5 tips × 50–75 words | Field-tested workflow refinements |
| 8 | Common mistakes to avoid | 4–5 mistakes × 50–75 words | Failure modes you are likely to hit, paired with the corrective move |
What each section contracts to deliver
1. Opening hook
Two or three sentences, no more. The first sentence carries a bolded validated statistic and a named source. The second sentence names the stakes for you directly. No “in today’s fast-paced digital landscape”, no empty adjectives, no warming up.
Content teams that deploy LLM-assisted briefs report a 58% reduction in brief-to-draft cycle time (HubSpot, 2024). If you publish four long-form pieces a month, that difference compounds to more than two working weeks per quarter — spent either shipping more, or shipping better.
2. The job to be done
One hundred fifty to two hundred words. Defines the specific marketing job the chapter addresses, in the language you would use to describe it. Names the binding constraint pre-LLM. Signals why the job gets harder as the team or the content volume grows. Addresses you as “you” at least twice.
If the section could describe any marketing job rather than the chapter’s specific one, it has failed. Specificity is the point.
3. Pain points LLMs solve
Three hundred to four hundred words across four or five pain points. Each pain point has a bolded short heading naming the pain in three to seven words, sixty to eighty words of body, and at least one citation — research, benchmark, or named industry source.
| Weak | Strong |
|---|---|
| ”Content creation is slow" | "Brief-to-draft cycle time is the binding constraint on publishing cadence — Forrester 2024 found content teams spend 42% of production time in the brief-and-outline phase alone" |
| "Voice consistency is hard" | "Voice collapses when multiple writers contribute — a Gartner 2023 analysis of 200 B2B blogs found brand voice coherence scores dropped by 31 points once teams exceeded four contributors” |
The strong version names a specific phase, cites a specific source, and quotes a specific figure. The weak version could be about any creative task. The Playbook ships only the strong versions.
4. Productivity gains
Two hundred to two hundred fifty words. Six to eight distinct validated data points quantifying the uplift. Every statistic bolded, every statistic cited (author or organisation, plus year). A mix of time-savings, quality-uplift, and volume-uplift figures — not all of one kind. At least two from Tier 1 sources per the evidence standard.
The section is woven into argument, never bulleted. Bulleted statistics without connective argument feel thin. The argument is what the reader remembers.
5. A real-world example
Three hundred to four hundred words, structured as a story with a turn:
- The situation before — what the team was doing, what was costing them
- The change — what they implemented, specifically
- The result — measured outcome with citation
- The caveat or lesson — what generalises, what does not
Every Section 5 case clears the case study standards and is logged in the subscriber-library register with a working source URL, the date window, and a vendor-origination flag where it applies.
6. Detailed prompting techniques
Four or five templates, each two hundred to three hundred words. Each template has a memorable name, a “what it does” line, a “when to use it” line, the full runnable prompt with {snake_case} variables called out, and a worked example with concrete inputs and described expected output. Every template ships with a runnable YAML file. The schema is documented in the prompt template schema.
The runnable test: copy the prompt into a fresh Claude or GPT-4 session, substitute the worked-example inputs, and run it. If the output does not match the described expected output within reasonable tolerance, the prompt has failed the standard.
7. Pro tips
Four or five tips, each fifty to seventy-five words. A pro tip is a specific workflow adjustment, a measurement heuristic, or a model-choice guide — something you would only learn after running the workflow in production for months. Generic advice (“always review LLM output carefully”) is not a pro tip. Restating the chapter’s main argument is not a pro tip. Vendor recommendations are not pro tips.
8. Common mistakes to avoid
Four or five mistakes, each fifty to seventy-five words. Each one names the mistake, says why it is seductive, and gives the specific corrective action. Where possible, mistakes pair with Section 7 tips without duplicating them. “Be more careful” is not a correction; the corrective move is always specific.
Treating the first output as the draft. The single most common mistake is accepting the LLM’s first generation as the working draft and editing from there. First generations regress to voice-neutral prose; they lock you into their structural choices before you have examined alternatives. Generate three drafts against the same brief, pick the strongest structure from one and the strongest voice from another, and compose from those components.
Word-count bands
The Playbook is honest about chapter length so you can plan your reading time. The body sections sit inside fixed bands; transitional and framing prose carries the rest.
| Section | Min | Target | Max |
|---|---|---|---|
| 1. Opening hook | 40 | 60 | 90 |
| 2. Job to be done | 150 | 175 | 200 |
| 3. Pain points | 300 | 350 | 400 |
| 4. Productivity gains | 200 | 225 | 250 |
| 5. Case study | 300 | 350 | 400 |
| 6. Prompt templates (4–5) | 800 | 1,100 | 1,500 |
| 7. Pro tips (4–5) | 200 | 275 | 375 |
| 8. Common mistakes (4–5) | 200 | 275 | 375 |
| Body total | 2,190 | 2,810 | 3,590 |
| Transitional prose, framing, closing | +3,000 | +5,000 | +5,400 |
| Chapter total | ~5,190 | ~7,810 | ~9,000 |
A chapter does not exceed nine thousand words without explicit reason. A typical chapter takes thirty-five to forty-five minutes to read end-to-end.
What the reader holds the Playbook to
If a chapter skips a section, re-orders the sections, drops the named case study to “a major retailer”, or ships a prompt template without a worked example — the chapter has failed this standard. The errata channel is the place to flag it.
The structure is the contract. The evidence is the trust. The runnable prompts are the work.