Audience theme pack: the bar is that the reader thinks 'that is literally my job title'

---
name: audience-theme-pack
description: Generates a complete content-strategy pack for one narrow audience: an audience research brief, a theme architecture with pillars, a list of 5 to 8 hyper-specific roles, a main topic, purpose bullets that mix goal types, and exactly five example artifacts written to five named archetypes. Enforces a specificity bar, a four-stage specificity ladder, and a structural-difference check so the five examples cannot collapse into the same shape. This skill should be used when a content theme, a segment strategy or a set of seed pieces has to be produced for a defined audience, and the examples will be used as the pattern for later generation.
---

# Audience theme pack

## The claim this skill is built on

A content theme is not a topic. It is a generation seed, and everything produced from it inherits its resolution. That is why the obvious approach fails in a specific and predictable way: asked to build a theme for an audience, the default output describes a category, writes a main topic that would fit any theme in the library, lists outcomes that are all the same kind of outcome, and then produces five examples that are five subjects wearing one shape.

None of that reads as wrong. It reads as fine. That is the problem. A category-level pack produces category-level pieces forever, and the failure shows up months later as engagement that is polite rather than personal.

Two things carry disproportionate weight. The first is the audience line, because a reader decides whether something was made for them in about a second and decides it on the roles, not on the argument. The second is the five example artifacts, because they are the thing a person judges the pack by and the thing downstream generation copies. A tell in one of five examples is a tell in roughly one fifth of everything the pack ever produces, and more than that if it sits in the archetype whose shape is easiest to imitate.

The order below is not decorative. Research precedes architecture because architecture written first turns research into a justification exercise, and you can recognise that failure later by the way every pain point maps suspiciously neatly onto a product feature.

## What the pack contains

Six components, produced in order.

1. **Audience research brief.** Platform-specific pain points, goals, actual vocabulary, demographic and psychographic profile, current blockers.
2. **Theme architecture.** The unique angle, the matching voice, 3 to 5 content pillars, and an honest complexity rating.
3. **Who this is for.** 5 to 8 hyper-specific roles.
4. **Main topic.** One or two sentences carrying the angle.
5. **Purpose bullets.** 5 to 8 outcomes, deliberately mixed across goal types.
6. **Five example artifacts.** One per named archetype, structurally distinct.

## Step 1. Research before architecture

Collect five things, and prefer the audience's own words to your summary of them at every point.

- **Platform-specific pain, not generic pain.** Generic: "struggles to get visibility". Platform-specific: "writes the weekly status update three times because the tool that stores it is not the tool the client reads". The second is a real Tuesday. The first is a category of complaint.
- **Goals as they would state them,** including the ones that are about status rather than outcome.
- **Vocabulary, verbatim.** Which word they use for the document at the centre of their work. Every industry has one word where an outsider picks the wrong synonym, and that one word is what gets you dismissed.
- **Profile.** Seniority, setting, team shape, tenure, what they are measured on and by whom.
- **Blockers.** What stops them acting on advice they already agree with. This is usually approval, budget cycle, or another team's queue, and it is the single most underused input in the whole pack.

If the vocabulary is guessed rather than sourced, stop and read the decision rule below before writing any artifacts.

## Step 2. The specificity bar

The bar is that a reader thinks "that is literally my job title". Not "that sounds like me". Not "I suppose I am in there".

A role line clears the bar when it carries at least two of these four: a function, a setting or company shape, a size or tenure band, and a recurring task or constraint. Job family alone never clears it.

| Category level | Sharpened | What was added |
|---|---|---|
| IT professionals | Disaster-recovery specialists at managed service providers who own a runbook nobody has tested since the last audit | Function, setting, and the specific unowned risk |
| Marketers | Demand-generation leads at 30 to 80 person software firms who carry a pipeline number without a sales team reporting to them | Function, size band, and the accountability that creates the pain |
| Freelancers | Second-year freelance motion designers with one retainer client covering most of their income | Tenure, discipline, and the concentration risk |
| Healthcare professionals | Practice managers at two to four clinician orthopaedic clinics who work the insurance pre-authorisation queue personally | Setting, size, and the task they actually do |
| Finance professionals | Financial controllers at 40 to 120 person services businesses closing the month in a spreadsheet a predecessor built | Title, size band, and the inherited spreadsheet |
| Ecommerce owners | Operations leads at homeware brands running a single third-party logistics contract with no backup warehouse | Category, function, and the single point of failure |
| Teachers | Secondary school science technicians who prepare practicals for four teachers and order the consumables | The role behind the role, plus the weekly task |
| Developers | Platform engineers who are sole on-call for a service they did not build | The rota and the inheritance |

Read the right-hand column and notice what none of them contain: adjectives about ambition. No "growth-minded", no "forward-thinking", no "ambitious". Those describe the writer's hopes about the reader. The sharpened lines describe the reader's week.

**Varying the titles without splitting the audience.** The 5 to 8 roles must be recognisably different people who share one pain. Write the pain out as a single sentence first, then require every role to be someone who would say it. Vary along three axes: seniority, so at least one person who owns the budget and one who does the work; setting, so in-house, agency or consultancy, and solo; and adjacency, so include the person who feeds the workflow and the person who receives its output. If two roles differ only by job title and would have identical days, one of them is padding.

**The over-sharpening edge.** A sharpened role still has to describe a reachable population. The test: name three places you could find these people, such as a searchable title filter, a community or forum, and a recurring event or trade publication. Three places, named, or the role is not an audience.

## Step 3. The specificity ladder

Four stages. Each has one question, and the question is what moves you up a rung.

**Stage 1, broad segment.** "Marketers." Ask: which industry has this problem worst, and what makes that industry different rather than just bigger?

**Stage 2, industry-specific.** "Software marketers." Ask: which role inside that industry personally carries the consequence when this goes wrong?

**Stage 3, hyper-specific.** "Demand-generation leads at 30 to 80 person software firms." Ask: what constraint or situation splits this role into the people who need this and the people who do not?

**Stage 4, ultra-niche.** "Demand-generation leads at that size who carry a pipeline number with no outbound team." Ask: is there still a reachable audience here, or have I described one company?

Climb the ladder across a library over time, not in one jump for one theme. A library sitting entirely at stage 4 has no coverage and no route in. A library sitting entirely at stage 1 has no recognition and no reason to be read. As a working distribution, no more than about a third of a library should sit at stages 1 and 2 combined.

**The stop rule, which is the non-obvious part.** Stop climbing at the rung where you can no longer write five structurally different artifacts for the audience. If every example you can imagine would make the same argument from the same evidence, the audience is no longer a theme, it is one article. Step back one rung. The detail you were about to add is not wasted: put it into the main topic, where it creates recognition without shrinking the reachable population.

## Step 4. Main topic

One or two sentences. Vivid, memorable, and written to trigger recognition rather than to summarise a category.

Apply the swap test. Lift the main topic, drop it onto another theme in the library, and see whether anybody would notice. If it survives the swap, it is describing content in general.

Weak: "Practical insights and strategies to help operations professionals work more effectively." Survives the swap onto any theme ever written.

Stronger: "The work that only exists because two systems disagree, and what it costs the person holding both of them open." Cannot be swapped, because it names a mechanism rather than a benefit.

## Step 5. Purpose bullets and the goal-type mix

5 to 8 outcomes. At least three distinct goal types must appear, and no single type may account for more than three bullets. At least one outcome must be visible inside about 30 days, and at least one must be a six-month positioning move. Active verbs, business consequence, no adverbs.

| Goal type | The question it answers | Weak | Stronger |
|---|---|---|---|
| Positioning | Who do they become in their peers' eyes? | Build thought leadership in the space | Become the person other teams cite when this decision comes up internally |
| Revenue or pricing power | What can they charge, or stop discounting? | Increase revenue | Justify a premium rate by naming a risk the cheaper option leaves with the buyer |
| Competitive differentiation | What can they say that a rival cannot? | Stand out from competitors | Take a public position on the trade-off most of the market avoids discussing |
| Lead generation | What arrives in the inbox? | Generate more leads | Attract inbound enquiries that already know which of the two approaches they want |
| Retention | Why does anyone stay or come back? | Improve engagement | Give existing clients language for defending the work to their own board |

A purpose list that is all positioning is the most common failure and the easiest to miss, because every bullet in it is individually defensible.

## Step 6. The five archetypes

Exactly five artifacts, one per archetype, each tagged with its archetype.

| Archetype | What it is | Its natural shape | Its characteristic failure |
|---|---|---|---|
| **Transparent insight** | An honest mistake, a cost, or a behind-the-scenes reality | Confession, then what it actually cost, then what changed | Vulnerability theatre: a flaw chosen because it flatters |
| **Tactical framework** | A repeatable system or sequence | Named steps, with the reason the order matters | A list of nouns with no mechanism between them |
| **Contrarian opinion** | A challenge to accepted practice, with reasoning | Claim, the strongest version of the counterargument, then why it still loses | Contrarianism about something nobody believes anyway |
| **Story with a lesson** | One episode with a specific time and place | Scene, turn, consequence, and what it changed | The tidy resolution where everything is learned and nothing is left open |
| **Data or results** | Specific outcomes with numbers | The number, its derivation, the part it does not explain | Rounded figures and multipliers, which read as invented because they usually are |

## Step 7. The structural-difference check

This is the enforced part, and it exists because the default failure is five subjects sharing one skeleton. Score every artifact on five axes.

1. **Opening move:** confession, direct claim, question, scene, or number. All five must differ.
2. **Body geometry:** linear sequence, chronological narrative, argument and rebuttal, before and after comparison, or annotated breakdown. All five must differ.
3. **Time direction:** past, present, future, or timeless. At least three distinct values across the five.
4. **Proof type:** personal experience, numbers, a named mechanism, a counterexample, or an observed pattern across others. All five must differ.
5. **Close type:** question to the reader, a rule to keep, a next step, an unresolved tension, or a restated number. All five must differ.

**The threshold: if any two artifacts share three or more axis values, one of them is rewritten, not adjusted.** Rewriting means changing the geometry. Changing the topic while keeping the shape is the cosmetic version of this fix and it fails the same check next time.

Add one more constraint at set level: the longest artifact should be at least twice the length of the shortest. Five pieces of near-identical length is itself a signature of generated sets, visible before a word is read.

## Step 8. The five calibration dials

Set these per artifact, not once for the pack.

- **Tone.** Where this piece sits between dry and warm. Vary it across the five.
- **Structure.** Already fixed by the axis check above.
- **Voice.** First person singular, first person plural, or observational third. At least two values across the set.
- **Depth, matched to audience sophistication.** This is the dial people get wrong. For a stage 3 or stage 4 audience of practitioners, use up to about four terms of art per artifact without defining them. Defining a term the audience uses daily is not clarity, it is a signal that the piece was written for their manager, and it reads as condescension. For a mixed-seniority audience, define on first use inside a clause, never in a sentence of its own.
- **Formatting.** Vary it. At least one artifact should carry no formatting at all: no bullets, no bold, no headers. A set where every piece is bulleted looks machine-assembled at a glance.

## Step 9. Hook, body, close, and the authenticity signals

Every artifact is built as a hook, a body that delivers on precisely what the hook promised, and a close. A hook that promises something the body does not deliver is worse than a flat hook, because the reader who stays is the one who now distrusts the pack.

Six authenticity signals. Every artifact carries at least three, and across the five all six appear at least once.

- **Unrounded numbers.** 43 percent, 11 days, 2,600 records. Not "over 40 percent", not "3x", not "double".
- **First-hand observation.** Something seen, not something known.
- **Appropriate vulnerability.** A cost that was actually paid, in the transparent-insight piece especially.
- **Industry-specific language.** The word the audience uses for the thing, taken from the research.
- **An actual opinion.** A stated preference between two defensible options, with the reason.
- **A real scenario.** A moment with a time, a place and a person in it.

On numbers: any figure ending in 0 or 5 should be treated as suspect unless it is naturally round, such as a team of five or a thirty-day month. Rounded statistics in an example artifact are the most contagious tell in the whole pack, because they are the easiest element for later generation to reproduce.

## The machine-tell catalogue

Scoped to seed artifacts, because a tell here propagates rather than merely embarrasses.

1. The negative-parallel construction: "it is not X, it is Y", and its variants.
2. Perfectly balanced triads, three items of matched length and identical grammatical shape.
3. Rounded statistics and multiplier claims.
4. Corporate abstraction: leverage, streamline, robust, holistic, empower, unlock, seamless, best-in-class, actionable insights, drive impact.
5. Throat-clearing connectives: "here is the thing", "let that sink in", "spoiler", "the result?".
6. Uniform paragraph length across a whole piece.
7. Universal openers: "in today's", "every [role] knows the feeling", "let me tell you about", "picture this".
8. Hedge then assert: qualifying a claim and then making it anyway in the next sentence.
9. Opinions attributed to unnamed groups: "many teams find", "most practitioners agree".
10. The tidy resolution, where the story ends with the lesson stated and nothing left unresolved.
11. Job-title stuffing: repeating the audience's job title back at them in every piece.
12. The long dash used as an all-purpose connector three times in a short piece.

**The set-level scan, which catches what piece-level reading misses.** Read only the first line of all five artifacts in sequence, then only the last line of all five. If the first lines share a grammatical shape, or if all five last lines resolve, the problem is the set rather than any piece, and no amount of line editing fixes it.

## The decision rule

For the audience you are about to build the pack around:

- **If you can name three places to find these people and the role line names a task they do at least weekly, build the full pack at this rung.**
- **If you can name the three places but the role is only a job family with no task or constraint attached, climb one rung** and ask that rung's question before writing anything.
- **If you cannot name three places, you are past the reachable edge.** Step back one rung and move the disqualifying detail into the main topic, where it earns recognition without costing reach.
- **If you cannot tell, meaning the audience is plainly real but you cannot establish how it talks or where it gathers, do not guess.** Build components 1 to 5 with the vocabulary marked as unverified, mark the five artifacts as blocked, and name the two inputs needed: three to five samples of the audience writing about this problem in its own words, and one item from their working day such as a template, a report, a queue or a checklist. Writing five examples on invented vocabulary is the worst available outcome, because they will be wrong and they will be imitated.

## Worked example, compressed

A project management tool wants a theme aimed at operations people.

**Ladder.** Stage 1, "operations people". Which industry has it worst? Professional services, where the product being sold is people's time. Stage 2, "operations at professional services firms". Who carries the consequence? The person who assigns people to projects. Stage 3, "resource managers at 50 to 200 person design and engineering consultancies who reassign people weekly". Stage 4 candidate: add "at firms reporting utilisation to a board monthly". Three places check on the stage 4 version: a title filter, yes; a professional body, yes; a trade publication, yes. Reachable. But the stop rule bites: with board reporting added, all five artifacts would argue the same thing about the same monthly meeting. **Hold at stage 3, move board reporting into the main topic.**

**Roles, 7 of them,** varied by seniority and setting: resource managers, studio managers at design consultancies, engineering delivery leads who own the staffing spreadsheet, operations directors at firms without a dedicated resourcing function, project accountants who reconcile forecast against actual, associate directors who negotiate for people they do not manage, and freelance resourcing consultants brought in during growth. One shared pain, seven different desks.

**Main topic, draft one:** "Helping operations leaders manage resources more effectively." Swap test: survives instantly. Fails. **Draft two:** "Utilisation is reported monthly and decided on Tuesdays, and the gap between those two facts is where the whole job lives." Cannot be swapped.

**Purpose bullets:** 7, covering positioning, competitive differentiation, retention and revenue, with two visible inside a month.

**Artifacts and axis check:**

| # | Archetype | Opening | Geometry | Time | Proof | Close |
|---|---|---|---|---|---|---|
| 1 | Transparent insight | Confession | Chronological | Past | Experience | Unresolved |
| 2 | Tactical framework | Number | Linear | Timeless | Mechanism | Rule |
| 3 | Contrarian | Claim | Argument and rebuttal | Present | Counterexample | Question |
| 4 | Story | Scene | Chronological | Past | Experience | Rule |
| 5 | Data | Number | Linear | Present | Numbers | Restated number |

Artifacts 2 and 5 collide on opening, geometry and close: three shared axes, over the threshold. Artifact 1 and 4 collide on geometry, time and proof: also three. Two rewrites. Number 5 is rebuilt as an annotated breakdown opening on a scene, closing on a next step. Number 1 is rebuilt as a before and after comparison in the present tense, resting on an observed pattern across other firms.

**Tell scan.** First lines of all five read in sequence: three begin with a declarative subject-verb-object claim of similar length. One rewritten to open on the staffing spreadsheet itself instead. Last lines: artifact 3 ended with "let that sink in", cut. Artifact 5 carried "improved throughput by 30 percent", replaced with the underlying figure, 27 percent, sourced from the brief.

**Verdict: pack ships at ladder stage 3 with seven roles across four settings, seven purpose bullets spanning four goal types, and five artifacts that differ on every required axis after two structural rewrites. Two machine tells removed, one rounded statistic replaced, and the board-reporting detail lives in the main topic rather than in the audience list, where it would have cost reach without adding recognition.**

## Failure modes

**The category audience.** The role list says "IT professionals" or "small business owners". From the outside it looks like mild approval: readers say the content is good, nobody says it was written for them, and nobody sends it to a colleague, because there is no colleague it obviously belongs to.

**The interchangeable main topic.** It reads well and it fits every theme in the library. You find out when two themes are put side by side in a planning document and nobody can tell which topic belongs to which.

**The single-goal purpose list.** Eight bullets, all positioning. Every bullet defends itself, and the pack still cannot answer what anyone gets paid for reading this. Typically caught only when someone asks what the commercial outcome is and the honest answer is awareness.

**Archetype collapse.** Five pieces tagged with five archetype names that all open with a claim, run a numbered list, and close with a rule. The tags are labels rather than structures. Looks fine piece by piece and obvious the moment you read the five first lines in a row.

**Tell propagation.** Every piece produced from the pack over the following quarter carries the same construction, and tracing it back lands on example three, which used it twice. This is the expensive failure, because by the time it is visible there are forty pieces carrying it.

**Terminology mismatch.** One word is wrong: the outsider's synonym for the thing the audience touches daily. A practitioner reads it, says "we do not call it that", and discounts everything else in the pack. The rest of the work being correct does not survive it.

**Over-sharpening past the reachable edge.** The pack is precise, elegant and describes about two hundred people, most of whom work at three organisations. It looks like success until someone asks where the audience is and the answer is a list of company names.

**Research as justification.** The angle was chosen first and the research was assembled to support it. The tell from outside is that every pain point maps cleanly onto a product feature, in order, with none left over.

**Depth mismatch.** Terms the audience uses hourly are defined in the artifacts. The audience reads it as written for their manager, and the pack loses the exact readers it was sharpened for.

## What this skill does not do

- It does not validate demand. Nothing in the procedure tests whether this audience has budget, authority or any intention of buying, and a perfectly specific pack aimed at people who cannot purchase is a well-built dead end.
- It does not do the research. It structures and sharpens what you give it, and given nothing it will produce fluent invented vocabulary that is worse than an empty field, because it looks finished.
- It does not produce publishable pieces. The five artifacts are demonstrations of five shapes, sized and shaped to prove the difference check, and they need editing and real specifics before anyone sees them.
- It does not match an existing writer's voice. Where there is a named author with a body of work, that is the voice-match-edit skill, and it should run over these artifacts afterwards rather than instead.
- It has no view on distribution, search demand or channel. A theme can be immaculate and target language nobody searches for, and the keyword and placement work is a separate job.
- It cannot see the library it is joining, so it will not tell you that this is the third theme aimed at the same person under three different job titles, which is the most common way a pack that is individually good makes a library worse.