The Complete Claude Marketing Guide 2026

Five files. Every chapter, prompt, template, workflow and audit reads them first.

Two hours once. It is the hard part and it is the part that decides output quality across everything else.

Why these come first

Skip them and you get a hundred components each inventing their own version of your positioning.

The failure mode in marketing output is not a wrong answer. It is a plausible one in the wrong voice, which is harder to catch and worse to publish.

marketing/
├── strategy-os.md      who we are, who this is for
├── content-os.md       how we sound, and the three pillars
├── distribution-os.md  the channels, and the ceilings
├── campaign-os.md      one job, one lever, kill criteria
└── reporting-os.md     the chain, and what is banned

marketing/proof/case-studies.md

strategy-os.md

# Strategy OS

## What we sell
One line, in the words a customer would use:
What a meeting with us is actually for:

## Who we sell to
Exact job title:
Company type and size band:
What that person is trying to do:
What they have already tried that failed:
Who this is NOT for:

## Derived ICP, from closed-won
Three attributes correlating with fast closes:
Three correlating with long or lost deals:
Job titles that actually signed, not the ones we target:
Size band with the highest win rate:
The searchable filter string:

## Positioning
The one sentence everything says:
Two beliefs we hold that our market mostly disagrees with:
Who this positioning deliberately repels:

## Competitors
Named, with the honest difference for each:
Where they are genuinely better:

## Banned
<never_use_these>
Claims we cannot evidence:
Numbers nobody approved:
Phrases we would never say:
</never_use_these>

The Banned block is the most useful section in the file. Without it, every component independently decides whether "trusted by hundreds of teams" is defensible. With it, the decision is made once.

content-os.md

# Content OS

## Voice
Three adjectives:
Two phrases we would never say:
Sentence rhythm: short and clipped, or long and winding:
Do we swear, joke, use rhetorical questions:
What genuinely annoys us:

## Three pillars
Each must be a repeatable topic, not a theme.
"Cold email" is a topic. "Growth" is a theme and produces nothing.

Pillar 1:
  Why our ICP cares, tied to what they tried and failed:
  Five angles inside it:
  Which proof points support it:
Pillar 2:
Pillar 3:

## Hard rules
- No em dashes
- Hook is lines 1 to 2 and must earn the See More click alone
- One idea per line, three lines maximum per paragraph
- No hashtags in the body
- Contents listed as objects, not outcomes
- One CTA per piece
- Proof only from proof/case-studies.md

## Reference pieces
Three of ours that performed, pasted IN FULL, with why:

The reference pieces do more than any rule list. A component reading three things that actually worked matches the voice better than one reading a description of the voice.

distribution-os.md

# Distribution OS

## Channels, ranked by what they actually produce
Channel | What it produces | Volume ceiling | Owner

## Organic social
Posting cadence. Never more than one post a day. Same-day
multiples suppress each other.
Engagement session: how many comments, how often.

## Outbound send budget
ACCOUNT-WIDE, not per campaign:
  Connection requests: 20/day or 100/week
  Direct messages: ~100/week free, ~150 premium
  Profile views: 80/day free, 150 premium
  Comments: 60/day

If more than one campaign is running, state the split here.
Three campaigns at 20 each means 60 attempts a day and a
restricted account.

## Email
Sending domain and warm-up state:
Deliverability thresholds: bounce under 2%, complaints under
0.10%, SPF, DKIM, DMARC configured.
Frequency ceiling per segment:

## Repurposing routes
One long asset becomes:
  Social posts, how many:
  Email, which sequence:
  Blog, which format:

The send budget section is where people go wrong. Recording it here means every component that proposes a volume checks the real ceiling first.

campaign-os.md and reporting-os.md

# Campaign OS

## Every campaign has one job
One action. If a campaign has two goals, it has none.

## Naming convention
[quarter]-[channel]-[offer]-[audience]
Used in the folder, the UTM, and the CRM lead source.
Never set lead source to the channel. Always the campaign.

## The five angle levers
Pick one per campaign. Never mix.
1 Effort removed      what they no longer have to do
2 Time to result      how fast the first useful outcome arrives
3 Outcome specificity the exact narrow thing they end up able to do
4 Risk removed        what they no longer have to commit
5 Identity            who they become, or stop being

State which lever, and who this angle repels. An angle that
appeals to everyone converts nobody.

## Required before launch
[ ] One job, named
[ ] One lever, named
[ ] Proof pulled, or none used
[ ] UTMs built
[ ] Send budget checked
[ ] Kill criteria set, WITH A DATE

# Reporting OS

## The diagnostic chain
Work in order. Stop at the first failure. Never change more than
one thing per week.

1 Awareness low        reach and engagement rate
2 Awareness fine,
  interest low         the hook, or the audience is wrong
3 Interest fine,
  intent low           the offer, not the content
4 Intent fine,
  action low           the page, or the ask
5 All fine,
  revenue low          price or fit, not marketing

## Banned in reports
- Impressions as a headline number
- Percentages without the absolute figure beside them
- Any metric with no action attached

Kill criteria set in advance is the discipline most teams skip. Deciding what failure looks like after seeing the numbers is not a decision.

proof/case-studies.md

One file. Every component in this guide pulls proof from it and nowhere else.

client | permission (YES / ANONYMISED / NO) | sector | size |
the problem | what we did | the result with a real number |
timeframe | service line | their quote if we have one

The result column needs a number you can defend. "Improved their rankings" is a sentence, not a result. Leave the row out.

The permission column is not optional. YES, ANONYMISED and NO are the only values. Components read it to decide whether to name or describe.

Twelve solid rows beat forty thin ones, because components pick the closest match and a thin match produces weak output.

The build order

1. The five OS files. Two hours. Everything else reads them.
2. The proof library. Anything that makes a claim needs it, and it will otherwise produce nothing or invent something.
3. The prompts. No install needed, and they tell you whether the OS files are good enough. Generic output means fix content-os.md, not the prompt.
4. The templates. Fixed formats for the recurring documents.
5. The chapters, in whatever order the work demands.
6. The audits, once there is something to audit.
7. The workflows last. Automate what already works manually, or you automate a broken process and it breaks faster.

Four chapters. Strategy, research, ICP, and the brand layer.

These decide everything downstream. A content calendar built on the wrong ICP is a well-run mistake.

01. Marketing Strategy for 2026

One thing changed that reshapes the rest.

Production became free. Distribution did not.

Anyone can produce a month of content in an afternoon. So the differentiator moved from whether you can make things to whether anyone sees them, and from volume to whether the thing is worth seeing.

Three consequences the whole guide is built around:

1. Distinctiveness beats polish. Everything is well-produced now. What is scarce is a position someone would argue with.

2. Proof beats claims. Claims are free to generate, which is exactly why nobody believes them. A number with a timeframe and a source does work that a paragraph of copy cannot.

3. Distribution is the constraint, not ideas. The ideas file is not the bottleneck. The 20 connection requests a day, the one post a day, the send budget: those are.

---
name: strategy-setter
description: Set the quarterly marketing strategy from evidence.
  Run quarterly, highest effort.
---
Read strategy-os.md and the last quarter's reporting.

OUTPUT one page:
- The one thing we are trying to change this quarter, stated as a
  number moving from X to Y
- The three channels that will do it, ranked, with what each
  produced last quarter
- The one channel we are stopping, and what it cost us
- The bet: what we believe that our market mostly does not
- What would have to be true in 90 days for this to have worked

FAIL CLOSED if last quarter's numbers are not available. Say so.
A strategy set without them is a preference.

Never list more than three channels. A four-channel quarter is a
quarter with no owner on any of them.

02. Market and Customer Research

---
name: voice-of-customer
description: Capture how buyers describe the problem in their own
  words. Run before writing any copy.
---
SOURCES, in priority order:
1. Their own posts and comments
2. Comments under competitor posts
3. Relevant forums and communities
4. Review sites for competing products
5. Your own call transcripts and support tickets

OUTPUT to marketing/research/voice-of-customer.md:
- Twelve verbatim phrases, quoted exactly
- Five words they use that we currently do not
- Three objections they raise unprompted
- What they say they already tried
- What they say did not work about it

NEVER paraphrase into marketing language. If they said "I can't
tell what any of it is doing", record that, not "attribution
challenges".

Source 5 is the richest and the least used. Support tickets are
the problem stated without any politeness filter.

---
name: competitor-brief
description: Build the honest competitive picture. Quarterly.
---
PER NAMED COMPETITOR:
- What they actually sell, in their words
- Who they position against
- Where they are genuinely better than us
- The trap question that surfaces the difference without naming
  them
- What never to say about them

The "where they are better" line is mandatory. A brief where we
win every row reads as marketing and gets discounted entirely,
including the rows that are true.

Never characterise their team, their pricing motive or their
roadmap.

Twelve real phrases beat any persona document. Copy written from them sounds like the market; copy written from a persona sounds like a workshop.

03. ICPs, Segmentation and Positioning

---
name: icp-from-closed-won
description: Derive the ICP from deals rather than a workshop. Run
  quarterly, highest effort.
---
Pull closed-won, last 12 months. Per deal: company size, industry,
buyer job title, deal size, days to close, the trigger.

OUTPUT to strategy-os.md:
1. Three attributes correlating with fast closes
2. Three correlating with long or lost deals
3. Exact job titles that actually signed, verbatim
4. Size band with the highest win rate
5. The searchable filter string
6. The gap between the stated ICP and the actual one
7. Who this is explicitly NOT for

FAIL CLOSED with fewer than 10 deals. Report the count and stop.

Point 6 is the output worth showing a founder. A real buyer who is
a different title from the one on the website changes the whole
plan.

---
name: positioning-line
description: Produce the one sentence everything else says.
---
OUTPUT:
- One sentence a stranger would understand
- Who this deliberately repels
- Two beliefs we hold that the market mostly disagrees with
- The named competitors, with the honest difference for each

A positioning line that repels nobody converts nobody. If the
"repels" field is empty, the line is too broad to build on.

The two contrarian beliefs are what the content pillars grow from.
Without them, content returns other people's arguments.

Segmentation rule. Three segments maximum. A fourth means one is not working and nobody has admitted it. Each segment needs a different problem, not a different size band.

04. Brand, Messaging and Offers

---
name: messaging-hierarchy
description: Build the message stack, from the one line down to the
  proof.
---
FOUR LEVELS:
1  The positioning line. One sentence.
2  Three message pillars. Each answers a different objection.
3  Per pillar, three proof points from case-studies.md.
4  Per proof point, the one-line version and the full version.

RULES
- Every level must survive being read alone. Someone will only
  ever see level 3.
- No claim without a row in the proof library.
- Check banned.md before writing anything.

OUTPUT the stack, and the one message you would keep if you could
only keep one.

---
name: offer-architect
description: Define an offer specific enough to sell without a call.
---
OUTPUT:

THE OFFER
- One sentence a stranger understands
- What they receive, as OBJECTS. Files, dashboards, sessions.
- What happens in week 1, week 2, week 4
- What is explicitly not included
- The price, and what it does not cover

THE WEDGE
- The free thing that proves competence before they pay
- Why it is useful even if they never buy

THE DISQUALIFIER
- Who this is not for, stated openly

RULES
- No "bespoke" or "tailored". Those words mean the offer is not
  defined.
- The price appears. An offer with no number is a conversation.
- Objects, not outcomes. "43 prompts, 6 sequences, 1 tracker" not
  "unlock your potential".

The disqualifier is the section people cut and the one that lifts
conversion most. A prospect who reads "not for you if" and keeps
reading has qualified themselves.

The order these four run in

Research before ICP. ICP before positioning. Positioning before offer.

Running them in any other order produces a positioning line derived from what you wish were true rather than from what closed.

Re-run chapter 3 quarterly against the last quarter's closed-won. If the winning job title has moved, chapters 5 through 10 need rebuilding, and finding that out in month six is cheaper than month twelve.

Three chapters. What gets made, whether it gets found, and whether it reaches anyone.

Content production is the easy part now. Chapters 6 and 7 are where the work is.

05. Content Strategy and Production

One long asset a month becomes everything else. Never write channel content from a blank page.

---
name: content-pillars
description: Define three repeatable topics. Not themes.
---
"Cold email" is a topic. "Growth" is a theme and produces nothing.

PER PILLAR:
- The topic in one line
- Why the ICP cares, tied to what they tried and failed
- Five angles inside it
- Which proof points support it

THREE MAXIMUM. A fourth means one is not working.

THE TEST: if nobody internally can talk about a pillar for twenty
minutes without notes, it is not yours. Replace it.

---
name: repurposer
description: Turn one long asset into a month of channel content.
---
EXTRACT 10 to 12 standalone ideas. One idea per line. No idea that
needs the others to make sense.

THEN PER IDEA:
- One social post, hook in lines 1 to 2
- Or one email, if the idea is better read than scrolled

PLUS:
- One carousel from the strongest section
- 4 to 6 comment-ready opinions for engagement sessions
- 1 email sequence of three, if the asset supports it

STATUS on everything: DRAFT. Never scheduled.

FAIL CLOSED if the asset yields fewer than 8 standalone ideas. It
is not substantial enough to repurpose, and forcing it produces
filler.

---
name: humanizer
description: The final pass on everything before it publishes.
---
WRITE THE BODY FIRST, THEN THE HOOK. A hook written first
constrains the body to fit it.

THEN STRIP:
- Numbers not all round. 85 to 11, not 85 to 9.
- No perfect parallelism in lists. One element differs.
- One genuine break in the flow of thought.
- Case studies carry one vague detail beside one hyper-specific
  one. Real memory is uneven.
- Not every piece resolves. One a week ends open.
- No em dashes anywhere.

FINAL TEST: could a competent person have written this in five
minutes? If it reads like it took a committee, that is what it
reads as.

Cadence. Three pieces a week, one a day maximum. Same-day multiples suppress each other. A missed day beats a thin piece.

06. SEO, AEO and AI Search Visibility

One thing to be straight about before anything else. Claude cannot see live rankings, volumes, backlinks or Core Web Vitals. Those come from a connected data source or a pasted export. Work from real data or say what you need.

---
name: technical-audit
description: The prioritised technical fix list. Run first, always.
---
CHECK, from connected data:
- Crawlability and index coverage
- Robots and sitemap against the latest crawl
- Core Web Vitals, and whether layout shift is the cause
- Redirect chains and broken internal links
- Canonical conflicts
- Schema present and valid

OUTPUT a list of TEN, prioritised by impact. Per item: the issue,
the evidence, the fix, the effort, and whether it needs a
developer.

Not eighty items. Eighty is an audit nobody actions.

ORDER OF WORK, always: technical first. Content work on a site
that cannot be crawled is wasted effort, and it is more fun to do,
which is why people do it first.

---
name: answer-readiness
description: Make a page quotable by an AI answer engine, not just
  rankable.
---
CHECK, per page:
- Do the opening 40 words directly answer the query
- Passage-level quotability: short standalone claims that can be
  lifted without the surrounding paragraph
- Entity clarity: is it obvious who and what this is about
- Source attribution on every figure
- FAQ coverage for the questions actually asked

OUTPUT the rewrite, passage by passage.

ONE HONEST CAVEAT. Observational research across roughly 2 million
citations found prompt-content alignment to be the dominant
page-level predictor, while the standard checklist of FAQ blocks,
structured data and Core Web Vitals collapsed toward zero once
domain effects were applied. That study is observational and
cannot establish causation. Treat schema as hygiene rather than a
lever.

The content verdict skill. For every existing page, return exactly one of KEEP, REFRESH, CONSOLIDATE or KILL, with evidence. A verdict with no evidence is a guess. A quarter's refresh list is five to ten pages, prioritised, not the whole site.

07. Social Media and LinkedIn

The chapter with the hardest constraint, and the one most plans ignore.

Comments are free. Sends are not.

RECOMMENDED CEILINGS, ACCOUNT-WIDE not per campaign
  Connection requests   20/day, 100/week
  Direct messages       ~100/week free, ~150 premium
  Profile views         80/day free, 150 premium
  Comments              60/day

Three campaigns at 20 each means 60 attempts a day and a
restricted account.

---
name: engagement-session
description: The daily 25 minutes that makes posting work.
---
Target: 8 to 10 likes, 5 comments.

SIX RULES
1. At most 2 of 5 comments mention our own work.
2. No two comments share a structure.
3. At least 1 of 5 outside our niche. Only-on-niche scores worse.
4. At most 1 promotional phrase across the session.
5. If you simply agree, like it. Do not comment.
6. At least 1 comment on a post with 200+ reactions.

Score each comment 1 to 10 before posting. Below 7 gets
rewritten, not posted. Target an 8.0 average.

RULE 6 is the highest return per minute available. A good comment
on a large post carries a 7 to 12x reach multiplier against a post
on a small account, and costs ninety seconds.

---
name: lead-magnet-post
description: The post that turns reach into a contactable list.
---
Never write before you have all five:
1. The resource, and what is literally inside it
2. The ICP, exact job title
3. Comment keywords, one per variant, same semantic family
4. Proof, a real number
5. The offer, one line, for the PS only

No proof is a valid answer. Write the curiosity version rather
than inventing one.

THREE VARIANTS, each a different formula, each its own keyword.
OUTBOUND, PIPELINE, SEQUENCE. Never numbered.

CTA BY CONNECTION STATE:
  Mostly connected:  Comment "[WORD]" and I'll DM it over.
  Mostly not:        Comment "[WORD]" and connect with me so I can
                     send it.
                     PS. If we're already connected, just comment.

PLAN THE DELIVERY BEFORE POSTING. A post pulling 500 comments is
500 people expecting a link against a 20 a day ceiling, which is
five weeks.

The delivery half is the part that cannot run by hand. Prosp imports commenters and likers straight from any post URL, including future reactions as they arrive, then runs the five-node sequence: import, a lead is 1st level condition, the message on the YES branch, a connection request then message on the NO branch, and a Reply Comment node so every public comment gets a public reply.

Connection requests auto-detect acceptance over two weeks, so no wait node is needed after one.

Where these three meet

The long asset from chapter 5 is the page chapter 6 optimises and the source chapter 7 posts from. One piece of work, three channels, which is the only way the cadence is survivable.

Run chapter 7's engagement session for two weeks before chapter 5 publishes anything. A post published into a feed where nobody recognises the name reaches only the followers you are trying to grow.

Four chapters. What happens after the click, what you pay for clicks, how you measure it, and what changes when you run it for someone else.

08. Email, Lifecycle and CRM

Every sequence needs three things before a word is written.

1  The trigger event that enrols someone
2  The EXIT event that removes them
3  The one action it exists to cause

No exit event, no sequence. One that keeps mailing people who
already converted is worse than none.

Read the real event names from the platform. Never guess one.

---
name: lifecycle-mapper
description: Find the missing flows before building new ones.
---
THE STANDARD SET. Most teams are missing two.
- Welcome, on signup
- Abandoned cart or abandoned demo
- Browse or content abandonment
- Post-purchase, with the second-action nudge
- Winback, at the category's repeat interval
- Re-engagement, before they go fully cold
- Referral request, at the moment of a result

OUTPUT which exist, which are missing, the timing of each, and
which missing one is worth most.

THE GAP IS THE FINDING. Build the missing flow before optimising
an existing one.

---
name: email-sequencer
description: Build a sequence with exit conditions.
---
PER EMAIL: the delay, the subject under 45 characters, the body,
the single CTA.

RULES
- Email 1 delivers what was promised. Link in the first two lines,
  not after a paragraph of thanks.
- Each subsequent email adds something the previous did not.
- Five emails maximum. After that, move them to the general list.
- Handle the empty first-name case. Nobody receives "Hi ,".
- Respect the frequency ceiling in distribution-os.
- No em dashes.

DELIVERABILITY, non-negotiable:
  Bounce under 2%
  Spam complaints under 0.10%
  SPF, DKIM, DMARC configured

These are enforcement lines, not targets. A sequence that lands in
spam is a sequence that does not exist.

The segment nobody works: people who downloaded something and never bought. Largest segment in most businesses and the warmest list available, because every one of them raised a hand. Build for that before any new acquisition work.

09. Paid Ads, CRO and Creative Testing

---
name: ad-copy
description: Write paid copy for a locked angle.
---
ASK which of the five angle levers this campaign is on. Do not
proceed without one.

PRODUCE per platform: five headline variants character-counted,
three primary text variants, one description, the CTA type.

RULES
- Every claim traces to the proof library. No proof, no claim.
- Check banned.md before writing anything.
- Variants differ in ANGLE, not wording. A synonym swap is not a
  variant.
- No superlatives without a number.

Name which variant to run first, with one reason.

---
name: landing-page-auditor
description: Audit a page against its single conversion goal. Run
  before sending paid traffic anywhere.
---
ASK what the ONE job of this page is. If the answer contains two
actions, stop and force a choice.

AUDIT:
- Does the headline match the ad that sent them here
- Is the form or CTA above the fold on mobile at 375px
- Is what is inside described as objects, or as outcomes
- Is there exactly one CTA action
- How many exits: nav, footer, other offers
- Is the proof real and attributed
- Does it load without layout shift

PASS, WEAK or FAIL each, with one line of evidence and one line on
what it costs.

END with the single change that would move conversion most. One
change, not a list. A list gets filed; one change gets done.

Two testing rules that save more money than any creative.

Do not touch anything in week one. Changing inside the learning phase restarts it and teaches you nothing.

Refresh creative when frequency climbs and CTR falls together. Either one alone is noise. Both together is fatigue and it is actionable.

10. Analytics, Automation and Growth

---
name: metric-narrator
description: Turn a number into a sentence someone can act on.
---
PER METRIC, four things:
1  THE NUMBER, with its absolute beside any percentage. "Up 40%"
   from 5 to 7 is a different sentence from 500 to 700.
2  AGAINST WHAT. Last month, last year, or the target. A number
   with no comparison is decoration.
3  WHY, if we know. If we do not, say we do not. A guessed cause
   presented as fact is the fastest way to lose trust in the whole
   report.
4  SO WHAT. The decision it should inform.

If a metric cannot answer point 4, cut it. Most reports are three
useful numbers and twenty decorative ones.

---
name: attribution-hygiene
description: Make the numbers answerable three months later.
---
LEAD SOURCE NAMES THE CAMPAIGN, never the channel.
"LinkedIn" tells you nothing in three months.
"q3-saas-founders-job-post-trigger" tells you which signal type
produced the revenue.

SET, per lead:
  lead_source    the campaign name, in convention
  signal_type    which trigger brought them in
  first_touch_date
  segment

THE SIGNAL_TYPE FIELD is what lets you answer which trigger
produced closed deals. Without it that question is unanswerable
and you keep running things that do not convert.

The growth loop that compounds. Every published piece is a list. Post engagers import as contacts, contacts become conversations, conversations produce the proof that makes the next piece land harder. The loop only closes if the import happens, which is why chapter 7's delivery step is not optional.

11. Agencies

What changes when the account is not yours.

PER CLIENT, ITS OWN FILES
  strategy-os.md   from THEIR closed-won, never yours
  content-os.md    three of THEIR pieces, pasted in full
  proof.md         their results, with a permission column

NEVER SHARE A CONTENT OS BETWEEN CLIENTS. Two accounts that sound
like the same person is the fastest way to lose both.

PER ACCOUNT, ITS OWN CEILING
  New account starts at 10 a day, not 20
  Build to 20 over the first fortnight
  Never aggregate. Eight clients is eight budgets.

DEDUPLICATE ACROSS CLIENTS. Two clients in adjacent niches will
share prospects.

RUN THE HUMANISER ON EVERYTHING before the client sees it. A
client who spots AI phrasing in a draft stops trusting the whole
deliverable.

The margin question nobody runs. Calculate gross margin per client monthly: fee received, delivery hours by discipline, loaded cost, account management hours, QA time, attributable tool cost.

Rank clients by margin percent, worst first. That order is almost never the same as the order by fee size, and agencies manage attention by fee size.

The order to run the chapters in, per client. Chapter 3 first, from their data. Then chapter 7's engagement for two weeks, nothing else. Then chapter 5. Then the rest as the engagement demands.

Where the sending happens

Chapters 7 and 11 produce specs, lists and copy. None of them sends anything, and that is deliberate for a client account.

Prosp is the send layer: cloud-based so nothing depends on a machine staying open, one account per client profile each created in its own browser session, all managed from one workspace. White-label with your own domain plus a sub-workspace per client.

The unified inbox spans every connected account with filters for campaign, replied and unread, which makes the daily reply pass one task rather than eight logins. Native HubSpot integration, 12 webhook events and a REST API supply the figures chapter 10 needs.

Ten categories, five each. No install needed. Paste and run.

Start here. If these produce generic output, the problem is content-os.md, not the prompt.

Strategy and research, 5

01  Read my last quarter's numbers and tell me the one thing I
    should try to change this quarter, stated as a number moving
    from X to Y. Then the three channels that will do it and the
    one I should stop.

02  Find the twelve phrases my buyers actually use to describe
    this problem. Quote them exactly. Never paraphrase into
    marketing language.

03  Read this competitor's last ten pieces. Where are they
    genuinely better than us? Answer honestly, it matters more
    than the rows we win.

04  Read these support tickets. What is the problem, stated
    without any politeness filter?

05  What does my market mostly believe that I think is wrong? Give
    me two, and tell me what someone would say against each.

ICP and positioning, 5

06  Pull my closed-won from the last 12 months. Give me the three
    attributes that correlate with a fast close, with the
    evidence. Fewer than 10 deals, say so and stop.

07  List the exact job titles that signed, verbatim. Then the
    titles I have been targeting. Show me the gap.

08  Write my positioning line in one sentence, then tell me who it
    deliberately repels. If nobody, it is too broad.

09  Who is this explicitly not for? Six disqualifiers, each with
    one line on why it fails.

10  Split my customers into three segments by the problem they
    have, not the size band they sit in.

Offers and messaging, 5

11  Write the offer: one sentence, what they receive as objects,
    week 1 week 2 week 4, what is not included, the price. No
    "bespoke", no "tailored".

12  Build the message hierarchy: the one line, three pillars each
    answering a different objection, three proof points per
    pillar.

13  List what is literally inside this resource. Objects, counts,
    sections. If I cannot list objects, tell me it is not
    finished.

14  Which of the five angle levers is this campaign on? Effort
    removed, time to result, outcome specificity, risk removed,
    identity. Pick one and tell me who it repels.

15  Read my proof library and tell me which claim I am making that
    I cannot actually evidence.

Content, 5

16  Define three content pillars. Each a repeatable topic, not a
    theme. If I cannot talk about one for twenty minutes without
    notes, tell me it is not mine.

17  Turn this long asset into 10 to 12 standalone ideas. One per
    line. No idea that needs the others to make sense.

18  Write five hooks for this piece. Then delete the body and tell
    me which hooks still feel complete. Those are summaries.

19  Strip the AI tells from this draft: round numbers, perfect
    parallelism, no break in the flow, everything resolving, em
    dashes.

20  Read this call transcript. Three claims my market agrees with,
    two it disagrees with, one I would defend under pushback. The
    last one is the piece.

SEO and AI search, 5

21  From this crawl export, give me ten technical fixes ranked by
    impact. Per item: issue, evidence, fix, effort, and whether it
    needs a developer. Ten, not eighty.

22  Rewrite the opening 40 words of this page so they directly
    answer the query.

23  Break this page into passage-level claims that could be lifted
    and quoted without the surrounding paragraph.

24  For each of these URLs return exactly one verdict: keep,
    refresh, consolidate or kill. Evidence required for each. A
    verdict with no evidence is a guess.

25  Build a content brief: the one thing the reader can do
    afterwards, the angle, the section outline, the proof rows to
    use, and what this deliberately does not cover.

Social and LinkedIn, 5

26  Find five posts published in the last six hours, past 200
    reactions and climbing, on my pillar topics. Rank by velocity,
    not total.

27  Write five comments for this session. No two share a
    structure. At most two mention my own work. At least one
    outside my niche.

28  Score this comment 1 to 10. Below 7, rewrite rather than post.

29  Write a lead magnet post. Three variants, three keywords in
    the same family. Never numbered. CTA depends on whether my
    audience is mostly connected.

30  This post will pull 400 comments. At 20 connection requests a
    day, how long to deliver to everyone? Give me the order.

Email and lifecycle, 5

31  Which lifecycle flows am I missing? Check the standard set:
    welcome, abandonment, browse, post-purchase, winback,
    re-engagement, referral.

32  Build this sequence. Before writing, tell me the trigger
    event, the exit event and the one action it exists to cause.
    No exit event, no sequence.

33  Write five emails. Subject under 45 characters, one CTA each,
    and each adds something the previous did not.

34  Check my deliverability against the thresholds: bounce under
    2%, complaints under 0.10%, SPF, DKIM, DMARC.

35  Write the sequence for people who downloaded something and
    never bought. Largest segment in most businesses and the
    warmest.

Paid and CRO, 5

36  Write five headline variants for this platform,
    character-counted. Each differs in ANGLE, not wording. A
    synonym swap is not a variant.

37  Audit this landing page. What is its ONE job? If the answer
    contains two actions, stop and force a choice.

38  Count the exits on this page: nav links, footer links, other
    offers. Tell me what each one costs.

39  My frequency is climbing and CTR is falling. Is that creative
    fatigue, and what is the evidence?

40  Give me the single change that would move conversion most on
    this page. One change, not a list.

Analytics, 5

41  For each metric in this report, give me the absolute beside
    the percentage, what it is measured against, why if we know,
    and the decision it should inform. Cut anything that cannot
    answer the last one.

42  Work the diagnostic chain in order and stop at the first
    failure. Do not report five stages when the first is broken.

43  Set lead source to the campaign name and the signal type,
    never the channel.

44  Which content produced conversations rather than reach last
    month? Tell me what those pieces have in common.

45  Give me one change for this month. One. Not three.

Agencies, 5

46  Derive this client's ICP from their closed-won, not mine. Show
    me the gap between who they say they sell to and who signed.

47  Build this client's voice file from three of their real pieces.
    Tell me if three is not enough to work from.

48  Calculate gross margin per client this month: fee, delivery
    hours by discipline, loaded cost, account management, QA,
    tool cost. Rank worst first.

49  Check for prospect overlap between two client lists. Same
    person, two accounts, is a problem I need to see first.

50  This client wants daily posts. Tell me why three a week is the
    right answer and how to say it without sounding like
    under-delivery.

Prompt 48 produces a ranking that is almost never the same as the ranking by fee size. Agencies manage attention by fee size, which is how the least profitable client quietly gets subsidised by the best one.

25 fixed formats for the recurring documents.

A template's job is to stop the same decision being remade every month. If a document gets written more than twice a year, it needs one.

Strategy and research, 5 templates

01. Quarterly strategy, one page. The number moving from X to Y, the three channels ranked with what each produced last quarter, the one channel being stopped, the bet, and what must be true in 90 days.

02. Voice of customer. Twelve verbatim phrases with sources. Five words they use that you do not. Three unprompted objections. What they already tried and what failed about it.

03. Competitor brief. Per competitor: what they sell in their words, who they position against, where they are genuinely better, the trap question, and what never to say.

04. ICP, derived. The seven outputs from closed-won, with the gap between stated and actual called out separately at the top.

05. Positioning one-pager. The line, who it repels, two contrarian beliefs, the named competitors with honest differences.

Template 03 fails if every row favours you. A brief where you win everything reads as marketing and gets discounted entirely, including the true rows.

Content, 5 templates

06. Content brief. The one thing the reader can do afterwards, the angle, the section outline, the proof rows to use, what this deliberately does not cover, and the internal links.

07. Editorial calendar. Four weeks visible at all times. Per slot: date, pillar, format, status, owner. Status values are IDEA, DRAFT, REVIEW, SCHEDULED, LIVE and nothing else.

08. Repurposing map. One long asset down the left, every derived piece across: social posts, email, carousel, comment-ready opinions. Nothing gets written from a blank page.

09. Post template library. The six hook shapes, each with a worked example from your own published work rather than a generic one.

10. Pre-publication checklist.

[ ] Hook is lines 1 to 2 and earns the See More alone
[ ] One idea per line, three lines maximum per paragraph
[ ] One CTA
[ ] Every claim traces to a proof row
[ ] Contents listed as objects, not outcomes
[ ] Humaniser pass complete
[ ] No em dashes

Campaign and channel, 5 templates

11. Campaign brief. One job, one angle lever, the audience, the proof rows, the UTMs, the send budget, and the kill criteria with a date.

12. Offer one-pager. One sentence, what they receive as objects, week 1 week 2 week 4, what is not included, the price, the wedge, the disqualifier.

13. Lead magnet post spec. The resource contents as objects, the three variants with their keywords, the CTA by connection state, and the delivery plan before anything posts.

14. Sequence spec. Per node: action, delay, exact copy, exit condition. Plus the trigger event, the exit event, and the one action it exists to cause.

15. Landing page spec. The one job, the headline matching the ad, what is inside as objects, the single CTA, the exits to remove, the proof with attribution.

Template 11's kill criteria field is the one people leave until later. Deciding what failure looks like after seeing the numbers is not a decision.

Reporting, 5 templates

16. Weekly report. The diagnostic chain result stopping at the first failure, one metric per stage with its action, and exactly one recommended change.

17. Monthly report. What changed, what it produced, which content drove conversations rather than reach, and which signal type produced closed deals.

18. Metric narration format. The number with its absolute, what it is measured against, why if known, and the decision it informs. Anything that cannot answer the fourth gets cut.

19. Experiment log. Date, what changed, the hypothesis, the result, and whether it can be attributed. One change per week or this column reads UNATTRIBUTABLE.

20. Attribution schema. The lead source convention, the signal type values, and the segment values. Written once, enforced everywhere.

BANNED IN EVERY REPORT
- Impressions as a headline number
- Percentages without the absolute beside them
- Any metric with no action attached

Agency, 5 templates

21. Client context file. The engagement, the contract dates, the fee, what it excludes, the ICP, the proof with permission, the send budget, and what other campaigns run on the same account.

22. Client voice file. Three of their pieces pasted in full, three adjectives, two phrases they would never say, and what genuinely annoys them.

23. Client report. Categories not raw counts. Interested versus timing versus objection. Plus the objection ranking, which is the most useful thing you hand over all month.

24. Margin tracker. Per client: fee, delivery hours by discipline, loaded cost, account management hours, QA time, tool cost, gross margin percent. Ranked worst first.

25. Handoff brief. Per booked call: the signal with its date and URL, the full thread verbatim, their problem in their words, the objection raised, and what was committed to.

Template 25's last field protects both sides. A founder walking into a call not knowing what was promised is how a good lead becomes an awkward thirty minutes.

Which five to build first

06  Content brief
10  Pre-publication checklist
11  Campaign brief
16  Weekly report
20  Attribution schema

Those five cover the documents that get written weekly. The other twenty are worth building when the work demands them, not in advance.

20 workflows on a schedule. Build these last.

Automate what already works manually, or you automate a broken process and it breaks faster.

Daily, 6 workflows

Time	Workflow	What it does, and the stop condition
07:00	High intent scan	ICP posting about the problem, or asking for a recommendation. Same-day or not at all. Nothing found means one line, not a report.
07:15	Borrowed reach scan	Posts under six hours old, past 200 reactions, on a pillar topic. Ranked by velocity, not total.
08:00	Scheduled publish	Today's piece goes live, then an auto-comment 20 minutes later. The gap is deliberate.
09:00	Engagement session	25 minutes, six rules, every comment scored before posting.
11:00	Reply classification	Every inbound sorted into seven categories. Below 8 confidence escalates. Drafts only, never sends.
17:00	Tag and log	Lead source, signal type and segment written against every new contact.

Start with three: the 07:00 scan, the 09:00 session, the 11:00 classification. Those run a complete daily motion, and adding the rest before those are reliable just multiplies the failures.

Weekly, 8 workflows

Day	Workflow	What it does
Mon 08:00	Diagnostic chain	Five stages in order, stop at the first failure, output ONE change.
Mon 09:00	Account health check	Sending accounts, domains, deliverability thresholds, anything capped or parked.
Mon 10:00	Warm signal scan	Anyone marked contacted or cold who has since engaged, viewed, or changed job.
Tue 09:00	Content batch	Next week's three pieces, drafted and humanised, status DRAFT.
Tue 11:00	List build and grade	New rows, filtered, graded A to F. Below B, rebuild rather than proceed.
Wed 15:00	Experiment audit	What changed last week and whether the lift can be attributed to it.
Thu 09:00	Sequence top-up	New contacts into active campaigns, within the daily split.
Fri 16:00	Weekly report	Every metric paired with an action. Drafted, not sent.

Wednesday's experiment audit only works if Monday produced one change. Change three things and the audit reads UNATTRIBUTABLE, which is the honest answer and a wasted week.

Monthly and quarterly, 6 workflows

Cadence	Workflow	What it does
Monthly	Signal attribution	Which signal type produced closed deals. One line, and it changes next month's targeting more than anything else.
Monthly	Objection ranking	The five most frequent, ranked. A positioning insight as much as a sales one.
Monthly	Long asset production	The one substantial piece the month's channel content comes from.
Monthly	Margin review	Gross margin per client or per line, ranked worst first.
Quarterly	ICP re-derivation	Against the last quarter's closed-won. If the winning title moved, everything downstream needs rebuilding.
Quarterly	Content refresh	Keep, refresh, consolidate or kill across the back catalogue. Five to ten pieces, prioritised.

The monthly signal attribution is the highest-value workflow here and the most likely to be skipped, because it produces one line rather than a dashboard.

The rules that keep 20 workflows readable

A quiet period produces one line, not a report. Any workflow that emits a full report every run regardless of findings gets ignored by week three. Build the nothing-found branch first.

Nothing auto-publishes or auto-sends. Every workflow that touches a person or a channel drafts and routes. A human presses the button.

Every workflow states what it could not assess. A diagnostic on partial data points at the wrong stage, which is worse than no diagnostic.

Errors notify with the full payload. A failed import you cannot replay is a lost list.

Where the sending workflows run

Four of the twenty touch a real person: the reply classification, the sequence top-up, the warm signal scan, and the account health check.

Prosp runs those. Cloud-based, so the schedule does not depend on a machine staying open, and the unified inbox spans every connected account so the 11:00 classification runs over one queue rather than several.

Lead states drive the Monday warm scan directly: Not Contacted, In Campaign, Contacted, Not Accepted after 14 days with auto-withdraw, Replied which stops the campaign, Duplicate, Blacklist, Failed.

Ten audits. Each produces a verdict, not a list.

An audit that returns eighty items is a document nobody actions. Every one of these caps its output and ends with a single recommended change.

01. The positioning audit

SIX CHECKS, each PASS, WEAK or FAIL with evidence:

1  Could a stranger say what you do in one sentence after five
   seconds on the homepage?
2  Does the positioning repel anyone? If nobody, it is too broad.
3  Is there a belief stated that your market mostly disagrees
   with?
4  Does the language match the twelve verbatim phrases, or the
   language of a workshop?
5  Is the named competitor difference honest, including where
   they are better?
6  Would the disqualifier survive a sales call?

VERDICT: one of SHARP, BROAD or ABSENT.

BROAD is the most common and the most expensive. It produces
content nobody argues with and a pipeline of people who were never
going to buy.

02. The ICP audit

COMPARE the stated ICP against the last 12 months of closed-won.

REPORT:
- The job title that actually signs, versus the one being targeted
- The size band with the highest win rate, versus the one in the
  plan
- The three attributes shared by fast closes
- The three shared by long or lost deals
- Whether the disqualifier list matches the actual losses

VERDICT: ALIGNED, DRIFTED or UNEVIDENCED.

FAIL CLOSED with fewer than 10 closed-won deals. Report the count
and say the answer is not yet in the data.

DRIFTED is normal after two quarters. It is only expensive when
nobody checks.

03. The content audit

PER URL, return exactly one verdict with evidence:

KEEP         performing, accurate, still on-strategy
REFRESH      right topic, stale detail or weak opening
CONSOLIDATE  competing with another page for the same intent
KILL         no traffic, no links, off-strategy, or embarrassing

A verdict with no evidence is a guess. Include the traffic, the
last update date, and the intent it serves.

CAP THE OUTPUT at ten REFRESH items per quarter, prioritised. A
list of eighty is a list nobody works.

CONSOLIDATE is the verdict people avoid because it means admitting
two pieces should have been one. It is also where the fastest
gains are.

04. The technical SEO audit

FROM CONNECTED DATA ONLY. Claude cannot see live rankings, crawl
status or Core Web Vitals. Work from an export or say what you
need.

CHECK: crawlability, index coverage, robots and sitemap against
the latest crawl, Core Web Vitals, redirect chains, broken
internal links, canonical conflicts, schema validity.

OUTPUT TEN items, prioritised by impact. Per item: the issue, the
evidence, the fix, the effort, whether it needs a developer.

VERDICT: BLOCKING, DEGRADED or HEALTHY.

BLOCKING means stop all content work until it is fixed. Content on
a site that cannot be crawled is wasted effort, and content work
is more enjoyable, which is why people do it first.

05. The AI search readiness audit

PER PAGE:
- Do the opening 40 words directly answer the query
- Are there standalone claims that could be lifted and quoted
  without the surrounding paragraph
- Is the entity clear: who and what this is about
- Is every figure attributed to a source
- Does it cover the questions actually asked, not the ones you
  wish were asked

VERDICT: QUOTABLE, PARTIAL or INVISIBLE.

ONE HONEST CAVEAT. Observational research across roughly 2 million
citations found prompt-content alignment to be the dominant
page-level predictor, while FAQ blocks, structured data and Core
Web Vitals collapsed toward zero once domain effects were applied.
That study is observational and cannot establish causation.

So: treat schema as hygiene rather than a lever, and spend the
effort on whether the page actually answers the question asked.

06. The profile and channel audit

NINE CHECKS on the primary profile, each PASS, WEAK or FAIL:
proof, empty adjectives, clarity, gaps, call to action, visual
consistency, mobile readability, searchability, contact hygiene.

THEN THE CHANNEL:
- Posting cadence against the ceiling. One a day maximum.
- Engagement session running, or not
- Which pillar each of the last 20 pieces served
- Whether any pillar has produced nothing in eight weeks

VERDICT: READY, LEAKING or BLOCKED.

LEAKING is the common one. The content is fine and the profile
loses the visit, which shows up as a low acceptance rate that
people then blame on the message.

07. The lifecycle audit

CHECK THE STANDARD SET. Most teams are missing two.
Welcome, abandonment, browse, post-purchase, winback,
re-engagement, referral request.

PER EXISTING FLOW:
- Does it have an exit event? No exit event is a failure, not a
  weakness.
- When did it last get touched?
- Does email 1 deliver what was promised in the first two lines?
- Does every email add something the previous did not?

THEN DELIVERABILITY:
Bounce under 2%. Complaints under 0.10%. SPF, DKIM, DMARC.

VERDICT: COMPLETE, GAPPED or AT RISK.

AT RISK overrides everything. A sequence landing in spam is a
sequence that does not exist, and building more flows on a broken
domain compounds the problem.

08. The funnel audit

WORK THE CHAIN IN ORDER. Stop at the first failure.

1 AWARENESS LOW        reach and engagement rate
2 AWARENESS FINE,
  INTEREST LOW         the hook, or the audience is wrong
3 INTEREST FINE,
  INTENT LOW           the offer, not the content
4 INTENT FINE,
  ACTION LOW           the page, or the ask
5 ALL FINE,
  REVENUE LOW          price or fit, not marketing

OUTPUT the failing stage, the evidence, and ONE change.

VERDICT: names the stage. Nothing else.

Do not report five stages of numbers when the first is broken.
A stage 4 recommendation delivered while stage 1 is failing is
why audits get ignored.

09. The attribution audit

CHECK:
- Does lead source name the CAMPAIGN or the channel? "LinkedIn"
  tells you nothing in three months.
- Is signal_type populated? Without it you cannot answer which
  trigger produced revenue.
- Are UTMs consistent with the naming convention?
- How many records have no source at all?
- Can you answer, right now, which campaign produced last
  quarter's best deal?

VERDICT: ANSWERABLE, PARTIAL or BLIND.

BLIND is more common than anyone admits, and it is the reason
teams keep running things that do not convert. The question is not
whether the data is perfect. It is whether one specific question
can be answered.

10. The agency margin audit

PER CLIENT, monthly:
  Fee received
  Delivery hours by discipline
  Loaded cost of those hours
  Account management hours
  QA and revision time
  Attributable tool cost
  Gross margin, in currency and percent

RANK BY MARGIN PERCENT, worst first.

THEN PER CLIENT AT THE BOTTOM:
- Is it scope creep, underpricing, or a delivery problem?
- What would have to change, specifically?
- What is the conversation, and when?

VERDICT per client: HEALTHY, WATCH or RENEGOTIATE.

The ranking is almost never the same as the ranking by fee size,
and agencies manage attention by fee size. That is how the least
profitable client quietly gets subsidised by the best one, for
months.

The order to run them in

BEFORE ANYTHING       02 ICP, then 01 positioning
BEFORE CONTENT WORK   04 technical SEO. BLOCKING stops everything.
QUARTERLY             03 content, 05 AI search, 06 profile
MONTHLY               08 funnel, 09 attribution
MONTHLY, AGENCIES     10 margin
BEFORE EMAIL WORK     07 lifecycle. AT RISK overrides.

Two of these gate the others. A BLOCKING technical verdict stops content work, and an AT RISK deliverability verdict stops email work. Running the others first produces recommendations you cannot action.

The guide builds it. Prosp runs the half that reaches people 💜

↳ Start a 14 Day Free Trial 💜 ↳ Watch Full Tutorial ▶️