The outreach engine: turn your foundation into pipeline that fills itself

Your team already has the outreach-engine and account-brief playbooks — the worked recipes for writing a grounded opener from a CSV and turning a pile of notes into a tight pre-call brief. Each is a good move, done once. This module is the layer that turns those separate moves into a single machine: a prospecting system that turns cold names into researched, on-narrative pipeline every week — whether or not anyone felt sharp that morning.

It’s Module 2 of the certifiable Sales track, and it inherits everything you built in Module 1. Your value-narrative.md, icp.md, and battlecards.md are the fuel this engine runs on. M1 mastered what you sell on, who you sell to, and how you handle the pushback; M2 builds the machine that turns that into pipeline — and assesses the machine, not a single lucky cold email.

If the foundation is the fuel, the engine is the machine that burns it on a schedule. A great cold email is a spark — it needs a good day, the right rep, and an account you happened to know something about. A system is a machine — a name goes in one end and a qualified, researched, on-narrative sequence comes out the other, every week, at a quality that doesn’t swing with who’s prospecting. The free playbooks teach the sparks. This module is where you stop being heroic and start being repeatable.

The system, not the spray

Most reps treat Claude as a slot machine — feed it a clever line, hope a good cold email falls out. It often does, which is the trap. A sharp opener from a one-off prompt feels like a win, but it isn’t a system: next week needs a fresh flash of inspiration, the quality swings with whoever’s prospecting, and nothing compounds. The spray fills a send queue, not a pipeline.

A system is five named steps that run the same way every time, so the output is consistent instead of heroic:

• Qualify — score the lead against icp.md before you spend a minute on it. STRONG / WORTH A CALL / DISQUALIFY. This is the cheapest step and it gates the expensive ones; a disqualified lead stops here, not after you’ve written it a beautiful email.
• Research — turn the account into a brief: where they stand, what they care about, the trigger event, the fit score, and the one specific hook. This is the step that decides quality; the rest is execution.
• Personalize — write the opener from the brief plus value-narrative.md. Every line cites a real account fact and earns its way back to a pillar. Because the research was done upstream, this step is just phrasing.
• Sequence — build the multi-touch follow-up off the same brief. Each touch adds a new reason; none assumes a yes the last touch didn’t earn. The sequence is what turns one opener into pipeline instead of a single shot.
• Log — record the account, the touch, and the next step in the CRM so the work compounds. The log is what turns a flurry of sends into a cadence a different person can pick up.

Each step has one job and one input, which buys you two things. First, a different person can own each step — the SDR who researches isn’t the rep who closes. Second, when a deal goes cold you can name the broken step: it’s a bad-fit lead (qualify), a flat opener (the brief was thin), or a sequence nobody followed up on — not a vague “cold outreach doesn’t work.” A system you can debug beats a spray you have to get lucky with.

You run the whole thing in the chat, no terminal on the main path. Open your sales folder in Claude Desktop, approve the read of the leads file and the foundation docs in the “Ask permissions” prompt, and work the steps one at a time.

The brief is the unit of quality

A system is only as good as what enters the opener, and what enters the opener is the brief. Get the brief right and the opener writes itself — grounded, specific, impossible to mistake for mail-merge. Get it thin and no amount of clever phrasing saves you; you’re just polishing flattery. The brief is where research becomes judgment, which is exactly why it’s the thing the free account-brief playbook can build but can’t teach you to judge.

A brief Claude can turn into a landing opener names five things — and a brief missing any one produces a confident, generic line:

• Where they stand. The relationship and the current state — net-new with no history, or a warm account mid-cycle; what they use today; who’s involved. An opener that doesn’t know where the account stands opens on the wrong foot.
• What they care about. The one or two things the buyer actually loses sleep over, in their words — not your feature list. This is the difference between “we help logistics firms” and “you’re running three branches off one Excel file on Saturdays.”
• The trigger event — the “why now.” The thing that makes this week the moment: a new branch, a hire, a deadline, a stumble. An opener with a real trigger reads like you noticed; one without reads like a blast.
• The fit score, from the M1 rubric. The brief carries the score, not just a vibe — STRONG / WORTH A CALL / DISQUALIFY and the signal that decided it. A brief that can’t say why this account is worth the effort is a brief for an account you shouldn’t be working.
• The one specific hook. The single true thing only this account gives you — the line you couldn’t say about any other prospect. This is what makes the opener theirs instead of the email every rep could send. It comes straight from the research; it’s never invented to fill the gap.

Write the brief well and you’ve front-loaded every judgment call into one short document, so the opener step is genuinely just phrasing. That’s the whole reason the brief is its own step: it’s where a person decides what’s true and what matters, so the opener can be where Claude produces.

Research once, sequence many

The cheapest outreach you’ll ever write is the outreach you’ve already researched. One account’s brief — the where-they-stand, the trigger, the hook — holds an entire sequence inside it: the opener, the proof touch, the angle shift, the breakup. The engine’s job is to re-cut that one piece of research across touches, not to invent fresh context for each one.

The distinction that matters is sequence vs. reproduce:

• Sequence is one researched account’s real context re-cut for a new touch and a new angle — the trigger cut, the proof cut, the cost-of-doing-nothing cut, the graceful breakup — all tracing back to the same brief and the same facts you already verified.
• Reproduce is asking Claude for “three more follow-ups for this account,” which quietly invents three new things about the prospect you now have to fact-check from scratch. At volume, that’s how a wrong headcount or an imagined funding round ends up in a buyer’s inbox before anyone notices.

So the rule is one brief → a full sequence, zero new facts. You feed Claude the brief plus value-narrative.md and ask for distinct touches off the same research — each adding a new reason, each on-narrative, none making a claim the brief didn’t. And the follow-up escalates: touch two adds proof, touch three re-anchors on the cost of doing nothing, the last touch is an honest breakup that never assumes a yes the account didn’t give.

And the log is the cadence. A handful of touches in a draft folder is not a sequence; a logged, dated sequence in the CRM — which touch goes out on which day, with the reply tracked — is. The brief is the unit of quality; the log is what turns one good brief into pipeline that a teammate can carry when you’re on a call.

Qualify before you personalize

Here’s the step reps skip because it feels like friction: scoring fit before writing a word. A beautifully personalized email to a disqualified lead is wasted effort dressed up as diligence — you researched, you crafted, you sent, and the account was never going to close. The most expensive opener you can write is the perfect one to the wrong buyer.

Qualify is step one for a reason, and the discipline has two edges:

• Score against the rubric, not the gut. Run the lead through icp.md — the fit signals you can check and the disqualifiers you named in M1 — and let the score decide whether the account earns the research step at all. A rep who personalizes everything is a rep who researches a 500-person enterprise with its own finance department as carefully as the perfect-fit owner. Same effort, a quarter wasted.
• The disqualifiers do more work than the fit signals. The whole value of qualifying first is the stop: the tempting-but-wrong account you walk away from before you’ve sunk an hour into its brief. A lead that trips a disqualifier doesn’t get a clever opener; it gets a DISQUALIFY and your time goes to the next STRONG.

This is also what keeps the system honest about volume. The spray sends to everyone because qualifying is work; the engine sends to fewer, better accounts because it qualified first — and that’s why its reply rate doesn’t collapse the moment the easy names run out. Qualify gates research; research gates the opener. Run the steps in that order and every expensive step is spent on an account worth it.

The bilingual cadence — a peer lane, not a translation

Here’s where most teams break the engine: they run the entire English system, then translate the openers into Arabic at the very end and re-time nothing. The result is a sequence that reads, in Arabic, like a translated cold email — technically correct, emotionally foreign — landing in a Gulf inbox on a Friday.

The fix is the rule M1 taught for the foundation — adapt, don’t translate — applied now to the outreach. The Arabic lane is its own narrative on its own cadence:

• Its own opener, not a translated one. An Arabic account gets an Arabic opener authored from value-narrative-ar.md: its own angle, and often a different leading pillar. For the Gulf owner the FTA-accepted VAT export is the proof that earns the first reply, so the Arabic opener can lead on Trustworthy where the English led on Speed. You adapt the brief’s hook to what lands here; you never adapt the finished sentence.
• The lead objection can come first. For the Gulf owner, “I already have an accountant” often arrives first and hardest — the family accountant is a real relationship, not a line item. So the Arabic opener pre-empts it warmly rather than waiting for the rebuttal, pulling the reframe straight from your Arabic battle-cards.
• Re-time the cadence to the work week. The Gulf week runs Sunday to Thursday — land your touches Sun–Thu, never Friday, and ease the pace and shift the send window during Ramadan. The same four-touch shape, re-timed; a sequence that lands on the regional rhythm reads like a peer, not a foreign vendor on a New York calendar.
• Same commercial gate, both languages. An Arabic quote, recap, or expansion ask clears the same deal-desk and manager sign-off as the English. The failure mode is the approver who can’t read it waving it through — so the gate is cleared by someone who actually reads Arabic.

Done this way, Arabic is a peer lane in your engine, not a post-processing step. In this region, that’s the difference between outreach that earns the reply and outreach that signals you’re selling into the market rather than building in it.

Your assignment

Build the engine for one product — your own (recommended: the output is a real prospecting system your team keeps) or the sample brand Mizan, the GCC small-business bookkeeping tool whose sales team runs throughout this track. Everything you produce inherits the three foundation files from M1; if you don’t have them yet, do Module 1 first. Open your sales folder in Claude Desktop, approve each read in the “Ask permissions” prompt, and work in the chat — no terminal needed.

Module 2 deliverable — the outreach engine

Inherits from M1: value-narrative.md + icp.md + battlecards.md
                  (+ value-narrative-ar.md)

1. The outreach SYSTEM   (the repeatable flow)
   - the five named steps, written so a teammate could re-run it:
     qualify → research → personalize → sequence → log
   - one input + one job per step; who owns each
   - the "name the broken step" map for when a deal goes cold
   - the commercial gate marked: pricing/terms NEVER a rep solo

2. Personalized openers   (grounded in real account research)
   - run a real leads CSV through qualify, then research the STRONG rows
   - ONE opener per qualified lead, each citing a real account fact
   - DISQUALIFY rows named; thin rows marked SKIP (+ data you'd need)
   - no invented funding round, headcount, or quote — zero new facts

3. A multi-touch follow-up sequence   (off the SAME brief)
   - 4 touches over ~12 days; each adds a NEW reason
   - escalates: opener → proof → cost-of-nothing → graceful breakup
   - never assumes a yes the last touch didn't earn

Bilingual teams: the AR lane carries its OWN openers — author from
value-narrative-ar.md, re-time the cadence Sun–Thu, lead on the pillar
that lands here; don't translate the English at the end.

The Foundation toolkit you installed in M1 already holds your CLAUDE.md and the three source-of-truth files. The toolkit additions at the end of this module give you the system and opener-plus-sequence templates, so you’re filling in structure, not staring at a blank queue.

How it’s graded — the rubric

This is the part the free playbooks don’t have, and the part that makes the credential mean something. Your three deliverables are scored against five criteria. Each is meets / nearly / not yet, and a “nearly” on any one is a revise, not a pass.

Outreach-engine rubric

1. The system is repeatable    The flow is five named steps anyone could
                               re-run — qualify, research, personalize,
                               sequence, log — not one clever cold email.

2. Openers are researched      Each opener cites a real, specific fact about
                               the account, not generic flattery. A thin
                               row is SKIPped, not faked. No invented stats.

3. On-narrative, qualified-1st  Every account is scored against icp.md BEFORE
                               it's worked; every opener serves a named
                               pillar from value-narrative.md. No effort
                               spent on a disqualified lead.

4. Data handled safely         Lead data stays in the approved workspace; no
                               raw lead PII pasted into a shared prompt;
                               [prospect]-style placeholders used. The
                               commercial gate is marked, not skipped.

5. The Arabic lane is a peer    Its own opener authored from value-narrative-
                               ar.md, its own cadence (Sun–Thu), proof and
                               lead objection localized — not the English
                               translated at the end. (Bilingual teams.)

The discipline is what a sales lead would demand before pointing a team at a target list. A system that depends on one rep’s good day, or a sequence that’s really English outreach wearing Arabic, fails quietly on every send — so it has to be caught here.

The bar, shown — a worked model answer (Mizan)

You don’t have to guess what “meets” looks like. Here are passing excerpts for the sample brand’s new-business prospect, Najd Logistics — yours won’t look identical, it just needs to clear the same bar. Mizan’s English outreach stays in English; the Arabic lane is authored, not translated.

account-brief.md — Najd Logistics (excerpt)

Where they stand
  Net-new prospect, no history. Owner-run logistics firm, 30 staff,
  3 branches, VAT-registered. Today: the owner does the books himself,
  in Excel, on weekends.

What they care about
  - getting his weekends back — month-end currently eats them
  - not getting caught by a VAT deadline (3 branches = 3x the mess)
  - numbers he can trust without paying an accountant to redo them

Trigger (the "why now")
  Opened the third branch last quarter — the Excel file that worked
  for one branch is buckling under three.

Fit score (from icp.md)
  VAT-registered ✓  owner in the books ✓  month-end/VAT pain ✓
  multi-branch ✓  no finance dept ✓  on spreadsheets ✓
  → 6/6 fit, 0 disqualifiers → STRONG FIT

The one hook (only this account gives you)
  Three branches, one Excel file, one owner doing it on Saturdays.
  THAT'S the line — not "we help logistics companies."

openers.md — Najd Logistics (3 grounded; pick one)

A — the trigger cut
  "Congrats on the third branch. The spreadsheet that handled one
  branch tends to crack around the third — three VAT positions
  reconciled by hand is where most owners lose the weekend. That's
  the exact point Mizan was built for."

B — the weekend cut  (chosen for T1)
  "Saw you run the books yourself on weekends across all three
  branches. Mizan keeps the three sets reconciled as you go, so the
  month-end close is an afternoon, not a Saturday — and the VAT
  export is FTA-ready without a redo."

C — the trust cut  (→ leads on Trustworthy)
  "Three branches, one VAT return, no room for a slip — a late return
  is an AED 1,000+ penalty. Mizan keeps every figure traceable to its
  source, so the export files clean the first time."

Every line cites a fact from the brief. Nothing invented; no flattery.

sequence.md — Najd Logistics (4 touches, Sun–Thu cadence)

T1  Day 0  (Sun)   Opener B (the weekend cut) — above.
T2  Day 3  (Wed)   Proof touch — "One owner running 3 retail branches
                   closed his first month on Mizan in an afternoon; the
                   VAT export was accepted by the FTA without amendment."
                   (real proof from the narrative — no NEW claim) + soft ask.
T3  Day 8  (Mon)   Angle shift, cost of doing nothing — "Three branches
                   means three chances to miss a deadline; one slip is
                   AED 1,000+. Worth 15 minutes to see the close?"
T4  Day 12 (Thu)   Graceful breakup — "I'll leave it here. If the weekend
                   books ever stop being worth it, the door's open."

Each touch adds a NEW reason; none assumes a yes T1 didn't earn. The
breakup is honest, not a guilt trip. No price is quoted — if he asks for
one, it goes to deal-desk BEFORE he sees it (the commercial gate).

openers-ar.md — Najd Logistics (note)

Authored, not translated. The Arabic opener LEADS on Trustworthy —
«أرقام تقدر تسلّمها بدون مراجعة» (numbers you can hand over, no redo) —
because for the Gulf owner the FTA-accepted export is the proof that
earns the first reply, and "عندي محاسب / I already have an accountant"
lands first, so the opener pre-empts it warmly. Cadence re-timed: land
Sun–Thu, never Friday; ease the pace and shift the window during Ramadan.
Proof local (FTA, the 9,000+ GCC count). Numbers Western.

What you keep — toolkit additions

These extend the Foundation toolkit you already installed — same folder, same CLAUDE.md, two more templates and a few more saved prompts. Drop them next to your foundation files, and every account in the system starts from structure instead of a blank queue.

# outreach-system.md — [Product] outreach

## The five steps   (anyone can re-run; one owner per step)
1. Qualify    — score the lead against icp.md → STRONG / WORTH A CALL /
                DISQUALIFY. Disqualified stops HERE. (cheapest step; gates the rest)
2. Research   — turn the account into a brief: where they stand, what they
                care about, the trigger, the fit score, the ONE hook.
3. Personalize— write the opener FROM the brief + value-narrative.md; every
                line cites a real account fact. Invent nothing.
4. Sequence   — build the multi-touch follow-up off the SAME brief; each touch
                adds a new reason; none assumes a yes.
5. Log        — record the account, the touch, and the next step so a teammate
                can pick it up.

## When a deal goes cold — name the broken step
bad lead → qualify   ·   flat opener → research/brief
no reply → sequence  ·   dropped after a reply → log

## The commercial gate   (never a rep solo)
Pricing, discounts, terms, and the expansion ask → deal-desk or a manager
BEFORE the buyer sees them. You draft the quote; a person agrees the terms.

# openers-and-sequence.md — [Account / batch]

## Account brief   (the unit of quality — fill BEFORE any opener)
Where they stand:    [relationship / current state / what they use today]
What they care about:[1–2, in their words]
Trigger (why now):   [the event that makes this the moment]
Fit score (icp.md):  [STRONG / WORTH A CALL / DISQUALIFY + the deciding signal]
The one hook:        [the single true thing only this account gives you]

## Openers   (grounded — each cites a fact from the brief above)
A [angle]: [≤2 sentences · one real fact · ties to a value-narrative.md pillar]
B [angle]: ...
(No "impressed by your work." Nothing real to say → SKIP + the data you'd need.)

## Follow-up sequence   (off the SAME brief; Sun–Thu cadence)
T1 Day 0 : [opener]
T2 Day 3 : [proof touch — a real win already in the narrative; no new claim]
T3 Day 8 : [angle shift — the cost of doing nothing]
T4 Day 12: [graceful breakup — never assumes a yes]
(AR lane = its OWN openers + cadence — adapt, don't translate.)

The saved-prompt library (Engine) — add to the Foundation set

Research an account into a brief
  "Read value-narrative.md and icp.md. Here's everything I have on
   [account]: [paste notes / thread / CRM row — only what's needed]. Write
   the account brief: where they stand, what they care about, the trigger
   event, the fit score against the rubric (and the signal that decided it),
   and the ONE specific hook only this account gives me. Cite only facts in
   what I gave you — invent nothing. Don't write the opener yet."

Write grounded openers from a leads CSV
  "Read value-narrative.md and icp.md. For each row in leads.csv, first score
   fit. For STRONG / WORTH A CALL rows, write ONE opener (≤2 sentences) that
   cites a real fact from that row and ties it to a named pillar. DISQUALIFY
   rows: write DISQUALIFY + why. Thin rows: write SKIP + the data I'd need.
   Invent no funding round, headcount, or quote. Output a table: company,
   fit, opener."

Build a follow-up sequence off the brief
  "Read the account brief and value-narrative.md. Build a 4-touch follow-up
   over ~12 days, Sun–Thu only: T1 the opener, T2 a proof touch (a real win
   already in the narrative — no new claim), T3 an angle shift to the cost of
   doing nothing, T4 a graceful breakup. Each touch adds a NEW reason; none
   assumes a yes the last touch didn't earn. Keep each under 60 words and
   on-voice. Don't quote a price — that's deal-desk's."

Build the Arabic lane (adapt, don't translate)
  "Read value-narrative-ar.md and icp.md. For [account], write the Arabic
   opener and sequence from scratch for a Gulf owner: its own angle, the
   proof that actually earns the first reply here (FTA-accepted export, the
   GCC count), and lead on whichever pillar lands — which may differ from
   the English. Pre-empt the 'I already have an accountant' objection warmly.
   Re-time the cadence to Sun–Thu, never Friday; ease it during Ramadan. An
   adaptation with its own copy, not a translation."

What you’ve proven — and what’s next

Clear the rubric and you’ve proven something a folder of clever cold emails never can: that your team turns cold names into qualified, researched, on-narrative pipeline predictably — a system, not a streak of good days. That’s the Engine stage of “Certified Sales with Claude.”

From here the track turns a filling pipeline into closed-and-growing revenue, each module assessed the same way:

• Module 3 — running the deal: discovery that earns the next step, the call recap that creates momentum, and the proposal drafted to the value narrative — the deal work the engine feeds.
• Module 4 — the big play, Module 5 — measure & forecast, then the capstone — one account taken from cold to won-and-growing, end to end, graded into the certificate.

First, make the engine reusable. The toolkit additions above — the system and opener-plus-sequence templates plus the saved prompts — turn the prospecting you just built into files your whole team runs from, extending the Foundation toolkit you already installed. And if you’re rolling this across a team, the operating guide is the data, deal-desk, and sign-off layer that belongs underneath the whole system.