# ICP-narrowing worksheet

Fill this in once, before you build any list. Every field has to have a real answer, not a placeholder. A vague field here is the reason a sequence stalls three weeks in.

## The four fields

1. Industry: the one or two industries you already understand well enough to spot a problem in ten minutes on their site.
2. Role: the exact title of the person who can say yes. Not "leadership," a real title: "VP of Operations," "founder," "head of finance."
3. Company-size band: a band you can actually verify from a company's site or LinkedIn (headcount range, revenue range, or store count).
4. Trigger condition: one concrete event that tells you this company needs help right now. Examples: a funding round in the last 90 days, a new leader in the role you're targeting, an office opening, a product launch, a hiring spike in a department tied to your service.

## Fill it in

- Industry: ______________________
- Role (exact title): ______________________
- Company-size band: ______________________
- Trigger condition: ______________________
- Where you'll find the trigger (one source, e.g. a funding database, a job board, a press release feed): ______________________

## Two worked examples

**Example A: fractional CFO practice**
- Industry: DTC consumer brands
- Role: Founder or CEO
- Company-size band: $2 million to $10 million in yearly revenue
- Trigger condition: raised a seed or Series A round in the last 6 months
- Source: a funding-announcement search alert

**Example B: ops consultant for logistics companies**
- Industry: regional trucking and freight brokerage
- Role: VP of Operations or COO
- Company-size band: 50 to 300 employees
- Trigger condition: posted 3 or more operations job openings in the last 60 days
- Source: a weekly manual check of a handful of company career pages, or a job board search

## Before you move to the list

Read the four fields back out loud. If any answer is a category instead of a specific thing (say, "small businesses" instead of a size band, or "decision makers" instead of a title), stop and narrow it further. A list built on a vague ICP produces a low reply rate that no template can fix.

# Initial outreach message: the five-part structure

Every first message to a prospect follows the same five parts, in this order. Parts two and three are the ones that take real work: they have to reference something true and specific about this one person or company, not a name swapped into a form letter.

## The five parts

1. **Name.** Their first name, spelled the way they spell it.
2. **A specific reason for reaching out.** Say why you're writing to this person now, tied to the trigger condition from your ICP worksheet (a funding round, a new title, a hiring push).
3. **A specific observation about them or their company.** Something pulled from their own site, a recent post, a job listing, an interview, an about page. Not "I saw your company is growing," but the actual detail: the product line they just launched, the specific role they just posted, the line from their own site.
4. **A concrete way you can help.** One sentence naming the exact problem you solve for companies like theirs, not a list of services.
5. **A low-pressure sign-off.** No "book a call this week." Ask a small, easy-to-answer question, or offer to send something useful with no ask attached.

## Template

Subject: [specific, short, references part 3, never "quick question"]

Hi [Name],

[Part 2: the specific reason, one sentence.]

[Part 3: the specific observation, one to two sentences, named detail.]

[Part 4: the concrete way you can help, one sentence.]

[Part 5: the low-pressure close, one sentence.]

[Your first name]

## Filled example

Subject: the ops role you posted last week

Hi Priya,

I noticed Brightline Goods closed a seed round in July and has three open operations roles up right now, including a Director of Fulfillment.

That kind of hiring push usually means the current process is already past what one person can track by hand, especially once order volume doubles.

I work with early-stage DTC brands on exactly that gap between headcount and process, before the new hires start.

Would it be useful if I sent over the two-page checklist I use with brands in that exact spot, no call needed?

Marcus

## What makes a message fail this structure

- Part 3 that could apply to any company in the industry ("I saw you're growing fast") instead of a detail unique to them.
- A part 4 that lists three services instead of naming the one problem that matches their trigger.
- A close that asks for 30 minutes on the calendar in the first message. Save that ask for later touches, once they've replied.

# Follow-up sequence: touches, spacing, and stop conditions

A prospect who does not reply to the first message gets a fixed sequence, never a guess. This file is that sequence: how many touches, how far apart, what each one says, and when to stop.

## The sequence

| Touch | Days after previous touch | Purpose |
|---|---|---|
| 1 | (the initial message, day 0) | introduce, one clear ask |
| 2 | 5 to 7 days later | add one new piece of value, don't just "bump" |
| 3 | 7 to 10 days later | a different angle: a short case, a stat, a question |
| 4 | 10 to 14 days later | the shortest touch in the sequence, one line |
| 5 | 14 to 21 days later | a genuine break-up message |
| 6 (optional) | 30 days later | a check-in tied to a new trigger, not the old one |
| 7 (optional) | 30 to 45 days later | final note, leaves the door open |

Total span: roughly 6 to 10 weeks from first message to last touch if the prospect never replies.

## Rules for every touch

- Every touch is under 80 words. If it's longer, cut it before you send it.
- Never repeat the same angle twice. Touch 2 adds something touch 1 didn't say.
- Never open a follow-up with "just following up" or "checking in." Say the new thing first.
- Each touch references something short from the earlier ones (a "still" or "as I mentioned"), so the thread reads as one conversation, not five cold opens.

## The touch bodies (template)

**Touch 2 (add value):**
Hi [Name], following up with [a specific resource, stat, or example relevant to their trigger]. Happy to send the full version if useful.

**Touch 3 (new angle):**
Hi [Name], a different thought: [one sentence naming a second, adjacent problem tied to their situation]. Worth a quick look?

**Touch 4 (shortest):**
[Name], still worth connecting on this?

**Touch 5 (break-up):**
Hi [Name], I'll stop here unless the timing's off rather than the fit. If it's the former, just say the word and I'll check back later this year.

**Touch 6 (new trigger, optional):**
Hi [Name], saw [new trigger, e.g. the new hire, the product launch]. Wanted to reopen this given the change. Still worth a look?

## Stop conditions

Stop the sequence immediately, the same day, when any of these happen:

- The prospect replies, in any form, even "not now."
- The prospect unsubscribes or asks not to be contacted again.
- You reach touch 7 with no reply.
- You learn the person has left the company or the role no longer exists.

Log the stop reason in references/pipeline-tracker.md the day it happens. Never let a scheduled touch go out after a stop condition has already fired.

# Pipeline tracker: the weekly log

One row per prospect. Keep every prospect in this sheet from the day they're added to the list through the day their sequence closes, whether that ends in a reply, a stop, or a closed deal.

## Columns

| Column | What goes in it |
|---|---|
| Prospect name | full name |
| Company | company name |
| Title | exact title, matches the ICP role field |
| Industry | matches the ICP industry field |
| Size band | matches the ICP size band |
| Trigger | the specific trigger that got them on the list, and its date |
| Channel | outbound, content-inbound, or referral |
| Batch date | the week this prospect's batch was pulled |
| Touch 1 date | date sent |
| Touch 2 date | date sent |
| Touch 3 date | date sent |
| Touch 4 date | date sent |
| Touch 5 date | date sent |
| Touch 6 date | date sent, if used |
| Touch 7 date | date sent, if used |
| Reply date | date they replied, blank until they do |
| Reply type | interested, not now, no, or unsubscribe |
| Outcome | open, closed-won, closed-lost, closed-no-fit |
| Notes | anything specific worth remembering for next time |

## Starter row (copy this structure into a spreadsheet)

| Prospect name | Company | Title | Industry | Size band | Trigger | Channel | Batch date | Touch 1 | Touch 2 | Touch 3 | Touch 4 | Touch 5 | Reply date | Reply type | Outcome | Notes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Priya Shah | Brightline Goods | Founder | DTC consumer brands | $2M-$10M revenue | Seed round, closed July 3 | outbound | 2026-09-07 | 2026-09-07 | 2026-09-14 | | | | | | open | sent the fulfillment checklist |

## Same structure as CSV

```
prospect_name,company,title,industry,size_band,trigger,channel,batch_date,touch_1_date,touch_2_date,touch_3_date,touch_4_date,touch_5_date,touch_6_date,touch_7_date,reply_date,reply_type,outcome,notes
Priya Shah,Brightline Goods,Founder,DTC consumer brands,$2M-$10M revenue,"Seed round, closed July 3",outbound,2026-09-07,2026-09-07,2026-09-14,,,,,,,,open,sent the fulfillment checklist
```

## Weekly maintenance rule

Before drafting a single new message each week, scan every open row for a touch date that falls due this week. Draft those follow-ups first, using references/follow-up-cadence.md. Only after due follow-ups are drafted and approved do you pull the next new batch.

The moment a reply date is filled in, set the outcome and stop drafting any further touch for that row, even if a touch was already queued for this week.

# Weekly content cadence

Two or three content slots a week, run alongside outreach, inside the same protected block of time. This is not a blogging exercise: every piece proves you can do the work you're pitching, and ends in a low-pressure way for a reader to reach out.

## Pick your slots

Before the first week, agree on the number of slots (2 or 3) and a fixed day and format for each. Common patterns:

- Monday: a short written post naming one specific problem your ICP has, with your own take on it.
- Wednesday: a one-page breakdown of a real (anonymized) situation you've worked through.
- Friday: a shorter observation or reaction to something happening in your ICP's industry that week.

Write the chosen slots here once and keep them fixed for at least a month before changing them:

- Slot 1: day ______, format ______________________
- Slot 2: day ______, format ______________________
- Slot 3 (optional): day ______, format ______________________

## The rule for every piece

Every piece is built around one specific problem the ICP has, not a general topic. State the problem plainly in the first two sentences. Show the actual thinking or the actual fix, not a teaser. End with a low-pressure call to action: a question, an offer to send more detail, or an invitation to reply, never a hard pitch.

## Before posting

- [ ] The piece names one specific, narrow problem, not a broad topic
- [ ] It shows real reasoning or a real example, not just an assertion
- [ ] It ends in a low-pressure call to action, not a pitch
- [ ] It has been shown to the person running the sequence for approval before it goes out; this cadence never posts on its own

## Sustainability check, run before committing to a cadence

Ask honestly: can this pace hold for eight straight weeks, including the weeks the delivery calendar is full? If the answer is no, cut to fewer slots or a shorter format before the first week, not after the second week is already missed. A cadence run at two slots a week for two months beats three slots a week that stops after two weeks.

# List-building: two methods for turning an ICP into a name list

Once the four ICP fields are set, the list gets built one of two ways: a straightforward method for a common ICP, or a custom method stitched together from a few sources for an ICP that doesn't fit a standard database. Pick one before starting, and use only one method per list so the batches stay consistent.

## Method 1: straightforward enrichment tool (use this when the ICP fits standard company and title filters)

Use this when the industry, role, and size band from the ICP worksheet can be searched directly as filters in an off-the-shelf contact database (for example, Apollo, Clay, ZoomInfo, or a similar B2B contact tool the user already has access to).

Steps:

1. Open the tool and filter by industry, exact title, and the company-size band from the ICP worksheet.
2. Add the trigger condition as a saved search or alert where the tool supports it (funding data, job postings, headcount changes). Where it doesn't, run the base filter first and check the trigger by hand for each name (step 4).
3. Export the raw list. Expect two to three times the target count before filtering, because a meaningful share will fail the trigger check.
4. Check every exported name against the trigger condition one at a time (a recent raise, a new hire in the target role, a job posting, an expansion). Drop any name that doesn't clear the trigger; a list padded with names that miss the trigger produces a lower reply rate later.
5. Stop once 100 to 250 verified names are on the list. Don't keep pulling past 250; a first list this size covers four to five weekly batches, and the reply data from batch one will shape the second list better than a bigger first pull would.

## Method 2: custom multi-source approach (use this when the ICP is narrow or unusual and no single database covers it well)

Use this when the industry is small, regional, or otherwise underrepresented in standard databases (for example, regional trucking companies, a specific franchise category, or a niche vertical).

Steps:

1. Pick two or three sources that each cover one part of the ICP: a trade association directory for the industry, a job board search for the trigger condition (open roles in the target department), and a search engine query for recent local news matching the trigger (an expansion, a new location, a leadership change).
2. Build a plain spreadsheet with one row per company: name, one contact at the target title (pulled from the company site's team page or a search-engine query for "[title] [company name]"), and the trigger evidence with its date and source link.
3. Fill in email addresses by checking the company's own site first (a team or contact page), then a standard email-pattern guess (first.last@domain) verified with an email-checking tool before it goes on the send list.
4. Work through sources in rotation, a fixed number of companies from each source per session, so the list doesn't end up dominated by whichever source was easiest to search first.
5. Stop at 100 to 250 verified rows, the same target as Method 1. A custom list this size usually takes several sessions; block the time for it the same way the weekly outreach block gets protected (Phase 1, step 6).

## Choosing between the two

Pick Method 1 whenever the ICP maps to filters a standard tool already offers; it's faster and the trigger data is often built in. Reach for Method 2 only when Method 1 turns up too few names to hit 100, or when the trigger condition isn't something any standard database tracks.

## Before the list goes into the tracker

- [ ] Every row has industry, role, size band, and trigger condition filled, matching the ICP worksheet exactly
- [ ] Every trigger has a date and, where possible, a source link
- [ ] Every email has been checked, not guessed and left unverified
- [ ] The list has 100 to 250 rows, not fewer and not padded past 250

# Reply-rate benchmarks: planning input only

These ranges exist to set a realistic expectation before the first batch goes out, not to promise an outcome to the user or to any client the user later pitches. Every number here is a range compiled from public cold-email and B2B outreach reporting, not a guarantee tied to this skill.

## Overall cold-outreach reply rate

Across general B2B cold email reporting, average reply rates cluster in the low single digits, roughly 3 to 4 percent. Well targeted campaigns, sent to a narrow ICP with a specific, real detail in each message, report reply rates reaching 10 percent or higher. The gap between the two numbers is exactly what the ICP-narrowing in Phase 1 and the five-part structure in step 10 are built to close.

## Where these numbers come from

- General cold-email reply-rate ranges: Instantly.ai's cold-email benchmark reporting.
- Reply-rate lift from narrow, well-targeted lists: LevelingUp's collected cold-email case studies.
- Referral and word-of-mouth reliance among small business owners (context for why a no-referral pipeline needs its own system): Entrepreneur.com / Alignable reporting.
- Broader marketing-for-consultants context: Consulting Success's marketing study.

Treat every figure above as a range from public reporting, not a promise. A specific campaign can land above or below any of these numbers depending on list quality, message specificity, and the ICP's own responsiveness that month.

## How to use this file

1. Phase 1, step 3: read the overall range to the user before the first batch goes out, so the first week's expectation is set by public data, not by hope. State it as a range, not a single number.
2. Never restate a number from this file as something the user's own list will definitely hit. The range calibrates expectations; only the tracker's logged replies show the user's real rate.
3. Once the user has four or five weeks of logged replies in references/pipeline-tracker.md, use their own logged rate instead of this file's range for every conversation after that; their own data always outranks a public benchmark once enough of it exists.

## Guardrail this file exists to satisfy

Never fabricate a reply rate, a client result, a testimonial, or a case-study number. Any reply-rate figure that reaches the user or a FAQ has to trace back to this file's sourced ranges above, or to the user's own logged tracker data, never to an invented or unnamed source.

# Referral partnership: finder's-fee proposal template

Use this once the user has at least one closed client and a real result to point to (Phase 3, step 13). The proposal goes to two or three adjacent, non-competing consultants or vendors who already serve the same ICP, not to random contacts; a partner who doesn't already sit in front of the same buyers has nothing to refer.

## Picking partners

Before drafting, help the user name two or three real people or firms who meet both of these:

- They serve the same ICP (industry, role, size band) but a different service, so there's no overlap or competition.
- The user has an existing relationship or a specific, real reason to reach out (a past project together, a shared client, a mutual introduction already made).

A cold pitch to a stranger asking for referrals is a different message than this one; don't use this template for that case.

## The proposal structure

1. **The connection.** One sentence naming the real, existing tie between the user and this person (a shared client, a past project, how they know each other).
2. **The result.** One sentence naming the specific closed-client result the user can now point to (the client's industry or situation, described generally enough to respect confidentiality, plus the concrete outcome).
3. **The ask.** A plain statement of what a partnership would look like: the user refers clients who need the partner's service, the partner refers clients who need the user's service, and a stated finder's fee or reciprocal-referral arrangement for either direction.
4. **The fee, stated as a number.** A specific percentage or flat fee, not "a fair cut" or "we'll work it out." Common structures: a flat percentage of the first project's value, or a fixed flat fee per closed referral. Confirm the number with the user before it goes in the draft; this skill never picks the number on the user's behalf.
5. **A low-pressure close.** An offer to talk it through on a short call, or to just start with one test referral each way before formalizing anything.

## Filled example

Subject: an idea for sending clients each other's way

Hi Daniel,

Since we worked together on the Renwick account last year, I've kept meaning to ask about this.

I just closed a fulfillment-process project for a DTC brand that grew from 40 to 200 orders a day in four months, the kind of client who's usually also looking for the bookkeeping help you do.

Would you be open to a simple referral arrangement? When I meet a client who needs a bookkeeper, I send them your way, and when you meet one who needs help with fulfillment process, you send them mine. I'd propose a 10 percent finder's fee on the first project's value in either direction, paid once the referred project closes.

Happy to hop on a fifteen-minute call to work out details, or we could just try it with the next referral each way and see how it goes.

Marcus

## Before sending

- [ ] The partner genuinely serves the same ICP with a different, non-competing service
- [ ] The result named in part 2 is a real, user-confirmed outcome, not a rounded-up or invented number
- [ ] The fee in part 4 is a specific number the user has confirmed, not a placeholder
- [ ] The draft is presented to the user for approval before it goes out, the same as any other outreach message in this skill

# Worked example: one batch, from list to logged follow-up

This is a single small batch worked all the way through, from three prospect rows on the list to a closed reply, so the shape of a real week is visible in one place. The names, companies, and details below are illustrative, built to match the ICP-criteria and outreach-templates examples elsewhere in this bundle, not any real company or person.

## Step 1: three rows pulled from the list (Phase 1, step 2 output)

| Prospect | Company | Title | Industry | Size band | Trigger |
|---|---|---|---|---|---|
| Priya Shah | Brightline Goods | Founder | DTC consumer brands | $2M-$10M revenue | Seed round, closed July 3 |
| Tomas Alvarez | Fieldstone Logistics | VP of Operations | Regional trucking | 50-300 employees | 3 ops job postings in 60 days |
| Renee Okafor | Marrow Studio | Founder | DTC consumer brands | $2M-$10M revenue | New head of ops hired in June |

## Step 2: three initial messages drafted (Phase 2, step 10, five-part structure)

**To Priya (matches the filled example in references/outreach-templates.md):**

Subject: the ops role you posted last week

Hi Priya, I noticed Brightline Goods closed a seed round in July and has three open operations roles up right now, including a Director of Fulfillment. That kind of hiring push usually means the current process is already past what one person can track by hand, especially once order volume doubles. I work with early-stage DTC brands on exactly that gap between headcount and process, before the new hires start. Would it be useful if I sent over the two-page checklist I use with brands in that exact spot, no call needed?

**To Tomas:**

Subject: the three roles Fieldstone posted this month

Hi Tomas, saw Fieldstone has three operations postings live right now, including a dispatch coordinator role. Usually that kind of hiring wave means the current routing process is stretched past what the team can track manually. I work with regional carriers closing exactly that kind of gap between headcount and process before the new hires land. Worth a fifteen-minute look at how two similar carriers handled the same stretch?

**To Renee:**

Subject: congrats on the new ops hire

Hi Renee, saw Marrow brought on a new head of ops in June, usually a sign the founder-run process needs a real system before it can hand off cleanly. I help DTC founders build that handoff so the new hire inherits a working process instead of tribal knowledge. Happy to send the short handoff checklist I use for this exact transition, no call needed.

## Step 3: batch approved, sent, logged (Phase 2, step 11)

All three messages were presented to the user in one pass. The user approved all three with no edits. Send dates were logged the same day in references/pipeline-tracker.md:

| Prospect | Batch date | Touch 1 |
|---|---|---|
| Priya Shah | 2026-09-07 | 2026-09-07 |
| Tomas Alvarez | 2026-09-07 | 2026-09-07 |
| Renee Okafor | 2026-09-07 | 2026-09-07 |

## Step 4: week two, follow-up touch due before the new batch (Phase 2, step 8)

By the following week, none of the three had replied. The tracker showed touch 2 due for all three (five to seven days after touch 1, per references/follow-up-cadence.md). Touch 2 drafts were written and approved before that week's new batch was pulled:

**Touch 2 to Priya:** Hi Priya, following up with the fulfillment checklist I mentioned, built for brands right around your order volume. Happy to send the full version if useful.

**Touch 2 to Tomas:** Hi Tomas, following up with a short breakdown of how two similar-sized carriers handled a hiring wave like this one. Happy to send it over if useful.

**Touch 2 to Renee:** Hi Renee, following up with the handoff checklist I mentioned, built for exactly this kind of ops transition. Happy to send it if useful.

Touch 2 dates were logged for all three: 2026-09-14.

## Step 5: a reply lands, the sequence stops (Phase 2, step 12)

Priya replied on 2026-09-16 asking for the checklist. Her row was updated the same day:

| Prospect | Touch 2 | Reply date | Reply type | Outcome |
|---|---|---|---|---|
| Priya Shah | 2026-09-14 | 2026-09-16 | interested | open, handed to active-client process |

No touch 3 was drafted for Priya. Tomas and Renee continued into touch 3 the following week, on schedule, since neither had replied.

## What this batch shows

- A batch this small (three prospects) is realistic for a short protected block; the shape scales the same way at twenty or forty.
- The due follow-up (step 4) was drafted and approved before the next new batch, matching the order in Phase 2, step 8 and the weekly-loop checklist.
- One reply stopped one sequence immediately, while the other two kept running on schedule; a reply from one prospect never pauses the batch for the rest.

Build one prospect list of 100 to 250 people who match a narrow ICP, then work a batch of 20 to 50 of them every week with a specific opening message, a short follow up sequence, and a proof of work piece of content published on the same cadence. The list and the weekly rhythm replace the referral network.

Because it was never protected time, it was leftover time, and delivery work always eats leftover time first. The fix is a fixed weekly block on the calendar for outreach and content that shrinks in a heavy week instead of getting bumped.

Public cold-email benchmarks put the average around 3 to 4 percent, with well targeted, narrow campaigns reaching 10 percent or higher (this skill's references/benchmarks.md carries the sourced ranges). Treat any range as planning input, not a promise; your own logged replies in the tracker will tell you more than any published range after a few weeks.

Plan for four to seven touches, each under eighty words, spaced about a week apart at first and then stretched out toward a month. Stop the sequence the moment a prospect replies, and mark it closed instead of sending the next scheduled touch.

Yes. Keep the weekly block running at a smaller size rather than dropping it to zero. Referrals are unpredictable by nature, and a pipeline that only restarts after the referrals dry up is back to feast or famine.

Aim for 100 to 250 prospects who match your ICP. That size covers roughly four to five weekly batches at 20 to 50 prospects each, enough runway to see real reply data before you need to build a second list.