the difference between your business ICP and your DM ICP
(spoiler: the second one is the one your setter actually needs)
Tasha opened a small boutique in West End Atlanta two years ago. Curated stuff. Pieces she handpicked at trade shows in New York and LA.
For her first season she ordered 50 dresses. Bought what the buying calendar told her was hot. Bohemian midi-length, neutral tones, sizes 4 through 14, weighted toward sizes 8 and 10 because “that’s the average size of an American woman.”
Six months later she still had 32 of those 50 dresses on the rack.
Sold out of the size 6 in week three. Sold out of the size 14 by month two. Except she only ordered two size 14s. Because “the data” said 8 and 10 were the popular sizes.
The women walking into Tasha’s boutique were not “the average American woman.”
They were Black women in Atlanta. Mostly 4-6 or 12-16. Mostly working corporate jobs Monday through Friday and church on Sunday. Mostly looking for pieces they could wear to brunch AND a board meeting AND their cousin’s wedding.
The buying calendar didn’t know any of that. The buying calendar averaged 50 million women into one woman who doesn’t exist anywhere except in a spreadsheet.
Tasha sold off the dead inventory at half price. Spent the next 3 months actually talking to her customers. Took notes. Looked at receipts. Studied who was buying what and what they were saying when they bought it.
Built what she now calls her “real customer file.”
By her second season her sell-through rate was 87%.
Same store. Same buyer. Same season. Different target customer.
What Tasha was originally doing is what most people do in their DMs.
Same opener. Same pitch. Same close. Sent to 200 prospects who are not the same person.
The 4 who happen to be a near-perfect fit reply. The 196 who needed a something different don’t.
The founder looks at the 4 replies and calls it a 2% reply rate. The 196 ignored messages get filed as “the algorithm” or “the market is slow.”
It’s not the algorithm. It’s not the market.
It’s the brochure problem. One message, four different types of people.
You spent May reading the autopsy of Reid’s $2K sequence. 7 messages. 12 diagnostic mistakes across 7 messages.
Today we fix the thing that comes BEFORE the message: the 60-second ICP classification.
→ Why your CRM ICP and your DM ICP are two different documents and what happens when you confuse them
→ The 5 specifics every DM ICP needs that your website ICP almost always misses
→ How to build the DM ICP in 90 minutes using prospects who’ve already bought from you
Let’s start with Tasha’s dress rack...
🏗️ Today’s Build | The 60-Second ICP Classification
Every prospect on your DM list is at one of two levels when your message hits their inbox.
The level isn’t about their company size. Their job title. Or their seniority.
It’s about where they are in the conversation that’s already happening in their own head.
1️⃣ LEVEL 1-2: Describing their situation.
Their last 3-10 LinkedIn posts sound like documentation. They’re naming what’s happening in their world without expressing frustration or naming a goal.
“Just hired our third account manager. Onboarding her this week.”
“Wrapped the Q2 reporting build. Two months of work. Worth it.”
“Spent Friday rebuilding how the team handles client onboarding. Lots to test.”
No goal stated. No frustration named. No identity claim. They’re documenting.
The conversation in their head is “here’s what we’re doing.” Not “here’s what I want.” Not “here’s what’s broken.”
✅ Send them a message that acknowledges what they actually said and asks how it’s going. They reply because the question is at their level.
❌ Send them a message about a goal they never named and they’ll ignore you. They weren’t asking. You projected.
2️⃣ LEVEL 3-4: Named a goal or frustration.
Their last 3-10 posts include language that signals what they want, what’s broken, or who they are. Active conversation already running in their head.
“Why is hiring senior designers taking 4 months in this market?”
“Bringing outbound back in-house after two agency disasters. Tired of paying for excuses.”
“Built our entire ops manual this quarter and it’s already obsolete. There has to be a better way.”
Goals, friction points, identity statements. The conversation in their head is “here’s what’s not working.” Or “here’s who I am.” Or “here’s what I want next.”
✅ Send them a message that references the specific thing they named and offers something useful (an insight, a small fix, a way of thinking about it). They reply because you spoke to where they already are.
❌ Send them a generic “hope your week is going well” check-in and they’ll ignore you. They wanted a real answer to the thing they were actively wrestling with. You gave them small talk.
3️⃣ The 60-second read.
Your setter (or VA, or you, depending on who runs the funnel) does this for every prospect before they touch the opener.
→ Step 1: Open the prospect’s LinkedIn profile.
→ Step 2: Read the most recent 3 posts. If they don’t post often, read the 3 most recent status updates or comments.
→ Step 3: Ask one question: Are they describing their situation, or have they named a goal or frustration?
→ Step 4: Tag the prospect. L1-2 or L3-4.
→ Step 5: Next prospect.
5 prospects = 5 minutes. 200 prospects = 200 minutes. That’s a setter’s morning.
The output isn’t a perfect score. The output is a folder of 200 prospects sorted into two buckets, each one routed to the opener built for THAT bucket.
What you’ve been calling “low reply rate” is actually a classification problem, not a copy problem.
Your sequence isn’t broken. It’s calibrated for one of the two levels and getting sent to both.
🔨 I’m running a free 7-day challenge
You’ll build a complete DM sequence from scratch, one message per day, scored against every diagnostic from this series.
A Notion workspace with daily build instructions. A Telegram group for feedback. The DM Sequence Grader to score every piece.
No course login. No replays. No upsell on Day 3.
Join the waitlist:
→ Join the free 7-Day Build Your Reply-Worthy Challenge Waitlist here
Here’s how to fix it in your own DMs:
1️⃣ Write your DM ICP from your last 10 closed clients, not from your strategy deck.
Pull the last 10 deals you closed. Open a doc. For each one, fill in:
→ trigger event (what happened in their world in the 90 days before they bought)
→ buying role (who actually signed), platform behavior (where they showed up)
→ internal language (what they called their problem), recent failure (what they’d just tried).
Find the patterns. 6 out of 10 had the same trigger? That’s your trigger.
7 of 10 used the same phrase to describe their problem? That’s your internal language.
The DM ICP is downstream from your actual buyers. Not upstream from your marketing strategy.
Build it from the data, not from the deck.
2️⃣ Carry the DM ICP doc in a single Notion page your setter pulls up every morning.
This isn’t a deliverable for a Q3 strategy off-site. It’s a working document that lives next to the inbox.
One page. Five sections. Updated whenever a new close reveals something the doc missed.
Your setter opens this doc before they open LinkedIn Sales Nav or the Manychat interface.
They check the prospects they’re about to message against the five criteria and triage.
5 out of 5 match? Top of the list. They get the most researched, most specific opener your team can write.
3-4 out of 5? Solid prospect. Standard opener, standard sequence.
1-2 out of 5? Still in the pipeline. Different opener. Lower research investment per prospect. But they don’t get skipped.
The match score tells your setter how much time to spend on the research, not whether to reach out at all.
3️⃣ Re-write your DM ICP every 90 days from the last quarter’s data.
Your DM ICP from January isn’t your DM ICP from April. The market shifts. Your offer shifts.
The trigger events that were hot last quarter cool off. The internal language migrates.
Re-pull the last 10 closes. Re-fill the five sections. Compare. Update the doc.
This takes 90 minutes per quarter. It’s the highest-leverage 90 minutes in your DM workflow.
The teams whose DM ICPs are 18 months old are the teams whose reply rates are dropping by quarter, and they don’t know why.
That’s it.
Here’s what you learned today:
→ Your website ICP and your DM ICP are two different documents. The website ICP fits on a slide. The DM ICP fits on a notecard.
→ Most teams write to the website ICP from their inbox, which is why their DMs feel generic. The DM ICP is built from real recent buyers, with five specifics: trigger event, buying role, platform behavior, internal language, recent failure.
→ The fix is pulling your last 10 closes, finding the patterns, writing the DM ICP on one page, and refreshing it every 90 days.
Start here:
Open a doc. Pull up the last 3 deals you closed.
For each one, write down: what was happening in their world 90 days before they bought, who actually signed, where they live online, what phrase they used to describe their problem, what they’d just tried that didn’t work.
If you can’t fill in those 5 fields from memory, you don’t know your DM ICP yet.
You know your CRM ICP.
Today’s mega-prompt builds your DM ICP from scratch.
Paste in details from your last 5-10 closed clients, and it pulls the 5 patterns: trigger event, buying role, platform behavior, internal language, and recent failure.
You get back a one-page DM ICP doc your setter can use tomorrow morning.
Paid members get to paste their closed client details and get back:
→ The trigger event pattern (what was happening in their world 90 days before they bought)
→ The buying role pattern (who actually signed, not who took the first call)
→ The platform behavior pattern (where they show up, how they engage)
→ The internal language pattern (the exact phrase they use to describe their problem)
→ The recent failure pattern (what they just tried that didn’t work)
→ A paste-ready one-page DM ICP doc formatted for Notion
→ The 60-second classification cheat sheet your setter runs before every first message
This isn’t the CRM ICP on your strategy deck. This is the inbox ICP built from the people who already paid you.
