Module 6 — Knowing Who to Hunt: ICP, Segmentation & Territory

"Most reps don't have a pipeline problem. They have a TARGETING problem wearing a pipeline problem's coat. Firing the reps doesn't fix the targeting. I have seen this mistake made. I have billed the cleanup." — Dr. Tanya Vex, keynote, $4,200 a ticket, open bar at the after-party, somehow not entirely wrong about this one thing

---

Here is the original sin of go-to-market, the one nobody goddamn confesses at the board meeting because the board slide is so beautifully designed with gradient arrows pointing at a number with a "B" in it: you are selling to the wrong people, on purpose, because the wrong people are easier to find than the right ones. The right account requires you to think — to pull real data, build a real profile, make uncomfortable decisions about who you are NOT going to sell to, and defend those decisions when a sales rep stands in your doorway waving a "hot" lead that meets zero of your ICP criteria. The wrong account just needs a pulse, a corporate card, and a slot in your demo calendar.

So your reps chase pulses. They chase logos that will never expand, never renew, never refer anyone — accounts that churn in eleven months and leave a smear on your net revenue retention like a bug on a windshield, and then they leave a three-star Trustpilot review that says "not the right fit" while your CS team quietly deletes the Slack channel and pours one out. And you let them do this, because activity feels like progress, and the existential dread of an empty pipeline is louder than the dread of a bad one. A bad pipeline looks like a full pipeline, and you will not realize the fucking difference until Q3 when the renewals start collapsing and the board wants answers you don't have.

Targeting is the cheapest leverage in the entire revenue machine. Not the CRM, not the methodology, not the AI SDR swarm that is currently sending five thousand emails a day on your behalf to people who want to receive those emails roughly as much as they want an unexpected root canal. Get the who right and everything downstream — conversion, cycle time, win rate, retention, NRR, all of it — gets measurably better. Get it wrong and no amount of methodology, comp design, executive coaching, or automation will save you. You will just generate garbage faster, at greater expense, with more confidence. The machine will be loud and expensive and completely full of shit, and it will generate beautiful dashboards that track exactly how fast you're burning money on the wrong accounts.

THE JOB

This function answers three questions in order, and the order is not decorative:

1. Who are we for? (ICP — Ideal Customer Profile.)
2. How big is that, really? (TAM / SAM / SOM.)
3. Who hunts which slice of it? (Segmentation and territory carves.)

ICP is a company-level fit definition — the profile of an account that buys fast, stays long, pays well, expands over time, and refers others instead of becoming a cautionary tale in your CS team's 1:1. It is NOT a buyer persona. The persona is the human you talk to inside the firm — that's a methodology question, covered in Module 4. ICP is the firm itself. Confuse the two and your targeting goes soft, your ABM becomes spray-and-pray with LinkedIn targeting, and you end up doing enterprise outreach at a company that has never bought software in its life because their Champion fit your persona deck.

Get ICP right and you've drawn the fence around the hunting ground. Segmentation and territory are how you divide that ground among the wolves so they don't tear each other apart over the same damn deer.

THE PLAYBOOK

1. Define ICP across three layers — not just firmographics, you lazy cowards. Three layers.

Most teams stop at "mid-market SaaS, North America, series B or later" and call it an ICP. They write it on a whiteboard in blue marker. The CEO takes a photo of the whiteboard. The photo goes in a deck. Nobody ever looks at the actual whiteboard again, and the deck photo is the last time anyone mentions ICP until the churn spike hits in Q4 and suddenly everyone wants to "revisit the targeting" — meaning, they want to figure out whose fault this is. That is not an ICP. That is a vibe with coordinates and a misplaced sense of confidence. A real ICP has three layers, all built from actual data you pulled from actual systems, not aspirational brainstorming where everyone names the logos they want on their LinkedIn:

Firmographic — the basics you cannot live without:

• Industry and vertical (and sub-vertical when it matters)
• Employee count range and annual revenue band
• Geography and language
• Growth stage and funding level
• Business model (PLG vs. sales-led vs. enterprise procurement — they are fundamentally different buying processes)

Technographic — what's already in their stack, which tells you more than most firmographic signals ever will:

• If you integrate with Salesforce, accounts already on Salesforce are warmer before you say a single word.
• If you displace or replace a specific tool, accounts running that tool are already in pain. Pain is intent. That technographic signal is intent you can buy from a data vendor instead of sitting around hoping to stumble into it.
• Technographic data is the most criminally underused targeting signal in B2B GTM. Stop ignoring it — it's free money on the table and your competitors are stepping over it. Use it or hand the edge to someone who will.

Behavioral and contextual — the "why now" layer that turns a target into a priority:

• Recent funding round (cash on hand, mandate to grow, increased tolerance for new vendor relationships)
• Hiring into relevant roles (they're scaling the exact function your product serves — that is a signal flare)
• Leadership change (a new VP of Sales means a new vendor review; old tools are suddenly not safe)
• Public initiative your product directly serves
• Recent expansion or M&A activity that creates the exact pain you solve

Derive your ICP from your own won AND churned data, not from a whiteboard brainstorm where everyone gravitates toward the logos they want to namedrop at a conference. Pull your best customers — high ACV, fast time-to-close, strong retention, expansion revenue, referrals — and find what they share across all three layers. Then pull your worst churners — the accounts that left angry, that never got value, that your CS team dreaded every QBR with — and find what they share. Your ICP is the intersection of "looks like my winners" and "absolutely cannot look like my churners." If you bullshit the ICP to include accounts you want rather than accounts that genuinely fit, the churn data will tell the truth eventually. The churn data always tells the truth. It is a merciless, indifferent bastard of a teacher, and it will wait as long as it needs to.

2. Size the market honestly. TAM → SAM → SOM. Do not lie to yourself. The board will do that for you; you don't need to help them.

• TAM (Total Addressable Market): Everyone in the world who could theoretically buy something like what you sell, under some imaginable set of conditions that may or may not involve favorable legislation and a shift in global procurement behavior. The number a VP of Strategy puts on slide four with a "B" in it and the board nods at because "B" sounds promising. It is largely useless for operational planning. It is a fantasy, and like all fantasies it is most dangerous when you start hiring to it.

• SAM (Serviceable Addressable Market): The slice you can actually serve right now — right geography, right segment, right product fit, right compliance posture. You are not in Germany yet. Your enterprise product cannot serve fifty-person startups. You don't handle healthcare compliance. Strip those out. SAM is real, SAM is defensible, and SAM is the number you should actually be proud of if it's honest.

• SOM (Serviceable Obtainable Market): What you can realistically win in the planning period given your current headcount, capacity, competitive position, sales cycle length, and the cold truth that breaking into financial services will take eighteen months from now and several references you don't yet have. SOM is the number you build territories on. SOM is the number that determines how many reps you can actually afford to hire without destroying your CAC. When SOM and headcount don't match, something will give — and it will not be the laws of economics.

Run both of these methods and reconcile the gap between them:

• Top-down: Industry report → filter to ICP criteria → multiply by your ACV. Clean, fast, and usually optimistic in a way that feels reassuring until it doesn't.

• Bottom-up: Count actual accounts matching your ICP using your data provider. Multiply by your realistic win rate (not your best quarter's rate, not the rate when you had a hot market and a tailwind — your real rate, the trailing twelve-month number, including the months that hurt) and your average ACV. The bottom-up number will be smaller than you hoped. It will also be the honest one. Build your goddamn hiring plan to the honest one, not the fantasy. The fantasy is for the board deck. The honest number is for your operating plan.

SOM = (# accounts matching ICP in territory)
      × (realistic win rate — trailing 12 months, no wishful thinking)
      × (average ACV)

If a guru hands you only a top-down TAM with a "B" in it and zero bottom-up reconciliation, hold your wallet and back away while maintaining steady eye contact. Do not blink. Chip Brennan made this exact mistake at not one but three startups, with increasing confidence and decreasing board patience each time. Chip is now in "strategic advisory" — which is what you call consulting when you don't want to explain the last job and nobody is asking the right follow-up questions yet.

3. Build an account scoring model. Fit × Intent. Stop treating all accounts as equal because they are profoundly, stubbornly, measurably not.

A-fit accounts close faster, expand more, churn less, and write the case studies that bring in the next A-fit accounts. C-fit accounts wave money at you during discovery, sign the contract enthusiastically, and churn in month nine with the energy of someone who was always going to churn, while leaving a review on G2 that your competitors will screenshot and send to prospects. You must score accounts or you are allowing reps to treat every name in the database as equally worth their goddamn time, which is how you get four hundred activities logged against three opportunities that were never going to close, and a pipeline coverage number that looks healthy right up until it doesn't.

Fit score: How closely does an account match ICP? Weight your firmographic and technographic attributes. Sample weights that you will tune to your own data:

• Industry and vertical match: 30%
• Company size and revenue band: 25%
• Technographic match (stack fit, displacement opportunity): 25%
• Growth signal (recent funding, hiring velocity): 20%

Output: A / B / C / D tier. Automate this in your CRM if you can make it happen without a six-month engineering project. If you can't, make a scoring spreadsheet, call it provisional, and replace it with the real thing when you have the budget.

Intent/signal score: Is this account showing in-market behavior right now? Intent data providers (Bombora, G2 Buyer Intent, Demandbase, et al.) track content consumption across the web. Layer in your own first-party signals: website visits, content downloads, trial activity, product-led telemetry if you have it.

Priority = Fit × Intent. The matrix is simple and brutal and will tell you things your reps do not want to hear:

The C-fit/high-intent account is the most dangerous object in your pipeline. They love the demo. They respond to every email within hours. They want to sign fast, they'll waive the security review, they'll skip the pilot — they are flattering and enthusiastic and they will absolutely churn in month nine because the product was never built for their use case, never delivered the value they were promised, and now they are telling their network that your product "doesn't scale" — a review that will circulate in a Slack community where your future A-fit buyers are watching and forming permanent opinions about you. Discipline means saying no to the C-fit, even when they're waving a signed order form and a nice-looking logo. This is the hardest fucking thing you will ever ask a quota-carrying rep to do — to turn down money in the name of fit — and the reps who manage it anyway, who have enough long-term thinking to say no to a bad deal, are the ones worth keeping.

4. Segment by motion, not just by size. The revenue band is a proxy. The motion is the actual rule.

The SMB / Mid-Market / Enterprise tiers are not arbitrary lines — they represent fundamentally different sales motions that require different reps, different cycles, different comp structures, different management approaches, different tooling, and frankly different personalities. Treating them as the same motion just on different budget levels is how you ruin good salespeople by misplacing them.

A rep wired for SMB velocity — accustomed to closing four deals a week, living on the dopamine of fast feedback loops, measuring herself in "meetings booked today" — will die in enterprise's silence. Three months of nothing happening is not failure in enterprise. It is a Tuesday. She will panic, discount prematurely, try to accelerate timelines that structurally cannot be accelerated, and ultimately push the deal into a state where procurement resents her pushiness and the champion is quietly embarrassed. That is a preventable tragedy that happens because someone put the wrong goddamn rep on the wrong motion.

An enterprise rep who manages six named accounts with patience, political sophistication, and a long view will starve on SMB's volume math. He cannot work ninety accounts simultaneously. He doesn't want to. He wasn't hired to. He will work five of them extremely well and let eighty-five go dark, and then you'll have both a bad pipeline and a confused AE wondering why everyone seems upset. Match the rep to the motion or watch talented people fail in the wrong patch and blame themselves when the real failure was yours. (Module 2 covers this. Read it. Seriously, read it before you place your first enterprise rep in an SMB territory and then wonder why she quit.)

5. Carve the territory. Pick a model, own its tradeoffs, and stop pretending any model is perfect.

There is no perfect territory model. Anyone who tells you there is one is selling a framework and probably a course to go with it. Every model has tradeoffs. The job is to pick the model whose tradeoffs you can live with, execute it consistently, and stop re-litigating it every time someone has a bad quarter.

• Geographic: reps own regions, states, countries. Simple, fair-feeling, travel-efficient. Weak when account density is wildly uneven — a rep in Montana has the same word "territory" as a rep in the Bay Area, and one of them has twelve viable ICP-fit accounts and one has twelve hundred. You know exactly which one is going to miss quota, file an objection, and blame the carve. You also know that rep is not entirely wrong, which is the most irritating possible outcome.

• Vertical/industry: reps own sectors instead of geography. Deep expertise, sharper messaging, the ability to name-drop customers within a vertical on the first call. Best for multi-vertical products with genuine differentiation by industry. Risk: verticals are different sizes and those size differences will generate resentment, appeals, and spreadsheet arguments by Q2 at the latest.

• Named accounts: each rep owns a fixed, curated list. Maximum focus, deep account penetration, classic for enterprise ABM. Risk: the politics over who gets the whale accounts will consume three weeks of your time every year and produce at least one awkward conversation with a rep who believes they deserved better than what they got.

• Hybrid (the real world, the actual answer): Enterprise = named accounts, assigned deliberately with logic you can defend. Mid-market = vertical pods with geographic tiebreakers. SMB = round-robin inbound with geo routing. This is messy and it also works. Accept the mess.

RULE No. 6: There is no carve that makes every rep feel fairly treated. Every rep thinks they got the shit end. Accept this. Stop chasing perfect fairness — it does not exist and the pursuit of it will cost you more leadership hours than it is worth. Chase coverage instead: every ICP-fit account in your SOM being actively worked by someone whose motion matches the account's segment. Coverage wins deals and drives revenue. Perceived fairness is a feeling. Feelings don't show up in the forecast.

6. Balance fairness vs. optimal coverage — and when in doubt, cover the market.

Every territory carve fights two forces: fairness (reps must believe they have a credible shot at their number) and coverage (every ICP-fit account actually gets worked). You will not reconcile these forces perfectly. Aim for defensible — roughly equal total opportunity value per rep (account count × average potential ACV), not equal account count and not equal headcount. A rep with 200 A-fit accounts and a rep with 200 C-fit accounts are not in equivalent territories. They just have the same number on a spreadsheet.

Then audit the whitespace. Within your SOM, which ICP-fit accounts have NO active opportunity, NO recent outreach, NO contact logged in the last ninety days? That is un-worked SOM sitting in your CRM right now, already paid for by your data provider subscription, waiting to be touched. The fastest pipeline you'll generate this quarter is not a new tool, not a new campaign, not a $50,000 conference sponsorship. It is the forty A-fit accounts already in your system that nobody has touched since Q3 because they got deprioritized during a bad month and then quietly forgotten while everybody chased new logos and ignored the gold sitting in the existing list. Map the damn whitespace. Assign it. Work it before you buy anything new.

7. The data you need — and where it goes to die if you don't take care of it.

ICP and territory operations run on: firmographics, technographics, intent signals, accurate account hierarchies, and clean lead-to-account matching. That last one will hurt you if you ignore it. Without lead-to-account matching, inbound leads from your named enterprise accounts get round-robined to the wrong rep, the territory model becomes a fiction, and your enterprise AE finds out three weeks later that a VP from her target account has been receiving automated sequences from an SDR who doesn't know the account exists. That relationship is already weird. It takes work to un-weird it, and sometimes it never un-weirds.

The full data layer — the warehouse, the enrichment pipeline, the matching logic — is Module 19. Read it; it will ruin several of your current assumptions. Until then: know that the best ICP model ever designed is only as good as the data powering it, and bad data doesn't announce itself with a warning label or a helpful Slack message. It just makes your scoring model confidently, precisely, and expensively wrong while everyone looks at the tiers and nods like they mean something.

HOW IT GOES TO HELL

The VP of Vibes defines ICP as "anyone with a budget and an interest in growing revenue." No tiers. No scoring. No tradeoffs acknowledged. Reps chase the loudest tire-kicker on the demo calendar because "they seem really excited." Win rates crater. Churn spikes at twelve months like clockwork, every single quarter, and he blames the product, then the CS team, then "market headwinds," then "the competitive landscape shifted," then eventually the CRM because he saw a dashboard he didn't understand and decided it was broken. After that he starts a podcast about authentic leadership and charges $2,500 per coaching session. The churn rate is not discussed on the podcast.

The Founder Who's Still the Best Salesperson insists "we sell to everyone." Every time someone tries to propose an ICP he says "but what about" and names a logo in a completely different segment and a different industry and sometimes a different country. He is not technically wrong that they could probably close that logo under the right conditions. He is spectacularly wrong about what chasing that logo costs in cycle time, churn risk, CS bandwidth, and organizational confusion when half the team is building for enterprise and half is building for SMB and both halves think they're right. Refusing to pick an ICP isn't ambition — it's cowardice wearing a growth-mindset disguise. A market of everyone is a market of no one, and your messaging becomes paste so beige, so safe, so inoffensive that it could be approved by seventeen corporate lawyers and read aloud at a shareholders meeting and still mean absolutely nothing to a real prospect with a real problem.

The Brilliant Jerk hoards the whitespace. He "owns" 400 named accounts and actively works nine of them with genuine skill and genuine results. The other 391 sit in the CRM under his name, accumulating digital dust, generating zero pipeline, locked out from any other rep because surrendering accounts feels like losing a personal asset, and he negotiated a wide territory when he signed his offer letter and will fight any revision with the focused, furious energy of a man whose entire identity is wrapped up in account ownership. This is a real category of waste and it costs you real pipeline. Cap account loads. Build a clear definition of "active" — a contact event in the last sixty days, an open opportunity, a documented next step with a date attached. Anything outside that definition at the 90-day mark is eligible for reassignment and you should say so in writing before he starts and not after the first territory review when it becomes a heated fucking argument about what he was promised.

The RevOps Martyr builds a beautiful, correct scoring model — weighted firmographic attributes, technographic signals, A/B/C/D output tiers, clean logic documented in the CRM — and zero people actually use it, because Sales leadership never trusted the input data, Marketing is routing leads on its own secret scoring that lives in a Google Sheet that updates only when someone remembers to update it, and the SDR team is working whatever list they were handed by the VP of Vibes last Tuesday. The model is correct, documented, and completely ignored, which is honestly worse than wrong and ignored — at least with wrong and ignored you have something to fix. Correct and ignored is a referendum on your organization's dysfunction and no scoring model can fix that. You have to fix the alignment first, which means getting Sales and Marketing in a room together to agree on what a good account looks like, which will feel like a hostage negotiation, because it is.

The annual reshuffle bloodbath: territories redrawn by committee every January via competing spreadsheets, ripping existing relationships mid-deal, every single rep believing they personally received the worst patch in the history of territory carves because their previous territory was specifically designed to screw them, three of them threatening to quit, one of them actually quitting and taking two good accounts with them because there was no coverage transition plan, and the whole catastrophe consuming six weeks of leadership bandwidth during the four-week window when you most need everyone focused on pipeline generation for Q1. Stop this. Carve thoughtfully, carve deliberately, with data. Grandfather in-flight deals with explicit rules about who owns the close. Give reps ninety days of visibility before changes take effect. Stop treating the territory carve like a hostage situation where you are simultaneously the negotiator, the hostage, and the one who brought the rope.

FIELD RULES

RULE No. 6-A: Derive ICP from your won AND your churned data, without exception. Both populations have a profile. Your worst churners are the most brutally honest teachers you have in the business, and they already paid for the lesson. Learn from the winners. Learn harder from the failures.

RULE No. 6-B: Fit beats intent, every single time. A C-fit account that desperately wants to buy right now is a future churn statistic wearing a promising Q2 costume. Say no. Say it with a referral to a better-fit vendor, say it with genuine respect for their problem, but say the damn no. Sell discipline, not desperation.

RULE No. 6-C: Segment by MOTION, not just revenue band. The size is a proxy. The motion is the real rule. Match the rep to the sales cycle or watch talented people fail in the wrong patch for reasons they'll never fully understand.

RULE No. 6-D: Balance opportunity value, not headcount. Fair-feeling and optimal are different gods and they will conflict. When they conflict, serve coverage — the market is not interested in your feelings about equity.

RULE No. 6-E: Whitespace in your CRM is the cheapest pipeline you own. It already matches your ICP. It's already in your system. Nobody has touched it in ninety days. Work it before you buy another tool or run another campaign.

RULE No. 6-F: No lead-to-account matching, no real territory. Without it, your territory model is a beautifully formatted map of a country that routes its own mail to random addresses and then acts surprised about the delivery failures.

---

WARNING — THINGS THAT CORRUPT THE HUNT:

• An ICP defined by committee consensus as "who we'd love to sell to" (not who actually buys and stays)
• A TAM built top-down with no bottom-up reconciliation — the "B" is a lie and the hiring plan built on it will be painful
• C-fit accounts in active Stage 3 because "they reached out and they're very enthusiastic"
• Named account lists last refreshed twelve months ago with rep who has since left the company
• Whitespace rotting in a single rep's territory for four months, zero activity, no accountability
• A territory carve that exists only in an Excel file that lives on one person's laptop

---

From the field: I once watched a company set a TAM with a "B" in it, present it to a board that clapped, hire sixty reps against it, set quotas to it, and then discover in month eleven that the actual SOM — the accounts that fit, had budget, could be reached in the current sales cycle with available headcount — was the rough size of a mid-sized parking lot on a Tuesday afternoon in January. The reps were not lazy. They were not incompetent. They were pointed at a fog bank and told to hunt, and they hunted the fog with great energy and beautiful tools, and the fog did not sign the paperwork. The market was not the problem. The map was the problem. The map was a fantasy in gradient blue with a "B" in the title and a confidence that wasn't earned. Expense the data provider. Run the goddamn bottom-up. Burn the TAM slide before the board sees it and mistakes it for a business plan and starts making compensation decisions based on it.