Aim First: A Three-Stage System for Knowing Exactly Who You Sell To

A few weeks into your first outbound push, the replies don't come - and your hands reach for the one part of the machine you can see: the words. New subject line. Punchier opener. Another silent batch, and the creeping conclusion that cold email doesn't work for your business.

Nearly always, the words were never the problem. There are only three ways a cold email can die, and prose accounts for just one of them:

You reached the wrong person - a reader the problem simply doesn't touch, which is a list problem rather than a writing one. A targeting failure.
You reached the right person but framed your value in terms that don't map onto their own mental model of the problem. A positioning failure.
You got both right and phrased it clumsily. A copy failure - real, but minor.

The first two are decisive, invisible, and slow to fix - so nobody works on them; the copy can be edited in ninety seconds, so it absorbs all the effort while the real defect sits upstream. Buyers also climb checkpoints in order - notice, prioritize, prefer, commit, champion - and a message must meet its reader at the checkpoint they occupy. Whether yours can is settled before you type a word.

This guide is a system for settling it well, in three stages run in sequence: Stage 1 - Listen (learn what buyers will actually pay for), Stage 2 - Draw the Line (let that learning decide who's in and who's out), Stage 3 - Frame It (describe your value so the right people recognize themselves). Stages 2 and 3 refine each other forever; Stage 1 feeds both.

Stage 1 - Listen: extract the truth from real buyers

Start from a humbling premise: however certain you feel, your ideal customer's identity - and what that customer will actually spend money on - is not something you currently know. You hold a hypothesis. The market holds the answer. Conversation is the only bridge - and most customer conversations, run wrong, produce politeness instead of truth.

Step 1: Ask about their life, never about your idea

The discipline is Rob Fitzpatrick's. His slim book The Mom Test turned into the standard reference on interviewing customers because it attacks the honesty problem head-on. His standard: a good question is one that even your own mother, who wants you to feel good, could not answer with a comforting lie. "What do you think of my idea?" invites kindness, and kindness funds the wrong product.

The repair is total: stop discussing your idea and investigate their world instead. The way they get the job done now. What genuinely frustrates them. What they've already attempted. All of it in the past tense - because what someone did is fact, and what they would do is fiction.

Founders are uniquely bad at this, so the rule must be strict. You love what you've built, so you'll drift toward it, angle for praise, and mistake courtesy for genuine appetite. The moment you start pitching, your data corrupts - your subject is now responding to you, not describing reality. Keep the product in your pocket. And when a prospect finally opens up, the classic YC advice applies: your main job is to stop talking.

Swap your questions accordingly:

Retire this question	Run this one instead
"What do you make of this idea?"	"How does your team deal with [the problem] right now?"
"Would you pay for something that did X?"	"Tell me about the last time this bit you. What happened?"
"What would this be worth to you?"	"What are you using at the moment, and what's the cost in hours and money?"
"Would you use this?"	"What fixes have you already tried? Why didn't they stick?"
"Would this be helpful?"	"Is this actively painful, or just mildly annoying?"

The right-hand column can't be faked. It pulls facts, reveals your true competitor (more often a spreadsheet, a workaround, or plain inertia than another vendor), and gauges whether the pain is sharp enough that anyone would pay to end it.

Step 2: Listen for the moment their workaround breaks

Knowing how to ask is half the skill; knowing what you're listening for is the other half. Nobody has framed that half more crisply than Bob Moesta, co-creator of the Jobs-to-be-Done framework with Harvard's Clayton Christensen, with a hand in launching over 3,500 products since:

"People don't buy products. They hire them to make progress." - Bob Moesta, Jobs-to-be-Done

What triggers a new "hire" is what Moesta calls a struggling moment - the instant when the current coping mechanism breaks down hard enough to force a change. These interviews exist to mine for that instant. No title, no industry label, nothing else gets you as close to a plain reading of what someone will open a budget to fix.

The corollary lands hardest on technical founders: better does not make people switch. Buyers move because their present situation pushes them, not because your feature list pulls them. If your interviews keep surfacing the same genuine struggling moment, then message and pricing evidence have both arrived in a single stroke. If they surface only mild irritation, no downstream tactic will conjure demand the market doesn't contain.

Decision rule for every conversation

Score each interview against a three-part test. It counted only if it produced at least one of:

Facts - hard specifics on their current process and its price in hours or cash.
Commitment - something spent that can't be recalled - time, reputation, budget: a booked follow-up, an introduction, a pilot.
Advancement - movement to a defined next step.

"Sounds great, keep me posted" produces none of these - only the sensation of progress, the most dangerous thing a startup can collect. Treat compliments as a null result and move on.

Stage 2 - Draw the Line: let the market define your ideal customer

The interviews now feed your first hard output: an ideal customer profile. Two clarifications first, because the common framing sends people sideways.

Step 3: Separate the company filter from the person filter

Drawing on the investor field guides from Unusual Ventures and Race Capital: your ICP is a description of companies - stage, headcount, sector, stack: the structural facts that turn your problem into an urgent one for that organization. Personas cover the humans within those companies - the individual whose job the problem worsens and who has standing to act. Sequence matters, and the six phases we run it in put it first for the same reason: nail the account filter first, because the list is the campaign. A perfect message aimed at a persona employed only at firms with no reason ever to buy is wasted craft. Founders tend to obsess over the persona and skip the account question - exactly backwards.

The deeper point is about epistemics. An ICP isn't something you announce from your desk; it emerges from evidence - the interviews you just ran and the response data your first sends will generate. Open with a sharp hypothesis; let reality amend it. And build the refusal side deliberately: the accounts you turn down shape a working ICP as much as those you go after. A definition that fails to rule out most of the market isn't a definition - it's a hope.

Step 4: Cut the segment down until it feels reckless

The most expensive mistake in this discipline is breadth, and in the moment it registers not as error but as prudence - you're just keeping options open. Why limit ourselves to one niche when the product technically serves anyone? Because from the buyer's side, "fits anyone" reads as "built for nobody."

Consider the arithmetic. Search a database for "DevOps Engineer" and you get about 1,000,000 results - an undifferentiated fog. Layer on the constraints that genuinely define your fit - region, headcount band, sector, the stack they run - and that million shrinks to roughly 2,000 people: a market a small team can work through name by name. A million-person "market" isn't a market; it's an excuse to stay generic.

"Selling to everyone is how you sell to no one." - Belal Batrawy, #deathtofluff

For a small team the stakes exceed "focus is good." You will never outspend a bigger competitor. Your only affordable edge is relevance - being so unmistakably made for one tightly drawn buyer that your email could pass for a note written inside their world. Relevance scales with narrowness: tighter target, more specific message - and specificity decides everything now that machine-generated outreach has cheapened the generic. The calibration: if the amount you're excluding makes you slightly queasy, the line is probably right.

Step 5: Treat early customers as evidence, not a template

A counterintuitive check on everything above, courtesy of Geoffrey Moore. His Crossing the Chasm is still the definitive study of how new technology finds its market, and its central lesson: the enthusiasts who buy earliest decide in ways the cautious mainstream never will - and the gulf between those two populations has drowned many a company that projected one onto the other. Your earliest buyers are unusually tolerant and unusually desperate - unrepresentative by nature. Copy them as a mold and you may march confidently toward a customer type that doesn't exist in volume.

Study them instead for the underlying pattern - rarely the obvious demographic. A widely cited example: Sprig's real sweet spot turned out to be companies whose products had millions of users - an odd, almost inconvenient cut that emerged only from watching which early customers extracted the most value and stayed. Nobody named that segment in advance. Through your opening months, the task is to accumulate enough conversations and customers that such a pattern can show itself - then follow it instead of your original guess.

Step 6: Segment on circumstance, not category

The most modern upgrade to targeting belongs to Jordan Crawford - Blueprint's founder, and the loudest voice for running go-to-market like a data-science discipline. His case: slice the market on evidence of pain and present circumstance rather than on vertical or job title. A vertical label - a SIC code, "SaaS," "healthcare" - is a permanent attribute; all it establishes is that a company is out there. Whether they're feeling your problem this quarter, it cannot say. A circumstance can:

A company that just made its first Head of Sales hire.
A team that recently churned off the competitor you replace.
An organization that just closed funding and must now scale a function.
A business newly caught by a regulation.

Every one of these is acute, time-boxed pain - and pain with a clock on it supports a specificity of message that a permanent attribute never can. The method: reverse-engineer your best customers, find the non-obvious circumstance they shared that predicted fit, then hunt down everyone else in it today. Note where this leads: a circumstance is at bottom a signal - something recently and visibly changed, which answers when to reach out as well as whom - and organizing lead lists that way is where this philosophy becomes an operating system.

Stage 3 - Frame It: position so the right buyer recognizes themselves

Now for the stage technical founders skip most eagerly and regret most bitterly. Positioning - the frame around what your product is and whom it serves - precedes every message you will ever send. Copywriting cannot rescue a muddled value proposition: when a reader can't rapidly place you (what is this, what does it displace, why should I care?), the polished email dies of a defect that masquerades as a wording problem.

April Dunford is the essential guide here. Over 200 technology products have passed through her positioning practice, and her book Obviously Awesome converted positioning from instinct into method.

Video: Obviously Awesome: How to Nail Product Positioning - April Dunford · Traction Conf · 29 min

Dunford's core warning: position your product or the customer will do it for you, from whatever they see first - and their guess tends to cast you as generic, or simply wrong. Positioning gives a buyer just enough context to avoid a bad conclusion.

Step 7: Keep the investor story out of the inbox

You own two stories, whether you've noticed or not. The category story - "our product rewrites the rules for an entire industry" - is sweeping, investor-thrilling, and for exactly that reason the one you'll be tempted to lead with. The buyer story is narrow and concrete: you have this specific, costly problem, and we handle it better than the thing you're using now. A stranger scanning a cold email cares nothing for your category ambitions; what matters to them is a problem alive this week, and whether you outperform whatever they're limping along with.

The rule: buyer story in outreach; save the category story for investors and boardrooms. And get the real question straight - no buyer weighs you in a vacuum; they weigh you against their other options, and the front-runner is nearly always the status quo. Muddling through as usual is the rival you genuinely have to out-argue.

Step 8: Interrogate every feature until it confesses its value

Here's a tool you can run this afternoon. List your features - the easy part, and where most founder messaging halts, which is also why it fails. Then put each feature through a chain of one repeated question: and why would a buyer care? Keep pushing until you reach an outcome the business feels.

Feature: "We use AI." - and why would a buyer care? It auto-sorts inbound support tickets. - and why would a buyer care? The support team recovers the two hours a day they spend triaging. - and why would a buyer care? Value: they absorb current volume without the extra hire they were budgeting for.

Run the full list and value themes emerge; the themes, never the features, are what belong in outreach. Two refinements separate adequate from sharp:

Translating is on you, not the buyer. Founders bank on prospects drawing the line from feature to business outcome themselves. Prospects never do. The customer knows their pain; you know your solution; the bridge between the two expertises is yours to build, and skipping it means the value never lands.
Only differentiated value counts. If every alternative can claim the same benefit, it's table stakes and decides nothing. Every CRM tracks deals; saying so persuades no one. Lead exclusively with value the alternatives cannot claim.

Filter your value themes through that second test, keep what is both valuable and uniquely yours, and what remains is the backbone of every outbound message you draft. One last note: your best-fit customers are by definition the buyers to whom your differentiated value matters most - which means ICP and positioning aren't consecutive one-time steps but a pair that co-evolve. Sharpening one sharpens the other. Freeze the ICP first and graft messaging on afterward, and you get positioning that reads fine and loses anyway. Work them together, forever.

The self-diagnostic: you know you've got this wrong when…

…your best interviews ended in compliments. "Love it, keep me posted" with no facts, commitment, or next step means you pitched instead of listened. Null result.
…you can't name the workaround you replace. If you can't say whether your true rival is a spreadsheet, a manual routine, or sheer indifference, you haven't done Stage 1.
…your ICP wouldn't disqualify anyone you're excited about. A definition that admits every plausible prospect is a wish wearing a lanyard.
…your target list looks like a job title plus a country. A million-name segment is spray-and-pray in disguise. Add conditions until the number scares you a little.
…you're cloning your first three customers. Early buyers are clues to a pattern, not the pattern. Find the non-obvious circumstance they shared before scaling the lookalike search.
…your opening line would excite an investor. Category-reinvention language in a cold email is the tell that the buyer story hasn't been written yet.
…your email leads with a feature you're proud of. If you can't trace it, in one chain of "and why would a buyer care?", to an outcome your rivals can't also claim - it doesn't go in the message.

Fix whichever you recognize, in stage order - listening first, definition second, framing last - because each stage inherits the defects of the one before.