How to figure out your first ICP before you waste a year selling to everyone

The short version. Figure out your first ICP by choosing the account where a specific buyer has visible pain with budget reach while your product creates a before-after story they would repeat to a boss. Start narrow. A broad market feels safe because every logo looks possible, yet it hides the pattern that makes founder-led sales teach you something useful before the calendar eats the year.

Before you start:

• Your first ICP is an account filter, so write the buyer title, trigger, exclusion, and current workaround in concrete language.
• Use buyer words from replies, call notes, lost-deal emails, and event conversations before another market map.
• Disqualification protects the calendar from polite meetings that cannot teach a repeatable sales pattern.
• A narrow outbound test beats broad activity when every reply ties back to the same account hypothesis.

Step 1: ask which beachhead has pain and state what done looks like?

Start smaller. Your first ICP should name the account where a specific buyer has visible pain, can reach budget, feels deadline pressure, and already owns the current workaround. General market labels like software companies give you false comfort, because they turn every sales call into a fresh research project with no compounding.

Salesforce's customer profile guide frames profiles around attributes a team can understand and reach, while Close's ICP guide treats fit as a practical sales filter. Use that split. The account is the unit of focus; the person is the route into the building, with LinkedIn, email, event chats, and a warm introduction acting as doors rather than definitions.

Done looks like a sentence your CRM can hold: "Series A security teams in the US that just hired a compliance lead and need audit evidence inside their existing workflow." If a cousin category also sounds plausible, park it in a later tab. The beachhead earns focus only when the pain words repeat without your coaching.

• Pain visible in buyer words
• Budget owner can be named
• Trigger has a date
• Use case repeats across accounts

Step 2: write the account hypothesis and state what done looks like

Write it down. A fuzzy ICP lives in the founder's head until a spreadsheet forces nouns onto the page: industry, company stage, buyer title, urgent trigger, current workaround, and tool environment. The page matters because your memory will protect exceptions that the market has already rejected.

Keep the hypothesis blunt enough for an intern to sort Apollo rows without asking you for strategy advice. If the row needs a debate, the field is soft. "Uses Salesforce" beats "mature revenue motion" because a tool is visible, searchable, owned by a team, and less likely to flatter your imagination.

Done looks like a living account sheet with inclusion fields and exclusion fields. The sheet should reject accounts as quickly as it accepts them, because early selling needs contrast more than coverage. I like a column for "why now" because timing exposes the difference between curiosity and pipeline.

• Account type
• Buyer title
• Trigger phrase
• Visible exclusion

Step 3: pull evidence from conversations and state what done looks like

Use live language. The cheapest ICP research usually sits in call notes, LinkedIn replies, demo objections, lost-deal emails, and Slack messages from people who almost bought during messy evaluation cycles. Scrape those nouns before you open another market map, because buyers reveal categories through complaints long before they reward your positioning deck.

Favor repeated sentences over compliments. A founder can collect praise from every corner of the market and still have no beachhead if each buyer wants a different workflow, security promise, procurement path, or integration. The useful phrase has a business consequence attached, such as "my team misses renewals" or "audit week breaks our spreadsheet."

Done looks like a small evidence file with buyer words copied verbatim and tagged by account type. No theater. When the same trigger shows up in different mouths inside the same account profile, the ICP has earned another week of focus. The file should include the account source so a loud anecdote cannot masquerade as a pattern.

• Exact buyer phrase
• Current workaround
• Business consequence
• Buying blocker

Step 4: score fit with disqualification and state what done looks like

Add a kill switch. Early ICP work improves when the founder writes the account types that deserve a quick no, even if the logo is shiny and the intro came from a friend. The exclusion list protects the calendar from polite meetings that create deck edits rather than revenue signals.

Score accounts on evidence you can see before a call. Website copy, hiring pages, funding announcements, job posts, security pages, and conference agendas usually tell you whether the pain has public shape. A hidden pain can still be real, but hidden pain makes cold outbound behave like prayer.

Done looks like a redline rule inside the CRM. If an account lacks the trigger, buyer access, visible workflow, or operational wound, it sits outside the first ICP until evidence changes. This rule feels harsh for a founder who needs pipeline, yet it creates the clean signal that general selling keeps smearing.

• Hard yes
• Hard no
• Needs evidence
• Parked for later

Step 5: run a narrow outbound test and state what done looks like

Keep the test boring. Use the same account profile, same buyer title, same pain statement, same offer, and same proof point long enough to see whether the market pushes back in a pattern. Changing every variable after each reply makes the founder feel responsive while the data turns to fog.

Outbound is useful here because it produces refusals from a defined surface area. A LinkedIn comment, a cold email, an event conversation, or a founder intro can all count if the account fits the sheet and the response teaches the same lesson. Random interest from a nearby segment goes into the parking lot rather than the roadmap.

Done looks like a test log where every reply is tied to the hypothesis it challenged. The goal is a sharper account definition, a cleaner pain sentence, a visible rejection rule, and a repeatable reason to believe the buyer will leave the current workaround for your product.

• Same account profile
• Same buyer title
• Same pain line
• Same offer

What goes wrong

Everyone feels reachable. That feeling ruins the first ICP because a founder begins to confuse access with fit, especially when investors, friends, former coworkers, and operators keep opening doors into unrelated accounts. A warm intro can be a trap if the pain belongs to a category you cannot serve repeatedly.

Another failure is borrowing an ICP from a competitor with a different product surface, price, sales motion, and trust level. Copying their logos gives you a costume. In the US market, a founder building from India also has to price in time zones, credibility gaps, procurement habits, and the lack of hallway proximity.

The quiet failure is mistaking persona work for ICP work. A VP of Sales persona leaves open which account has the urgent wound, which system holds the data, who blocks security review, or what budget gets raided. Buyer psychology matters after the account pattern has teeth in real pipeline.

• Pretty logos
• Friendly intros
• Persona decks
• Vague verticals

Step 6: turn the ICP into operating rules and state what done looks like

Make it operational. Your ICP should change the accounts you load, the pages you write, the events you attend, the partners you court, and the objections you prepare for discovery. If the ICP only lives in a slide, the business will keep selling to whoever answered yesterday.

Put the profile into the CRM as fields a seller can filter. Put the same language into the homepage, LinkedIn posts, outbound snippets, event follow-up, and AEO answers so the market hears the same wound from multiple surfaces. Consistency makes pattern recognition faster because each channel stops inventing its own buyer.

Done looks like a weekly operating rule: accounts outside the profile need written evidence before they enter active pipeline. The ICP can evolve when the market teaches you, but it should never drift because the founder got bored or a polite prospect praised the demo during a quiet week.

• CRM fields reflect fit
• Content names the wound
• Outbound uses the trigger
• Events match the beachhead

Common questions

How do I figure out my ICP as an early-stage startup?

Start with the account where the pain is visible before your call and the buyer has a reason to move now. Write a hypothesis, gather buyer language, reject weak-fit accounts, then run a narrow outbound test. Your first ICP is a working filter for sales time, content, product proof, and event choices.

What should an ICP include?

An ICP should include account type, buyer title, urgent trigger, current workaround, buying blocker, and visible exclusion. Keep each field observable before a call when possible. A phrase like "needs SOC evidence for enterprise deals" gives sales a sharper filter than a soft phrase like "quality-conscious buyer."

How is ICP different from a buyer persona?

An ICP describes the account worth pursuing in a product category at a specific moment. A persona describes the person inside that account and how they evaluate change. Early founders should anchor on the account first, because account pain decides whether the sales motion repeats.

When should I change my ICP?

Change the ICP when repeated market evidence shows the same account profile, same pain, same workaround, and same blocker with stronger pull elsewhere. A loud prospect from another segment should enter a parking lot until the pattern repeats. Drift is expensive when every sales call is still teaching the company.