ICP and wedge

skills/icp-and-wedge/SKILL.mdView on GitHub

Goal: one specific segment the company goes after first, written down on one page, with the rules that tell a salesperson or an agent whether a lead fits. The riches are in the niches. Pick one, nail it, then expand.

This is the best informed guess, not a research project. Write it in a session or two, then go test it in the market. It gets revised when real conversations say so.

Before you start

This skill works on its own. If the person has their own marketing brain folder (the one they set up with the marketing-brain skill, or any folder of company notes they point you to), read the company, offers, competitors and goals files in it first. If they do not have one, do not look for one and do not assume one. Ask for three things only: what they sell in one sentence, who has bought it so far, and what revenue target the next 12 months has to hit. Do not start with an interview longer than that.

Tools this skill uses when the person has them. None are required; a spreadsheet works for all of it.

Two situations change how you run this:

• No customers yet, or under 10. Run the full scoring in step 3. Evidence is thin, so weight access and clarity of pain.
• Customers exist. Start from the wins. The pattern in the best 5 to 10 customers (fastest to buy, fastest to value, happiest, most likely to refer) is the first candidate segment. Scoring confirms or challenges it.

Steps

1. Gather the evidence. List every customer, won deal and lost deal you can find in the person’s notes, a CRM export (HubSpot or Pipedrive), or what they tell you. For each: industry, size, role of the buyer, how they came in, how long it took to close, and whether they are happy. Note where inbound interest is already coming from. If they record calls (Fathom, Granola), ask for the last five to ten customer calls and pull the words customers use for the problem; the call-mining skill does this properly if they want it done at scale.

2. List candidate segments. Three to six, no more. A segment is specific enough to put in a search: industry plus company size plus the role you sell to plus geography plus the situation they are in. “Mid-size companies” is not a segment. “Commercial HVAC contractors in the US Midwest with 20 to 100 field technicians who bid on new construction” is.

3. Score each segment 1 to 3 on seven questions. Build the table and fill it with the person, out loud.

   Question	3	1
   Is the pain clear and named?	They describe it unprompted in their own words	You have to explain they have a problem
   Do they already pay to solve it?	A tool, a vendor, a person or a workaround with a cost	Nothing; it is tolerated
   How short is the sales cycle?	Weeks	Two quarters or more
   How many people decide?	One or two	A committee with procurement, IT and legal
   Are there enough of them?	Hundreds or thousands you can name	A few dozen
   Do you have access or proof already?	Customers, referrals or a network in this segment	None
   How fast do they get value?	Days after buying	Months of setup

   Add the scores. Then apply two overrides before you pick: any segment that scores 1 on pain is out whatever its total, and a segment where the company already has traction gets the tie-break. The top scorer is the wedge. The second is the backup, kept on the list, not worked.

4. Write the wedge in one sentence: “We are going after [segment] because [the two or three scores that decided it].” Public and private can differ. Position publicly around the wedge so the website, outreach and content are specific. Behind the scenes the company can still say yes to a good deal outside it; that is the founder’s call, deal by deal, and it is not a reason to widen the ICP.

5. Build the ICP card. One page, these headings, filled with specifics:

   • Firmographics: industry, revenue band, employee band, and the operating measure that actually predicts fit for this business (number of users in the buying function, number of locations, units of volume, seats, trucks). Give a range, not a checkbox.
   • Geography.
   • Situation and triggers: what is true about them right now that makes this the moment (new hire in the role, a lost bid, a regulation, a funding round, a product they just outgrew).
   • Disqualifiers: the two or three things that mean no, every time.
   • Stretch: who sits just outside the card that the company would still take, and on what terms (for example, larger companies start with one or two seats and expand).
   • Tiers if the company serves more than one size: what each tier gets. Do not disqualify a company for being big or small; route it to the right offer.

6. Map the buying committee for the wedge. One row per role, even when one person holds several roles in a small company (say so). Columns: role, title it usually has, what they care about, what they are afraid of, what they need to hear, whether they can say yes alone. Typical roles: the champion who feels the pain, the economic buyer who signs, the daily user, and the blocker (IT, legal, finance, an incumbent vendor). This table feeds positioning-doc directly.

7. Write qualification rules. Three to five questions a salesperson or an agent can answer in under five minutes from the company website, LinkedIn and one call, plus the source for each answer. Every lead gets one of three labels with a one-line reason: fits, needs review, does not fit. Nobody gets dropped silently. Say which data points enrichment tools (Clay, Apollo, LinkedIn) get wrong for this business and where a human looks at the website instead. If they use HubSpot or Pipedrive, suggest two custom properties, ICP tier and ICP reason, so the label travels with the record.

8. Check the numbers. Count roughly how many companies fit the card in the geography. Use Apollo, Clay or Sales Navigator filters for a quick count, an association member list if one exists, or Google Maps for local segments. Then run the goal math backwards: closes needed this year, divided by close rate, divided by meeting rate, equals conversations needed. If the segment is too small to supply that many conversations, widen one dimension (usually geography) before widening the industry.

9. Set the test. What the next 30 to 60 days have to show for the wedge to hold: for example 20 conversations in the segment, 5 of them turning into a pilot, and the pain described in the same words three times. Put a review date on it. If the test fails, move to the backup segment; do not add the backup on top.

Rules

• Never invent customers, deal data or market counts. A blank is a question for the person.
• One wedge at a time in public. Two or three segments can be tested in outreach, but the website and content speak to one.
• An ICP is a range and a situation, not a checkbox. If the CRM has a yes-no ICP field nobody uses, replace it with the tier and the reason.
• The ICP belongs to sales, marketing and product together. Show it to whoever runs sales before calling it done.
• Deals outside the ICP are allowed, flagged and watched. If the company keeps winning outside the card, the card is wrong; revisit it.
• Plain language. A new salesperson should read the card on day one and know who to call.

Output

Return, in this order: the wedge sentence, the ICP card, the scoring table, the buying committee table, the qualification rules, the market count and goal math, and the 30 to 60 day test with its review date. Offer the scoring table and tracker as a CSV or Google Sheet.

If the person has their own marketing brain folder, save to it:

• 1-context/ideal-customer.md: the ICP card, buying committee and qualification rules (replace the draft, keep a line on what changed)
• 3-strategy/positioning.md: the wedge sentence at the top, as the “who it is for” answer
• 5-current/priorities.md: the test and review date
• 5-current/decisions-log.md: one line, the date, the wedge chosen and the reason

If they do not have one, return the ICP card as a standalone file they can keep, and mention once that the marketing-brain skill gives every later skill this context without re-explaining it. Do not set one up unless they ask.

Next: run positioning-doc with this card open. outbound-list-and-sequence and crm-setup use the qualification rules as written here.