Building an icp for outbound

A few years ago, I watched a team build an outbound list around “B2B SaaS companies with 50 to 500 employees.” The list looked reasonable. It also gave reps almost nothing to work with.

There was no reason to contact those accounts this month, no obvious person to start with, and no indication that the companies had the operating structure to buy. The team blamed the sequence when meetings did not appear. The real mistake was treating a market segment as an ideal customer profile.

An outbound ICP should help a rep make three decisions:

• Is this company capable of buying?
• Why might it care now?
• Who inside the company can create or approve a purchase?

If the profile cannot answer those questions, it belongs in a strategy deck, not in the workflow used to build a call list.

Start with evidence from your own accounts

The best source of an ICP is usually the last 20 to 40 closed won deals. Do not compare only industry, revenue, and employee count. Look at what happened before the purchase, who created momentum, which tools the customer already used, how long the deal took, and whether the account retained or expanded.

A sales engagement vendor might find that its best customers were not simply software companies. They were B2B software firms with 80 to 250 employees, a recently hired VP of Sales, several open SDR roles, and a CRM already used by a small sales team.

That combination matters. The new VP had inherited a pipeline target. The hiring plan showed that outbound was becoming a real function rather than a founder side project. The CRM indicated that the company had enough process to feel the cost of poor prospecting. Those details tell you both who to target and what situation to mention.

Closed lost deals are just as useful. Suppose five companies matched the industry and employee range but had no sales manager, no SDR team, and a founder still handling most sales conversations. The product may have fit on paper, but the companies lacked the operating conditions to adopt it. Put that pattern in the exclusion criteria.

For an early-stage company with no customer base, use discovery calls as provisional evidence. Favor prospects who described the problem without prompting, had already tried another solution, and could explain who would need to approve a change. Those details reveal pain, urgency, and buying mechanics even when there is no revenue history to analyze.

Separate fit from timing

Most ICPs describe what stays relatively stable: industry, company size, geography, business model, and technology. That is the fit layer.

Outbound also needs a timing layer. A company may be a good customer eventually and still be a poor prospect today. The timing layer identifies the event that changed the account's priorities.

For a compliance monitoring vendor, a 120-person fintech company becomes more interesting after hiring its first chief compliance officer and posting several risk and audit roles. The company was already a possible fit. The hiring pattern suggests that compliance work is becoming a formal operating responsibility, with new processes and deadlines likely to follow.

That gives the rep something better than “I work with fintech companies on compliance.” The message can refer to the new compliance function and ask how the team is handling the work created by that expansion.

Useful triggers tend to be observable events that create a mandate. A new executive may be expected to replace an inherited process. A cluster of job postings may show that a problem is becoming expensive. A CRM migration may create reporting or integration work. A new regulation, audit requirement, or enterprise customer may impose a deadline.

A funding announcement from four years ago is background. A Series B announced last month, followed by ten operations hires, is a reason to look closer. The trigger should explain why the account belongs near the top of the list, not serve as a token personalized sentence.

Map the people who can move the deal

A company-level profile is incomplete without a contact-level view. The person who feels the problem may not control the budget. The person who approves the spend may not understand the day-to-day cost of doing nothing.

Take a data infrastructure vendor selling to mid-market ecommerce companies. The VP of Data may own unreliable reporting. The director of engineering may assess implementation risk. Finance may approve the contract. Contacting only the VP of Data can create interest without creating a deal.

The SDR should know which role to approach first, who could champion the project, who controls the budget, and who might block it. That does not mean emailing every senior title in the account. It means giving each relevant person a reason to care about the same problem from their own perspective.

Titles need context. “Head of Operations” at a 40-person logistics company may be the buyer, the user, and the implementation owner. At a global manufacturer, the same title may have little influence over a regional technology purchase. The account's size, structure, and trigger determine what the title means.

My view is that teams get multi-threading wrong by treating it as a volume exercise. More contacts do not compensate for a weak account hypothesis. Start with the role closest to the trigger, then involve the economic buyer or technical approver when the conversation gives you a reason to do so.

Use technology and exclusions to narrow the list

Technographics are useful when the existing stack says something about the problem or the account's ability to adopt.

A sales analytics product may perform better with companies using Salesforce and a sales engagement platform than with companies still tracking deals in spreadsheets. The tools do not prove buying intent. They show that a sales operation exists and that managers may have enough process to notice reporting gaps.

The same data can reveal a poor fit. A company that completed a major rollout of a competing platform last quarter may match every other criterion and still be a bad near-term prospect.

Write those exclusions down. An anti-ICP might rule out companies below the minimum contract size, businesses without a relevant owner, industries with repeated implementation failures, or accounts already committed to a competing system. Without explicit exclusions, every list builder makes a different judgment about who belongs. The result is a list full of plausible accounts that consume rep time and rarely progress.

The profile should fit on one page and include enough detail to name real example accounts. It should specify the firmographic fit, required or disqualifying technologies, primary triggers, buying roles, the problem in customer language, and the opening position likely to make sense to the buyer.

If the document needs five pages to explain who belongs, the team probably has not made the hard decisions yet.

Test one narrow slice before scaling

An ICP is a hypothesis until prospects respond.

For example, a team selling sales hiring software could test 150 US-based B2B software companies with 50 to 200 employees, a VP of Sales hired within the last six months, and at least one open SDR role. The list should stay narrow enough that the message can refer to the same underlying situation. Do not mix fintech, agencies, and ecommerce companies just to make the sample larger.

Track positive replies and meetings, but read the replies. A segment can generate activity while still being wrong. If prospects respond because they misunderstood the offer, the targeting has not been validated. If they mention the hiring pressure or pipeline expectation that prompted the outreach, the account hypothesis is gaining support.

When the test performs poorly, check the layers before rewriting the copy. Did the accounts actually meet the criteria? Was the trigger recent and verifiable? Was the contact close to the problem or the decision? Did the message describe a situation the account could recognize?

Only then should you change the sequence.

The cold outreach motion should reflect the quality of the signal. An account with a fresh trigger may justify research and contact across two or three relevant roles. An account with firmographic fit but no active signal may belong in a lighter motion or outside the campaign entirely.

The profile also needs to live where work happens. Put account filters in the data workflow, define how reps verify a trigger, and show persona notes during research. Review new wins and losses on a fixed schedule, then update the criteria when the evidence changes. The broader ideal customer profile describes who the business serves well. The outbound version must tell the rep why this account, why this person, and why now.