Imagine two companies that look identical in a prospecting database.

They operate in the same industry, have a similar number of employees, use the same CRM, sell through phone calls, and work in the same geography. Both match the ideal customer profile.

Inside the first company, call volume has doubled. Missed follow-ups are visible. A new sales manager owns the problem. The team recently missed a target, and leadership wants a pilot running this quarter. The manager can provide call samples, show the current CRM process, and define what an improved outcome would look like.

Inside the second company, calls are occasional. No one measures follow-up quality. CRM is mostly an administrative archive. The problem is interesting but has no owner, trigger, deadline, or budget logic.

A static ICP places the accounts next to each other. A first-customer motion should not.

Firmographics tell us what kind of company we are looking at. They do not tell us whether a specific workflow has become painful, owned, measurable, and ready to change.

For early products, pilots, and new offers, the ICP needs a second layer: workflow state.

What A Traditional ICP Still Does Well

Industry, company size, geography, business model, technology, and role are useful filters. They reduce the market to a plausible set of accounts and help a team understand where a problem could exist.

For example, a product built around phone-sales intelligence may reasonably prioritize teams with meaningful call volume, managers who own conversion or service quality, and a CRM or operating record that the workflow can improve. A legal operations product may prioritize firms with high document and deadline volume. A financing offer may require businesses with recurring revenue.

Those filters matter. The mistake is asking them to do more than they can.

An industry code cannot show whether the problem is active. Employee count cannot show who owns it. A technology signal cannot show whether the company trusts the system enough to use it as an operating record. A job title cannot show whether the person has authority to change the workflow.

Static attributes describe potential fit. First-customer work depends on current readiness.

Define The ICP As A Workflow State

A workflow-state ICP describes the conditions under which a problem becomes buyable.

The unit of analysis is no longer only the company. It is the interaction between a company, a recurring workflow, an accountable person, and a moment in time.

The state can be described through seven dimensions:

1. repetition;
2. consequence;
3. owner;
4. trigger;
5. budget logic;
6. workflow readiness;
7. commitment behavior.

Together, these dimensions answer a practical question: is this account merely capable of having the problem, or is it ready to do something about it?

1. Repetition: Does The Problem Happen Often Enough?

Frequency creates both pain and evidence.

A workflow that occurs twice a year may be frustrating but difficult to validate through a short pilot. A workflow that occurs daily creates visible cost, repeated user behavior, enough examples for review, and a clearer path to measurement.

Frequency should be described in operating terms. How many calls, cases, documents, approvals, reports, or follow-ups happen each week? How many people touch them? How often does the failure occur? Is the work seasonal or continuous?

For an AI workflow, repetition also affects learning. A model that processes one rare case cannot build a meaningful review loop quickly. A frequent workflow produces corrections, edge cases, and acceptance data.

Repetition alone is not sufficient. A frequent task can still be too cheap or inconsequential to justify change. It simply tells us whether the team has enough exposure to notice and test the problem.

2. Consequence: What Does The Current State Cost?

Pain becomes commercially relevant when it creates a consequence someone cares about.

The consequence may be lost revenue, delayed response, rework, management blindness, compliance exposure, error risk, lower conversion, poor customer experience, or expensive manual effort. The important point is not to inflate the number. It is to connect the workflow failure to a decision-relevant cost.

For a phone-sales team, “calls are hard to review” is a friction statement. “Managers can review only a small sample, so missed next steps and repeated objections remain invisible until opportunities stall” describes the operating consequence.

The consequence should also be attributable enough to measure. If every business outcome depends on ten uncontrolled factors, the first pilot may need an operational proxy rather than a revenue promise. Follow-up speed, CRM completeness, accepted output, or manager review time may be more credible than claiming immediate revenue lift.

3. Owner: Who Feels And Can Change The Problem?

A problem without an owner becomes research, not a deal.

The owner is not always the user. A rep may experience the workflow, a manager may own the performance standard, operations may control the system, IT may approve the integration, and finance may approve the budget. First-customer work requires a clear view of these roles.

The most important owner has both consequence and authority. They can see the cost of the current state and can change at least part of the process if the evidence supports it.

This is especially important when the proposed product affects an operating record. If no one in the buying group can authorize CRM changes, approve new QA rules, alter a follow-up process, or assign review responsibility, a pilot may generate interesting outputs without changing work.

Owner quality can be tested. Will the person show the current workflow? Name the decision? Involve required stakeholders? Attend reviews? Resolve blockers? Those actions are stronger than enthusiasm.

4. Trigger: Why Is The Problem Active Now?

Many companies live with known inefficiencies for years. A trigger creates movement.

Common triggers include rapid team growth, a missed target, leadership change, a new channel, increased call volume, CRM migration, audit, expansion into a new market, a customer complaint pattern, rising labor cost, a failed implementation, or a board-level efficiency mandate.

The trigger should be connected to a deadline or decision. “AI is a priority” is broad. “The operations team must choose a workflow for the next automation program before the Q4 planning cycle” is actionable.

Triggers also improve messaging. A message based on industry relevance says, “Teams like yours often struggle with call visibility.” A trigger-aware message says, “When call volume increases faster than manager review capacity, CRM notes become a weak substitute for what buyers actually said.” The second message is useful only if the state is real.

This is why timely field research matters. The team should look for observable changes, then confirm them rather than assuming urgency from a generic signal.

5. Budget Logic: What Existing Decision Does The Spend Compete With?

Early offers often fail because the buyer understands the value but cannot locate the budget.

Budget logic is not the same as asking, “Do you have budget?” It asks what existing cost, risk, initiative, or operating line makes the purchase legible.

A call-intelligence pilot may relate to sales operations, CRM improvement, QA, manager productivity, missed-call recovery, or revenue operations. An AI legal workflow may compete with manual processing time, external support, error exposure, or an internal automation initiative. A validation sprint may compete with premature build cost or the opportunity cost of pursuing the wrong segment.

The team should understand who owns that budget, when it is planned, what evidence releases it, and what alternative use of funds will be compared with the offer.

Without budget logic, even a real problem can remain an unfunded concern.

6. Workflow Readiness: Can The Account Run A Meaningful Test?

A buyer can have urgent pain and still be unready for a pilot.

Workflow readiness includes access to users, usable data, process visibility, systems, approvals, consent, review capacity, and a practical path for handling exceptions. It also includes the willingness to define a baseline and participate in corrections.

OpenAI Academy’s use-case prioritization guidance emphasizes a clear owner, real users, dependencies, available data, readiness, approvals, a smallest test, and success or stop signals. These conditions separate an exciting use case from a testable one.

Readiness should not become an excuse to sell only to perfect companies. Early pilots often include manual work and incomplete systems. The point is to know which gaps are part of the test and which make the evidence uninterpretable.

If CRM is not trusted, for example, the pilot may first need to establish a controlled record. If consent cannot be obtained for calls, the workflow cannot be tested as designed. If no manager can review outputs, the team cannot establish acceptance.

7. Commitment Behavior: What Is The Buyer Willing To Do?

Commitment behavior is the strongest early signal because it costs the buyer something: time, access, political capital, data, coordination, or money.

Useful behaviors include showing the real process, sharing a representative data sample under appropriate controls, introducing the decision owner, agreeing to a baseline, assigning users, reviewing an artifact, defining success criteria, signing an LOI, or paying for a scoped pilot.

The commitment should match the stage. Asking for payment before the buyer understands the workflow may be premature. Continuing discovery after the buyer refuses every small commitment is also a signal.

The 6sense 2025 Buyer Experience Report is relevant here. Its study found that B2B buyers often form and rank shortlists before engaging sellers, and that buyers usually initiate the first contact. The practical implication is that a meeting does not mean the buyer is starting from zero. By the time they engage, they may already be testing whether a preferred option can satisfy requirements and survive internal validation.

Commitment behavior shows whether your offer is entering that real decision or only producing interest.

A Sales Black Box Example

Consider a prospect for Sales Black Box.

A strong workflow-state fit could look like this:

• the team handles enough calls each week to produce repeated evidence;
• missed follow-up, weak CRM records, or limited manager review creates a visible consequence;
• a sales or operations manager owns the outcome;
• a recent target miss, volume increase, or process change creates urgency;
• the project can connect to an existing revenue, CRM, QA, or operations budget;
• the team can provide an eligible call flow, consent path, CRM access, and weekly review owner;
• the buyer is willing to define a baseline and paid-conversion criteria.

A weak fit could have the same firmographics but a different state. Call volume is low. The CRM is not used to operate the sales process. No manager owns call quality. The buyer cannot provide samples or involve the relevant decision maker. The conversation remains at the level of general curiosity about AI.

The weak account may become relevant later. It should not receive the same pilot attention now.

A Simple Workflow-State Score

For the next 25 accounts, score six core dimensions from 0 to 2:

Dimension	0	1	2
Pain and consequence	vague or low-cost	visible but not quantified	material and decision-relevant
Frequency	rare or unclear	recurring	high enough for rapid evidence
Owner	absent	user or sponsor without authority	accountable owner with influence
Trigger	none	general priority	current event with timing
Readiness	key inputs unavailable	gaps can be resolved	owner, data, review, and approvals available
Commitment	opinion only	limited time or access	workflow, data, stakeholders, baseline, LOI, or payment

Budget logic can be added as a seventh score when the sales motion is mature enough to qualify it reliably.

The total is not scientific. Its value is comparative. It forces the team to distinguish an account that resembles the market from an account that can produce a useful commercial test.

The score should influence both priority and message. A high-trigger, low-readiness account needs a readiness conversation. A high-pain account with no owner needs stakeholder mapping. A strong-fit account with weak commitment may need better proof or a smaller next ask.

Use The Score To Change The Week’s Work

An ICP framework is only useful if it changes behavior.

After scoring the accounts, compare the result with the original prospecting order. Which familiar logos fall? Which smaller accounts rise because the workflow state is clearer? Which missing information becomes the focus of the next call? Which segment repeatedly shows pain but no budget path?

Then change the outreach. Instead of writing one message for an industry, write for a workflow state. Instead of asking for a generic discovery call, ask to inspect one process, baseline, or decision. Instead of advancing every interested account, ask for a stage-appropriate commitment.

This is closely related to the approach in How To Get Your First Paying B2B Customers: first-customer work is not a scaled demand-generation exercise. It is a structured search for a narrow group willing to help prove the problem, workflow, offer, and payment path.

The Practical Consequence

Your ICP may not be wrong. It may be incomplete.

Keep the industry, size, geography, technology, and role filters. Add the state of the work: repetition, consequence, owner, trigger, budget logic, readiness, and commitment.

The static layer tells you where to look. The workflow-state layer tells you where to spend this week’s attention.

At Proof Engine Grow, this is how we think about early GTM and first-customer work. The objective is not to create the largest plausible list. It is to identify which accounts can produce the next useful evidence and which offer, message, or pilot can earn a real commitment from them.

FAQ

What is a workflow-state ICP?

A second layer on top of firmographics that describes when a problem becomes buyable. The unit is not the company alone but the interaction between a company, a recurring workflow, an accountable owner and a moment in time, described through repetition, consequence, owner, trigger, budget logic, readiness and commitment behaviour.

Does this replace firmographics?

No. Industry, size, geography, technology and role still reduce the market to a plausible set of accounts. They just cannot show whether the problem is active, who owns it, or whether the buyer can change the workflow. The static layer says where to look; the workflow state says where to spend this week.

How do you score accounts by workflow state?

Take the next 25 accounts and score six dimensions from 0 to 2: pain and consequence, frequency, owner, trigger, readiness and commitment. The total is not scientific; its value is comparative. It separates an account that resembles the market from one that can produce a useful commercial test.

What is the strongest early signal from an account?

Commitment behaviour, because it costs the buyer something: showing the real process, sharing a data sample under appropriate controls, introducing the decision owner, agreeing to a baseline, signing an LOI or paying for a scoped pilot. The ask should match the stage.

Sources And Continuation Paths

• 6sense: The B2B Buyer Experience Report for 2025
• OpenAI Academy: Evaluate AI Workflow Readiness
• Strategyzer: Designing Strong Experiments
• Proof Engine: Grow and Methodology
• Proof Engine blog: How To Get Your First Paying B2B Customers and What Is Sales Black Box?