An ideal customer profile (ICP) is a testable description of the accounts that are most likely to obtain value, buy appropriately and remain economically viable for your business. Segmentation separates customers into groups whose needs or buying behaviours justify different strategies. Neither exercise is simply an exercise in describing everyone who has ever bought.
For SaaS and B2B services, a useful ICP connects the buyer’s problem to product fit, operating constraints and long-term value. It helps marketing choose relevance, sales identify legitimate priority and delivery know what to expect.
Executive takeaways
- Separate account fit, the specific buying trigger, and current intent. A desirable account may not be in-market; an eager prospect may be a poor fit.
- Use actual customers, lost deals, product usage and delivery experience. Demographics or industry tags alone can overstate precision.
- Treat exclusion criteria as operating safeguards, not as moral judgements about prospects.
- Segment only where a meaningful difference changes product, message, route or commercial decision.
- Update the ICP through a governed feedback loop; do not silently adjust scoring to make dashboards look better.
1. ICP, persona and segmentation are different tools
The ICP usually describes an organisation or account: its environment, constraints, readiness and likely fit. A buyer persona describes the relevant people and their roles in a purchase or adoption. Segmentation divides the addressable population into groups with meaningfully different needs or responses.
A buyer may be a founder, marketer, finance approver or system owner. Those roles can sit inside the same account. A title cannot by itself establish fit. Equally, treating every business above a certain headcount as ideal ignores whether its problem, budget, procurement and implementation capacity match the offer.
The right question is not “Who looks like our customers?” but “Which customers repeatedly experience the problem we solve, can adopt the solution and produce a viable relationship?”
2. Start with observed problems and switching triggers
Use recent interviews to understand the job behind the purchase. Harvard Business Review’s jobs-to-be-done account explains why knowing customer attributes alone does not reveal the circumstances that cause a choice. That theory supports investigating purchase context; it does not replace validation in your actual category.
Ask: What changed before the buyer started looking? Which task had become urgent? What workaround was being used? What risk made the existing approach unacceptable? What internal event, such as a product launch or a new sales team, created a budget conversation?
Record each trigger in the buyer’s own words. A trigger can distinguish “interesting to market to” from “plausible to buy this quarter” without assuming that the latter will convert.
The four linked decisions
Could the organisation benefit?
Why does this problem matter now?
What evaluation behaviour is real?
Can they buy and achieve value?
3. Establish account-fit dimensions
Evaluate dimensions that are both observable and linked to value: use case, operating environment, existing systems, data maturity, number of stakeholders, implementation resources and budget ownership. Company size and region matter only where they affect these requirements.
For a CRM project, integration complexity and an accountable sales-process owner might be more predictive of implementation readiness than employee count. For self-serve SaaS, the presence of repeat usage and a clear first-value event may matter more than contract size. Different business models demand different factors.
| Dimension | Evidence to collect | Possible exclusion | Review owner |
|---|---|---|---|
| Problem fit | Repeated urgent workflow | No relevant use case | Product / strategy |
| Operating fit | Systems, access, data quality | Incompatible critical dependency | Delivery / technical |
| Economic fit | Budget owner, feasible value | Unaffordable cost to serve | Finance / sales |
| Timing and intent | Trigger, evaluation action | No buying window yet | Marketing / sales |
| Adoption capacity | Sponsor, implementation owner | No owner for change | Customer success |
4. Do not collapse fit and intent into one score
A fit score is useful for defining who could benefit. Intent signals indicate what a prospect may be doing now. Combining them without keeping the inputs visible can create false precision: a large target account downloading one blog post is not automatically a sales-ready opportunity.
Gartner’s 2025 B2B buyer survey found that 73% of its 632 respondents avoided suppliers that sent irrelevant outreach. The finding supports caution in using weak intent signals for aggressive prospecting; it is not an ICP scoring threshold.
Define separate outcomes: nurture relevant-but-not-ready accounts, prioritise fit-plus-valid-intent, and route low-fit enquiries to an honest alternative. Record reasons rather than creating an opaque algorithm.
5. Build segments that change a decision
A segment is useful if its members need a different message, product entry point, buying route or success plan. Example: founder-led SaaS teams and mid-market RevOps teams might share a broad revenue problem but differ in integration constraints, proof requirements and decision-making process.
Limit the number of initial segments to those the team can understand and serve. A 20-row industry taxonomy without meaningful operational differences makes reporting elaborate but does not improve execution. Define which segments are primary, secondary and intentionally out of scope. Revisit those assumptions when evidence contradicts them.
6. Turn the ICP into channel and content decisions
The ICP should determine which questions appear in a qualification form, which topics deserve educational material, which objections a sales page must answer and how a handoff is routed. Do not make every visitor complete an exhausting questionnaire to see basic information.
Evidence should come from multiple sources: interviews, actual CRM outcomes, service-delivery notes, product analytics where available and lost-deal themes. Avoid using only closed-won deals; that excludes people who might have bought had the offer or process been different.
7. Measure whether segmentation improves the system
Useful measures include accepted demand by segment, opportunity progression, time to first value, retention by cohort, and cost to acquire and serve. Compare like-for-like periods and guard against selecting winners from tiny samples.
Signals and what to investigate
Educate; do not force a sales call
Next check
Check trigger and time horizon.
Qualify honestly and explain scope
Next check
Check use-case and delivery constraints.
Route with context and ownership
Next check
Check accepted demand and adoption.
Where reliable outcomes do not yet exist, begin with qualitative evidence and a small number of observed signals. Do not invent a universal ICP fit-score cutoff. A score is a communication tool, not a causal model.
build a usable qualification contract
Translate research into fields. A qualification form should ask only what changes routing or the ability to respond meaningfully. Company, business email and a concise description of the challenge may be sufficient for first contact; detailed system inventories can follow when needed. Avoid collecting unnecessary personal information or pretending that a form validates budget, urgency and fit by itself.
Keep disqualification reversible. Document the reason a prospect is not currently suitable: missing use case, unverified integration, absent project owner or unsuitable timing. A prospect can become relevant after its operating situation changes. Separate “not now” from “not a fit”; otherwise a CRM status turns a time-sensitive decision into a permanent claim about the account.
Audit the feedback loop. Each month, review accepted and rejected enquiries with sales and delivery. Look for segment-specific patterns. A high-fit account rejected due to missing integration support suggests a product boundary; an apparently high-intent account that never responds may reveal weak signal quality. Revisit the definition only after considering sample size and cohort maturity.
| Qualification outcome | Recommended handling | Information to retain |
|---|---|---|
| Relevant, not evaluating | Educational follow-up if permitted | Problem and timing assumption |
| Relevant, evaluating | Owned conversation with context | Trigger, route and next step |
| Unsuitable today | Honest scope response | Reason and review condition |
The ICP is useful only when it changes operational behaviour: the editorial plan talks to a defined problem, the service page qualifies scope clearly, and the receiving team knows when to act. A 100-point score without reliable underlying data often creates a veneer of accuracy. Start with understandable criteria and improve sophistication only when a larger sample supports it.
build an ICP that changes routing decisions
A usable ideal customer profile is a hypothesis about organisations likely to benefit and succeed, not a list of company-size fields. Start with five to ten recent wins, losses and stalled evaluations when available. Record the original trigger, existing workflow, who led evaluation, resource constraints, reason for progressing or stopping, and whether the customer achieved value after purchase. If the sample is small, say so and keep the ICP provisional. Do not train a statistical score on a handful of selectively remembered successes.
For a hypothetical marketing-automation provider, two companies can have the same industry and headcount yet differ profoundly: one already has dependable account ownership and needs reporting; the other lacks any agreed lead definition. A polished automation pitch may attract the second company, but that prospect could require foundational CRM work first. The distinction should change discovery questions, qualification and the proposed scope. Firmographics alone will not reveal the delivery prerequisite.
Separate the four decisions that generic lead scoring collapses
Fit asks whether the account has a problem the offer can solve and the prerequisites to realise value. Trigger asks why the problem has become consequential now. Intent describes observed evaluation behaviour with a defensible source. Readiness asks whether ownership, budget process, security requirements and implementation resources make a next step possible. Keep these as separate recorded fields rather than mixing them into an unexplained 0–100 score.
An account downloading a guide is not necessarily in-market. A high-fit account researching a new regulatory requirement may require education, not a meeting request. Conversely, a low-fit account can express urgent intent without becoming a suitable customer. Gartner’s surveyed B2B buyers provide a warning about irrelevant outreach, but the survey cannot supply a conversion probability for a particular account.
Run an ICP learning loop that can overturn assumptions
For each accepted and rejected enquiry, collect an agreed, privacy-conscious reason. Compare segment progression using mature cohorts: accepted enquiries, sales-stage movement, implementation completion and eventual customer value where measurable. Avoid a false win from changing the denominator: “higher qualification rate” is not comparable if qualification rules or intake sources changed during the period. Review reasons monthly, and revise a segment only when repeated evidence points in the same direction.
Practical output: a one-page ICP card containing entry criteria, disqualifying dependencies, trigger questions, buyer roles, evidence examples, routing action and date of last review. Keep the persona separate: the same account can include a technical evaluator, an economic buyer and an operational champion whose questions need different content.
8. Illustrative SaaS decision case
Imagine a collaboration SaaS product with enquiries from agencies and enterprise IT teams. Agencies activate quickly but rarely adopt multi-team governance; enterprise teams ask for controls the product has not validated. Calling both “ideal customers” because they show intent would produce conflicting product priorities.
Separate the two into testable segments. Examine value achievement, integration effort, security needs, retention and cost to serve. Choose an initial segment only after reviewing that evidence; do not confuse a high deal value with profitable customer fit.
9. A practical 90-day operating sequence
Days 1–30: compare real wins, losses and stalled customers; interview buyers and delivery owners; document evidence quality. Days 31–60: define one primary ICP and up to a few meaningful segments; adjust messaging and intake; train teams on fit-versus-intent distinctions. Days 61–90: review acceptance, adoption and disqualification reasons; refine only criteria whose change can be explained. These are illustrative phases, not universal timelines.
10. Failure modes and trade-offs
A rigid ICP can exclude future categories before the business has learned; an overly broad ICP can force the product and commercial teams to make incompatible promises. Firmographic precision often feels objective but may be weakly related to the underlying job. Treat segmentation as a living hypothesis with versioned evidence.
A strong governance question is: Which decision would be different if this account belonged to another segment? If the answer is none, the segmentation may not be earning its complexity.
11. Frequently asked questions
How often should the ICP change? After meaningful evidence or business-model change, with a documented review cadence. Avoid weekly changes driven by one anecdote.
Is a persona the same as an ICP? No. Persona describes the people involved; ICP describes the account fit and relevant context.
Can we use intent software to replace interviews? No. Observable behaviours can complement, not prove, the buyer’s context or willingness to adopt.
What if our best customers differ by region or size? Investigate the underlying reason: buyer process, regulation, complexity or economic fit, and segment only where it changes the strategy.
12. References and next steps
Harvard Business Review on jobs to be done provides the problem-context lens. Gartner’s B2B buyer research informs outreach relevance. These external sources do not validate any specific Bargaon segment; the matrices are editorial tools for diagnosis.
Take one real won deal and one lost deal, compare their trigger, fit and adoption capacity, and write the smallest ICP definition the evidence supports. To discuss how audience strategy connects with demand, contact Bargaon.