## Why It Matters

Most hiring goes wrong before the first interview. The team never agreed on what "great" looks like, so every interviewer measures candidates against a private, shifting standard — and "gut feel" quietly fills the gap. The result is slow decisions, disagreement at debrief, and bias that hides behind the word *fit*.

An **Ideal Candidate Profile (ICP)** fixes this. It's a short, role-specific definition of the skills, traits, and experience a person needs to **succeed** in the job — written down and agreed on *before* you source or interview anyone. Defining tangible, role-relevant criteria up front (instead of relying on intuition) is one of the most replicated findings in hiring research: it raises the predictive validity of your process and shrinks differences between demographic groups.

## ICP vs Job Description vs Persona

These three are easy to confuse. They serve different purposes and different audiences.

| | What it defines | Audience | Lives where |
|---|---|---|---|
| **Job description** | What the *job* is — responsibilities, qualifications | Candidates (external, marketing) | Your career portal |
| **Ideal Candidate Profile** | Who *succeeds* — outcomes, competencies, must-haves | Your hiring team (internal) | Your hiring tool / intake doc |
| **Sourcing persona** | Where that person is found — titles, companies, channels | Recruiters / sourcers | Sourcing playbook |

The key distinction: a job description **describes the role**; an ICP **defines the person who will thrive in it**. Write the ICP first — it's the source material for everything else, including the [job description](/docs/writing-effective-job-descriptions).

> [!TIP]
> **Build the ICP, then write the job post.** The job description is the public, polished translation of your internal ICP. Doing them in that order keeps the posting honest and your interviews aligned to the same bar.

## The Scorecard: A Proven Structure

The most rigorous ICP format is the **scorecard**, popularized by Geoff Smart and Randy Street in *Who: The A Method for Hiring*. A scorecard isn't a job description — it's a set of **outcomes and competencies that define a job done well**. It has three parts:

1. **Mission** — One or two sentences on why the role exists. *Example (sales): "Grow client base and revenue through relationship-building and strong sales technique."*
2. **Outcomes** — Three to seven specific, measurable, time-bound results the hire must deliver. Rank them by importance. *Examples: "Grow the client base by 30% and revenue by 25% in the first year," "Respond to every inbound customer within 24 hours."*
3. **Competencies** — The capabilities and behaviors needed to hit those outcomes. *Examples: strong interpersonal skills, proven sales record, prospecting, product knowledge.*

The power of this structure is that it forces you to define **success** (outcomes), not just **activity** (a task list). Most job descriptions list tasks but never say what a great hire actually accomplishes.

### Skills vs Competencies

These aren't the same thing, and a strong profile separates them:

- **Skills** are *what* a candidate can do — write SQL, close deals, design layouts.
- **Competencies** are *how* they do it — judgment, collaboration, resilience under pressure.

Outcomes tell you the destination; competencies tell you whether this person can drive there.

## Write Criteria as Observable Behaviors

The single biggest quality lever is specificity. Abstract qualities can't be assessed consistently — every interviewer interprets them differently.

| Useless (abstract) | Useful (observable) |
|---|---|
| "Communicates clearly" | "De-escalates a frustrated stakeholder without escalating to their manager" |
| "Team player" | "Gives direct feedback in code review and incorporates it gracefully when receiving it" |
| "Detail-oriented" | "Catches edge cases in a spec before they reach production" |

Aim for **five to seven concrete behaviors**, not a wall of adjectives. Behavioral anchors like these are what make a rating rubric reliable — they reduce halo effects and bias against under-represented candidates, because everyone is scoring against the same observable bar.

## Must-Have vs Nice-to-Have

Forcing this split is what makes a profile usable. If everything is required, nothing is prioritized.

- **Must-have:** Limit to a handful of genuinely non-negotiable items. If a strong candidate lacking one would still be worth interviewing, it's not a must-have.
- **Nice-to-have:** Skills that help but are learnable on the job, or that only some great candidates will have.
- **Disqualifiers:** It's also worth naming backgrounds or patterns that typically *don't* work for this role — it sharpens sourcing and screening.

> [!WARNING]
> **Don't let nice-to-haves leak into scoring.** A common failure is treating a "preferred" skill as a tiebreaker that quietly becomes a requirement. Decide the weight up front and hold to it.

## Build It Collaboratively

An ICP written by one person in isolation is just one person's bias on paper. Build it in an **intake meeting** with the hiring manager, recruiter, and key interviewers before sourcing begins.

- The **hiring manager owns** the profile; the recruiter **shapes and challenges** it. This is collaboration with clear ownership, not flat co-authorship.
- **Calibrate against reality.** Look at people who have actually excelled (and struggled) in similar roles. What did the great ones do? What did the profile's must-haves actually predict?
- Agree explicitly on **what separates an okay hire from a great one** — that gap is the heart of the profile.

## Culture Fit vs Culture Add

"Culture fit" is where good intentions most often become bias. Research on elite firms (Rivera, *American Sociological Review*, 2012) found more than half of evaluators rated cultural fit their single most important criterion — and that they "construct merit in their own image," favoring candidates who shared their own tastes, backgrounds, and class-coded hobbies. Hiring for fit-as-similarity produces a homogeneous team and quietly disadvantages anyone who isn't already like the people in the room.

The fix is **culture add**: a candidate who aligns with your **values** *and* brings new skills and perspectives the team lacks.

| Culture fit (risky) | Culture add (better) |
|---|---|
| "Would I want to get a beer with them?" | "Do they share our values of X and Y?" |
| Personality and affinity | Demonstrated values and complementary strengths |
| Rewards sameness | Rewards difference that strengthens the team |

Treat values as **assessable competencies** with observable behaviors — not a vibe you "just know" when you meet someone. If culture must be scored, give it a defined, modest weight rather than letting it override skills.

## How the ICP Feeds Structured Interviews

The ICP only pays off if it drives the actual evaluation. Translate each competency into a **structured interview**: the same questions for every candidate, scored against the same rubric.

- **Standardized rubric.** Define what an *outstanding*, *solid*, *borderline*, and *poor* answer looks like for each competency, so every interviewer shares one definition.
- **Score independently, then discuss.** Each panelist rates each competency on their own first; the panel then meets to talk through significant disagreements. Independent-first scoring stops one loud voice from anchoring the room.
- **Equal weighting by default.** Weight competencies equally unless you have a clear, documented reason not to — it's the most defensible choice.

The evidence for this is strong. Structured interviews roughly **halve interviewer bias** compared to unstructured ones (a meta-analysis found bias effect sizes of d≈.23 vs d≈.59) while raising validity and rater agreement. In the most-cited natural experiment, switching orchestras to **blind auditions explained an estimated 30–55% of the increase in women hired** — concrete proof that *how* you structure evaluation changes *who* gets selected.

### A Simple Scoring Scale

A widely-used government standard (U.S. OPM) is a five-level proficiency scale, with behavioral anchors written for each level:

| Level | Label | Meaning |
|---|---|---|
| 5 | Expert | Deep mastery; sets the standard for others |
| 4 | Advanced | Handles complex cases independently |
| 3 | Intermediate | Solid working capability |
| 2 | Basic | Functional with guidance |
| 1 | Awareness | Familiar only |

Use **one scale across all competencies**, and have people who know the role write example behaviors for each level so ratings stay consistent.

## A Living Document

Profiles go stale. After a team sees a few real candidates, criteria drift — the bar moves, people screen from memory, and the original doc becomes a fossil. Revisit the ICP partway through a search: are the must-haves actually predicting quality, or just filtering out good people? Update it deliberately, with the team, rather than letting it erode silently.

## Common Anti-Patterns

| Anti-Pattern | Why It Fails | Fix |
|---|---|---|
| **Unicorn / purple squirrel** | A profile demanding every skill at expert level finds no one and stalls the search | Cut to genuine must-haves; rank outcomes |
| **Proxy credentials** | "Degree required" or "must come from a top company" screens for privilege, not capability | Require demonstrated skills and outcomes, not pedigree |
| **Cloning the team** | "Someone like us" rebuilds the team's blind spots and homogeneity | Hire for culture *add* and complementary strengths |
| **Requirement bloat** | 15+ requirements signal an unclear role and deter strong applicants | Limit must-haves; move the rest to nice-to-have |
| **Abstract criteria** | "Rockstar," "great communicator" can't be scored consistently | Rewrite as observable behaviors |
| **Affinity / recency bias** | Profile shaped by the last great (or bad) hire, or by who the interviewer likes | Calibrate against multiple past performers and data |
| **Vague "culture fit"** | Masks similarity bias, harms diversity | Define values as scored, observable competencies |

## Quick Checklist

Before you start interviewing, verify your ICP:

- [ ] **Mission** states why the role exists in 1-2 sentences
- [ ] **Outcomes** are 3-7 specific, measurable, time-bound results
- [ ] **Competencies** are separated from skills (the *how*, not just the *what*)
- [ ] **Criteria** are observable behaviors, not abstract adjectives
- [ ] **Must-haves** are limited to genuine non-negotiables
- [ ] **Nice-to-haves** are separated and won't creep into scoring
- [ ] **Built collaboratively** with the hiring manager owning it
- [ ] **Calibrated** against people who actually succeeded in similar roles
- [ ] **Culture add**, not culture fit — values defined as scorable behaviors
- [ ] **Rubric** defines outstanding / solid / borderline / poor for each competency
- [ ] **Scoring** is independent-first, equal-weighted, then discussed
- [ ] Treated as a **living document** to revisit mid-search