> Building a Persona — Copilot first, then manual

TL;DR: Pick the ICP this Persona belongs to. Define title patterns, seniority, function, tenure, person-level signals. Save as Live to start scoring.

The Copilot path

From any page:

> for the Mid-market SaaS ICP, build a Persona for CROs and VPs of Revenue,
5+ years in role, exclude IC roles

Copilot drafts:

• A Job Title list — CRO, VP of Revenue, Chief Revenue Officer, and the close variants of each
• A seniority requirement — VP and above
• A tenure requirement — 5+ years in current role
• A Disqualify rule — title contains "Intern" or "Associate"

Confirm to save. Refine in conversation: > add a +2 for anyone with "RevOps" in their title.

You can also build from outcome data:

> what Persona patterns convert best inside Mid-market SaaS?

Copilot pulls historical reply rate, meeting rate, signup rate, and purchase rate by title patterns, then proposes a Persona overweighting the highest-converting attributes.

The manual path

Open Qualification → switch the toggle from By ICP to By Persona if you want a Persona-first view. Click + Add → Persona. The drawer opens.

Top section

• Parent ICP — required. A Persona always belongs to exactly one ICP. Pick from the dropdown of your Live ICPs.
• Name — short, plain-English (e.g. "Founder", "VP Sales", "CRO", "Security Lead"). Becomes the persona_code.
• Status — Draft / Live / Paused.

Parameter list

The drawer groups every Persona parameter into exactly three categories:

Category | Parameters
**Professional Identity** | Job Title, Past Titles, Seniority, Function, Profile Keywords, Country, State / Province, City, Total Years Experience, Years in Current Role, Languages Spoken, Education Level
**Behavioral** | LinkedIn Premium, Content Creator, LinkedIn Followers, LinkedIn Connections
**Custom AI Research** | Custom AI Question

There is no separate "Exclusions" category — exclusion is a mode you set on any parameter, the same way you set Required match or Score points. Person-level signal conditions are not Persona parameters either; you filter on signals when you build an audience.

Matching on titles

Job Title is one free-text list. Type as many titles as you want and add them individually. There is no wildcard syntax, and there is no separate "semantic" or "pattern" input type — close variants match automatically, so VP Sales also catches Vice President, Sales.

Past Titles is a distinct list covering roles the person no longer holds — the way to reach ex-founders and former operators.

Because variant matching is deliberately broad, a title list in Required match mode can exclude more than you intend. Put it on Score points first and look at what actually scores before you make it a hard requirement.

Custom AI Research

One free-text question, answered per record by the research model. Useful when title alone doesn't capture intent. Examples:

• "What are they trying to solve right now?"
• "What stage of buying cycle are they in?"
• "What's their current vendor for X?"

The LLM answers by reading public profile data, recent posts, and any captured signals. The answer surfaces on the lead detail page under Persona custom question and is searchable.

Save

Click Save and run. If Live, the Persona runs against all records inside the parent ICP within a few seconds.

Promotion: Draft → Live

Same flow as ICPs: Draft for editing, Live to start scoring. A Persona only scores records that already qualified for its parent ICP — Persona scoring is scoped, not workspace-wide.

Sanity checks after save

• Filter Leads by the new persona_code, sort by persona_score descending. Spot-check the top 20. If names and titles look right, the Persona is dialed in.
• Check No persona fit for the parent ICP. A high count there means your Persona is too narrow — there are records that qualified for the ICP but no Persona inside it.
• The Reachable % on audiences using this Persona should be >40%. Low Reachable % usually means the Persona qualifies records the enrichment waterfall can't find a contactable email for.

Common patterns

• One ICP, three Personas — covers most B2B motions. The buying committee usually has 3 roles: economic buyer, technical evaluator, end user.
• Persona-specific custom question — when the same title means different things at different ICPs. Custom question disambiguates.
• Tenure floor of 1+ year — eliminates people in role for <1 year; they're often still learning their job and harder to convert.

Related

• ICP vs Persona — the difference
• Building an ICP — Copilot first, then manual
• Parameter modes — Required match, Score points, Disqualify
• Reading ICP and Persona scores