Build evidence-labeled personas with ak:persona

Use ak:persona to turn approved audience inputs into an ICP profile or audience analysis. A persona is a decision aid: it summarizes evidence and hypotheses for a defined use case, but it does not prove that every customer behaves alike.

Choose the action that matches the artifact

Use ak:persona when

• create should gather audience inputs and draft a new persona.
• analyze should identify segments and patterns in supplied audience or behavior data.
• update [name] should revise an existing persona with new evidence.
• list should enumerate persona artifacts available in the working context.

Choose another workflow when

• You need fresh market evidence before forming a customer hypothesis. Use ak:marketing-research.
• You need competitor positioning rather than a customer profile. Use ak:competitor.
• You need to turn approved personas into channel, message, owner, and measure decisions. Use ak:marketing-planning.

Prepare lawful, relevant inputs

Define the product, market, decision, geography, timeframe, and intended use of the persona. Supply only data you are authorized to use, preferably aggregated or de-identified. Useful inputs include interview themes, survey results, support themes, product-behavior summaries, sales notes, and current market research with source dates.

The bundled create flow asks about age, role, industry, pain points, challenges, goals, motivations, buying behavior, and preferred channels. Include a field only when it is relevant, lawful, supported, and necessary for the stated decision. Do not infer protected or sensitive traits from proxies.

Invoke a persona action

/ak:persona create "Draft an ICP for admins at 20-100 person software companies from the supplied interview themes. Cite each finding, label assumptions, exclude names and sensitive traits, and do not contact anyone"

/ak:persona create "Draft an ICP for admins at 20-100 person software companies from the supplied interview themes. Cite each finding, label assumptions, exclude names and sensitive traits, and do not contact anyone"

$ak:persona create "Draft an ICP for admins at 20-100 person software companies from the supplied interview themes. Cite each finding, label assumptions, exclude names and sensitive traits, and do not contact anyone"

The canonical actions are create, analyze, update [name], and list.

Understand the stages

1. Set the decision scope. State what the persona will help decide and which population, product, market, geography, and period it represents.
2. Gather inputs. For create, the workflow asks for audience attributes, problems, goals, behavior, and channels. Missing fields remain unknown rather than being filled from stereotypes.
3. Analyze evidence. The workflow can use the lead-qualifier, researcher, and, for behavior data, analytics-analyst Agents when available.
4. Separate evidence from interpretation. Preserve source, date, sample, conflicts, and limitations for observed findings; label inferred patterns and proposed needs as hypotheses.
5. Create or revise the artifact. Persona profiles are written to assets/leads/icp-profiles/{persona}.md.
6. Review and validate. Ask a human owner to check evidence quality, privacy, fairness, usability, and whether more customer validation is needed.

Protect people and approval boundaries

Reading a supplied dataset, searching the web, or calling analytics systems can have privacy, network, credential, and provider effects. Approve each access separately. Creating a persona does not authorize outreach, audience upload, tracking, ad targeting, CRM changes, publication, or spend.

Verify outputs and evidence

A decision-ready profile should include:

• Intended decision, scope, population, timeframe, and owner.
• Source list with collection dates, sample context, and permitted use.
• Observed needs and behaviors separated from interpretations and assumptions.
• Conflicting evidence, excluded data, confidence limits, and validation gaps.
• Goals, problems, buying context, and channels only where relevant and supported.
• Artifact path, review status, expiry or refresh trigger, and unresolved questions.

Troubleshoot and know the limits

The Skill source does not define a persona schema, minimum sample size, statistical method, confidence threshold, expiry policy, or validation score. Its generic demographic prompts are optional inputs, not evidence. Continue with the Marketing Kit overview or Projects, artifacts, and checkpoints.