Design, analyze, or optimize a funnel with ak:funnel

Use ak:funnel to create a funnel design, analyze an existing funnel, or recommend optimizations. The Skill coordinates funnel architecture, performance analysis, and conversion guidance, then writes a design, audit, or test artifact. It does not build pages, send campaigns, change checkout, or spend budget.

Choose ak:funnel for a whole journey

Use ak:funnel when

• You need a new lead-magnet, webinar, product-launch, evergreen, or tripwire funnel design.
• You have stage-level counts or rates and need to locate observed drop-off.
• You want prioritized recommendations and A/B test suggestions across traffic, landing, capture, nurture, sales, checkout, and follow-up stages.

Choose another workflow when

• One non-signup form is the bounded problem. Use ak:form-cro.
• The post-signup first-run experience is the bounded problem. Use ak:onboarding-cro.
• You already have one experiment candidate. Use ak:ab-test-setup.
• You need implementation, campaign activation, paid traffic, or publishing. First approve the funnel artifact, then use the relevant execution workflow.

Run a conversion-optimization workflow

Begin with observed behavior, not a preferred redesign. The chain is ak:analytics → ak:funnel → the relevant CRO Skill → ak:ab-test-setup → ak:analytics.

Preserve the baseline query, dashboard definition, hypothesis, approved variant, instrumentation version, experiment ID, and decision rule across the chain. Never choose a winner from an incomplete window or replace a failed primary metric with a favorable secondary metric. Shipping the chosen change remains a separate approved implementation workflow.

Prepare journey and measurement context

For design, provide the offer, audience, funnel type, traffic assumptions, required stages, constraints, and success definition. For analyze or optimize, provide a dated stage map, unique-user counts, conversion definitions, attribution window, segments, acquisition sources, costs where relevant, and known tracking changes. Redact personal records and secrets.

Complete Onboarding, and confirm Marketing Kit is installed for the runtime and scope you are using.

Run the Skill

The first argument is an exact action. design also requires one supported funnel type; analyze and optimize do not declare a type argument.

/ak:funnel design webinar "Create a draft funnel for the supplied audience and offer. Define every stage and metric, state traffic and cost assumptions, and do not publish, buy traffic, or configure providers"

/ak:funnel design webinar "Create a draft funnel for the supplied audience and offer. Define every stage and metric, state traffic and cost assumptions, and do not publish, buy traffic, or configure providers"

$ak:funnel design webinar "Create a draft funnel for the supplied audience and offer. Define every stage and metric, state traffic and cost assumptions, and do not publish, buy traffic, or configure providers"

For an audit, use /ak:funnel analyze in Claude Code or Cursor and $ak:funnel analyze in Codex. Include the same metric definitions and time window used by the supplied data.

Understand the workflow stages

1. Parse the action. The Skill validates design, analyze, or optimize and reads the funnel type when the action is design.
2. Map the journey. A design can cover traffic source, landing page, lead capture, nurture sequence, sales page, checkout, and follow-up.
3. Define measurement. The Skill names the population, event, denominator, time window, and metric for each included stage.
4. Analyze observed flow. For an existing funnel, it calculates stage rates from supplied evidence and identifies drop-off without inventing benchmarks.
5. Prioritize changes. Optimization combines architecture and conversion guidance, ranks recommendations, and proposes bounded tests.
6. Write the artifact. The Skill uses the documented Marketing asset paths so the result can be reviewed before separate implementation.

Keep recommendations separate from external effects

• Use real stage definitions and consistent unique-user denominators; do not mix sessions, users, leads, orders, and revenue without labeling them.
• Separate observed data, assumptions, external benchmarks, and recommendations.
• Do not use deceptive scarcity, hidden continuity, forced opt-in, misleading social proof, or an obstructed cancellation path to improve a stage rate.
• Include lead quality, refund, complaint, unsubscribe, retention, accessibility, and unit-economic guardrails where the journey can affect them.
• Treat industry benchmarks as dated context only when a source and comparable population are provided.

Review outputs and evidence

The source declares these exact output patterns:

Review the artifact for stage ownership, metric definitions, evidence dates, assumptions, recommendation priority, testable hypotheses, dependencies, consent, and separately approved implementation steps. The file is the Skill's deliverable; a working live funnel is not implied.

Troubleshoot the workflow

Know the limits

• The Skill does not guarantee traffic, conversion, revenue, profit, or a complete implementation.
• Attribution, cross-device identity, long sales cycles, seasonality, and offline steps can make stage analysis incomplete.
• The source names a campaign workflow and specialist Agents, but their use does not authorize external campaign actions.
• The documented package releases contain identical ak:funnel actions, types, outputs, and invocations.

Continue with ak:ab-test-setup for a selected test or review the Marketing Kit overview.