Claude (Anthropic) — Funnel Overview

Funnel Summary

• Total steps: 33 screens (the most comprehensive onboarding in the AI assistant space)
• Funnel type: Conversational onboarding with progressive feature disclosure
• Time to complete: 5-8 minutes (estimated based on 33 screens)
• Data collected: Email/SSO, name, use case/topic selection, preferences, feedback on generated content
• Payment timing: Free tier available. Pro at $20/month, Team at $25/month, Enterprise custom. Pricing presented mid-flow (after value demonstration), not at start.
• Personalization level: Heavy — use case selection, topic preferences, tone/length controls, per-Project custom instructions, team collaboration settings

Funnel Flow

Step 1-3: Account Creation & Authentication

• Sign up screen with email or SSO (Google)
• Account selection interface
• Terms agreement
• Name entry

Step 4-6: Personalization & Preferences

• Onboarding question — initial use case inquiry ("What do you want to use Claude for?")
• Topic selection interface — choose areas of interest
• User preference settings — tone (Formal/Casual), length (Short/Detailed)

Step 7-11: Product Introduction & Value Demonstration

• AI generator demonstration — shows Claude creating content
• Sample prompt submission — user types a real prompt
• Response preview with details — Claude generates a response
• Feedback submission mechanism — teaches users they can rate/shape outputs
• Artifact publishing workflow — introduces Artifacts (code, documents, diagrams) as side-panel outputs

Step 12-14: Plan Selection

• Pricing & plans presentation (appears AFTER value demonstration)
• Plan selection screen
• Account upgrade options

Step 15+: Advanced Feature Discovery

• Projects feature introduction
• Team collaboration features
• Advanced artifact types
• Additional capability demonstrations

What Works Well

1. Comprehensive Value Demonstration Before Pricing

Claude's most distinctive strategy: pricing appears at screens 12-14, AFTER the user has experienced multiple demonstrations of value. By the time users see the price, they've already typed a prompt, seen a response, submitted feedback, and explored artifacts. This creates strong anchoring — the perceived value exceeds the price.

2. Feedback Mechanism Teaches Product Shaping

Including a feedback submission step during onboarding teaches users that they can influence outputs. This is a retention mechanism — users who know they can shape Claude's responses are more likely to persist through initial imperfect outputs.

3. Artifact Publishing as "Wow" Moment

The artifact workflow (where Claude generates viewable/editable code, documents, or diagrams in a side panel) is introduced during onboarding as a showcase feature. This differentiates Claude from simple chatbots and demonstrates advanced capabilities.

4. Conversational Tone Throughout

Claude's onboarding uses warm, conversational language ("Let me know if you have questions"). This creates emotional rapport and reduces the "talking to a machine" feeling that can cause AI tool abandonment.

5. Per-Project Customization Introduced

Users learn about Projects (organized workspaces with custom instructions) during onboarding. This teaches a power-user workflow early, increasing the likelihood of deeper product adoption.

What Could Be Better

1. 33 Screens May Cause Drop-Off

While each screen serves a purpose, 33 is significantly more than any competitor (ChatGPT: ~10, Perplexity: ~3, Gemini: ~3). Users seeking quick answers may abandon the flow. A "skip to product" option could reduce drop-off.

2. No Interactive Demo Prompt Early

Unlike ChatGPT (which prompts users to "Ask anything" at Step 6), Claude's interactive element appears later in the flow. Moving the "try it" moment earlier could increase engagement.

3. Complex Feature Set May Overwhelm

Introducing Projects, Artifacts, feedback, publishing, AND team features during onboarding may create cognitive overload. A staged approach (core features in onboarding, advanced features via progressive emails) could be more effective.

4. Limited Social Proof Integration

The onboarding doesn't prominently feature user testimonials, case studies, or adoption metrics. Given Claude's growing enterprise presence, this is a missed opportunity.

Key Psychological Principles Used

Sunk Cost / Escalation of Commitment

With 33 screens, users who've progressed through 10+ screens feel invested in completing the flow. Each completed step makes abandonment feel like "wasting" effort already invested.

Reciprocity

Claude demonstrates value (generating responses, showing artifacts) before asking for anything (plan selection). This creates a reciprocity dynamic — users feel they "owe" engagement in return for value received.

Endowed Progress Effect

By completing early onboarding steps, users feel they've made progress toward "mastering" Claude. This perceived progress motivates continued engagement.

Feature Anchoring

Showing advanced features (Artifacts, Projects, team collaboration) during onboarding anchors users' perception of Claude's value at its maximum capability, even if they initially use only basic chat.

Mid-Flow Pricing (Decoy Effect Potential)

Presenting pricing mid-flow (after value demonstration) allows Anthropic to show the free tier alongside paid options in a context where users have just experienced premium-quality outputs. This makes the Pro tier feel like unlocking capabilities they've already previewed.

Relevance to Twofold

Direct Transfers

1. Mid-Flow Pricing Presentation: Twofold should show pricing AFTER the user has experienced a sample note generation, not on the landing page. This ensures users understand the value before seeing the cost.

2. Feedback Mechanism in Onboarding: Teaching clinicians they can edit/refine AI-generated notes during onboarding sets expectations correctly — AI augments, not replaces, clinical judgment. This reduces "accuracy anxiety."

3. Artifact-Style Note Preview: Like Claude's artifact side panel, Twofold should show the generated clinical note in a formatted preview panel alongside the recording/transcript. This visual separation makes the output feel professional and editable.

4. Conversational Onboarding Tone: Using warm, respectful language during onboarding ("We know your time is valuable. Let's get you set up in 3 minutes.") reduces the clinical coldness that clinicians experience elsewhere.

Adaptations Needed

1. Shorter Flow: Claude's 33 screens are too many for busy clinicians. Twofold should aim for 8-12 screens maximum, with advanced features introduced via progressive onboarding emails.

2. Specialty-First Customization: Claude's topic selection is general. Twofold should lead with specialty selection (therapy/psychiatry/PT) to immediately customize note templates and onboarding examples.

3. Clinical Context: Claude's demonstrations show general writing/coding. Twofold's equivalent should show clinical note generation specifically — a sample therapy session → SOAP note would be the most compelling demo.