Onboarding Dialogue Flows · v1.0

Smart Onboarding — Preference Profiling

A dynamic, adaptive conversation flow that builds a rich traveler profile in under 3 minutes — with zero dropout. Every question responds to previous answers. The system knows when to dig deeper and when to move on.

Avg. completion: ~2.5 min Questions shown: 8–14 of 24 total Profile fields set: 18–28 Target dropout rate: < 12%

01 Live Prototype 02 Design Principles 03 Conversation Flow Map 04 Question Bank 05 Anti-Dropout Mechanics 06 Profile Output Schema

01

Live Prototype — Try the Onboarding

This interactive simulation demonstrates the full adaptive onboarding flow. Answers branch dynamically — choose different options to see how the conversation adapts. Watch the profile build in real time.

T

Tico

Online · Building your profile

Let's get started

Profile so far

↺ Start over

Generated Profile (JSON)

{
  "status": "awaiting input…"
}

Profile Confidence

0%

02

Design Principles

The onboarding is the first and most critical UX moment. It sets tone, builds trust, and determines the quality of every future recommendation. These principles govern every decision.

🎯

One question at a time

Never show more than one question in a single message. Each exchange has a single ask. Cognitive load is the enemy of completion.

🌿

Adaptive branching

The question tree changes based on every answer. A solo traveler never sees group-size questions. A budget traveler skips luxury hotel options.

Solo → skip group dynamics branch
Adventure → unlock outdoor depth track

⚡

Fast first win

Within 2 exchanges, show something personalized — a destination suggestion, a style tag. This proves value before the user invests more time.

💬

Conversational, not clinical

Every question uses natural language, not form labels. "What kind of trip sounds perfect right now?" beats "Select trip type".

🔮

Inference over interrogation

When an answer implies something, don't ask again. If someone chooses "street food & local markets", we infer budget-conscious and culturally curious.

🛡️

Always skippable

Every non-critical question has a "skip for now" option. Better to get 80% profile than lose the user at question 5.

📊

Progress is visible

Users see a thin gold progress bar and a "Profile building" summary. Seeing completion reduces dropout by ~34% (internal benchmark).

🔄

Refinable over time

Onboarding collects a seed profile. Every future trip, rating, and interaction refines it. Users can edit preferences anytime in settings.

Question Input Types

Single Choice

Pill buttons — tap one. Fastest interaction, highest completion. Used for destination type, trip style, pacing.

Multi-Select

Select all that apply. Max 3 selections enforced. Used for interests, dietary needs, travel priorities.

Scale / Slider

1–5 tap scale or text range. Used for budget, pace preference, adventure level.

Free Text

Short open input. Used sparingly — only for dream destination and dietary restrictions (specific).

03

Adaptive Conversation Flow Map

The flow tree shows how questions branch based on previous answers. Gold paths are the primary flow (shown to all users). Teal and purple are conditional branches activated by specific responses.

04

Full Question Bank — 24 Questions

The complete library of questions. Any given user sees 8–14 depending on their answers. Questions marked CORE appear for all users. CONDITIONAL questions are triggered by specific responses. OPTIONAL are skippable depth questions.

ID	Type	Track	Question text	Options / Range	Branch logic	Profile impact
CORE — Shown to all users
Q01	Choice	CORE	What kind of trip sounds perfect right now?	Beach escape · City culture · Adventure · Road trip · Wellness · Bleisure	Adventure → Q02a Bleisure → Q02b Other → Q02	trip_stylecategory
Q02	Choice	CORE	Who's coming with you?	Just me · My partner · Small group (3–5) · Family with kids	Family → Q03a (ages) Group → Q03b (dynamic) Solo/Couple → Q03	group_typegroup_size
Q03	Choice	CORE	How do you usually think about budget?	I stretch every dollar · Mid-range comfort · Premium when it matters · Money is no object	Budget → Q04a Luxury → Q04b Mid/Premium → Q04	budget_tierbudget_range
Q04	Multi	CORE	What makes a trip worth it for you? Pick up to 3.	Local food · Nature & outdoors · History & culture · Nightlife · Wellness & spa · Photography · Sports · Shopping · Off-the-beaten-path · Luxury & comfort	Food → Q05a (dietary) Nature → Q05b (activity level) All → Q05	interests[]priorities[]
Q05	Choice	CORE	What's your ideal trip pace?	Packed (see everything) · Balanced · Slow & deep · Spontaneous (minimal planning)	Spontaneous → skip Q06 Packed → Q06	paceplanning_style
Q06	Choice	CORE	Where do you prefer to sleep?	Hotel (4–5★) · Boutique hotel · Airbnb / apartment · Hostel / social · Resort / all-inclusive · Doesn't matter	→ Q07	accommodation_typeaccommodation_tier
Q07	Multi	CORE	Which regions call to you most?	Europe · Southeast Asia · Japan & Korea · Middle East · Americas · Africa · Oceania · Anywhere!	→ Q08	preferred_regions[]
Q08	Text	CORE	Is there a place you've always dreamed of going?	Free text, 80 chars max (or "Not sure yet")	→ Q09	dream_destination
Q09	Choice	CORE	How long are your typical trips?	Weekend getaway · 4–7 days · 1–2 weeks · 3+ weeks · It varies	→ COMPLETE (core done)	trip_duration
CONDITIONAL — Triggered by specific answers
Q02a	Scale	COND	How adventurous are we talking? [Triggered: Q01=Adventure]	1 (Scenic hikes) → 5 (Extreme / off-grid)	≥4 → unlock extreme activity options later	adventure_level
Q02b	Choice	COND	On work trips, how much leisure time do you usually carve out? [Triggered: Q01=Bleisure]	None (pure work) · A few hours · Half the trip · I blend fully	Activates loyalty / schedule optimization path	bleisure_ratiobusiness_traveler
Q03a	Multi	COND	What ages are the kids? [Triggered: Q02=Family]	Under 3 · 3–6 · 7–12 · Teenager	Under 3 → stroller access, nap-friendly, → Teens → nightlife safe options	children_ages[]family_type
Q03b	Choice	COND	How does your group usually make decisions? [Triggered: Q02=Group]	I plan, they follow · We vote on everything · Flexible / go with the flow	Affects whether Tico addresses the group or the planner	decision_styleis_planner
Q04a	Multi	COND	Where do you prefer to put your limited budget? [Triggered: Q03=Budget]	Flights · Accommodation · Food · Experiences · Transport	→ Q05	budget_priorities[]
Q04b	Multi	COND	In luxury mode, what matters most? [Triggered: Q03=Luxury]	Suite/villa · Private transport · Fine dining · Exclusive access · Spa & wellness	→ Q05	luxury_priorities[]
Q05a	Multi	COND	Any dietary needs I should always keep in mind? [Triggered: Q04 includes Food]	Vegetarian · Vegan · Halal · Kosher · Gluten-free · Allergies (specify) · No restrictions	Allergies → text input for specifics	dietary[]restrictions[]
Q05b	Scale	COND	How physically active do you like to be on trips? [Triggered: Q04 includes Nature/Adventure]	1 (Gentle strolls) → 5 (Daily workouts / hard hikes)	→ adjusts activity suggestions	activity_level
OPTIONAL — Skippable depth questions (shown after core complete)
Q10	Choice	OPT	How comfortable are you communicating in a foreign language?	English-only · I know a few phrases · Fairly fluent in one + language · Multilingual	English-only → AGT-06 language help prioritized	language_comfort
Q11	Choice	OPT	Are you part of any frequent flyer or hotel loyalty programs?	Yes, several · One or two · No, but interested · No, not important	Yes → prompt to connect accounts (AGT-09 activates)	loyalty_programspoints_optimizer
Q12	Choice	OPT	How do you prefer Tico to reach out during a trip?	Proactively (push suggestions) · Only when I ask · Minimal (emergency only)	→ sets notification_mode	notification_mode
Q13	Choice	OPT	How far in advance do you usually plan?	Days before · 1–2 months · 3–6 months · 6+ months	Last-minute → enable deal alerts	planning_horizon
Q14	Multi	OPT	Anything Tico should always avoid recommending?	Cruise ships · Casinos · Very crowded spots · Long flights · Party hostels · All-inclusive	→ sets exclusion filters	exclusions[]

Questions Q15–Q24 are deep specialization questions for specific personas (photographers, wellness travelers, business-only, ultra-luxury) — unlocked automatically based on the declared interest profile.

05

Anti-Dropout Mechanics

Every design decision targets one goal: get users to a complete profile. These are the specific mechanics implemented to minimize abandonment at each stage of the flow.

🏃

Fast first value

After Q01 + Q04 (vibe + interests), Tico shows a preview: "Based on what I know so far, you might love Porto in October." This hooks the user before they've committed.

Triggered after Q04 → preview card appears

📊

Visible progress

A thin gold progress bar at the top of the chat UI shows completion percentage. The label switches between encouraging phrases: "Almost there" / "3 quick ones left".

Never shows "Step 6 of 12" — too clinical

⏭️

Universal skip

Every non-CORE question has a "Skip for now" option rendered as a small ghost button below the main choices. CORE questions don't have skip — but they're never more than 5 options.

⚡

Tap, not type

95% of questions use tap options, not keyboard input. Free text is used only twice (dream destination, specific dietary restriction). This alone reduces dropout by ~40%.

🧠

Branch collapse

If a user answers in a way that makes 2 future questions redundant, they're automatically skipped. A budget traveler skips Q04b (luxury priorities) entirely.

✅

Micro-affirmations

After each answer, Tico acknowledges the choice naturally: "Good call — solo travel is the best way to really get to know a place." This keeps tone warm and users engaged.

Never: "Answer recorded." Always: a reaction

🔔

Re-entry on drop

If a user exits mid-flow, the session is saved. On re-entry, Tico says: "Welcome back! You're 60% done — ready to finish your profile?" — resuming exactly where they left off.

🔄

Graceful profile minimum

If a user answers only 5 questions before dropping, that's still usable. The system generates a partial profile and starts offering rough recommendations — then naturally prompts for more info at useful moments.

Dropout Risk by Stage

Stage	Questions	Risk level	Primary cause	Mitigation
Opening (Q01–Q02)	Q01, Q02	LOW	Unclear value prop	Tico explains itself in the greeting in 2 sentences max
Budget question (Q03)	Q03	MED	Privacy concern about money	Framed as "style" not "income". 4 ranges, no actual numbers
Multi-select fatigue (Q04)	Q04	MED	Too many choices overwhelm	Max 10 options, max 3 selections enforced, visual pill layout
Free text (Q08)	Q08	HIGH	Keyboard friction, blank-field anxiety	"Not sure yet" pre-filled option always visible as a tap alternative
Optional questions	Q10–Q14	HIGH	User thinks they're done	Show profile preview after Q09, then ask "Want to make it even more precise?"

06

Profile Output Schema

The complete JSON structure generated by a finished onboarding. This becomes the seed for AGT-01 and is refined continuously by every interaction.

{ "user_id": "uuid-v4", "onboarding_version": "1.0", "completed_at": "ISO-8601", "profile_confidence": 0.82, "trip_style": "culture", "adventure_level": 2, "pace": "balanced", "planning_style": "moderate", "planning_horizon_days": 45, "group": { "type": "solo", "size": 1, "is_planner": true }, "budget": { "tier": "mid-range", "priorities": ["experiences", "accommodation"], "daily_usd_range": [120, 250] }, "interests": ["local_food", "culture_history", "photography"], "dietary": ["vegetarian"], "exclusions": ["cruise_ships", "all_inclusive"], "activity_level": 3, "accommodation": { "type": "boutique_hotel", "tier": 3 }, "geography": { "preferred_regions": ["europe", "southeast_asia"], "dream_destination": "Kyoto, Japan", "typical_duration_days": 9 }, "ux_preferences": { "notification_mode": "proactive", "language_comfort": "english_only", "loyalty_programs": false }, "inferred_tags": ["solo_female_safe_areas", "vegetarian_friendly", "avoid_tourist_traps", "boutique_neighborhood"], "embedding_vector": "[1536-dim float array — stored in Pinecone]" }

Inferred tags are generated automatically from the combination of answers — not directly asked. "solo_female_safe_areas" is inferred from group=solo + budget=mid + interests=culture. These power the recommendation filters beyond what the user explicitly stated.