Growth Design
Funnel Redesign
Conversion

~20% → 50% conversion

New parents were abandoning before typing a single message. The product worked, the funnel didn't. Moving registration to after the first value moment changed everything.

A 6-screen registration wall became 2 post-value steps. No changes to the product itself. Only a sequence change and a copy reframe.

Role

Product Designe 0->1

Year

2026

Product

B2C

Platform

Progressive Web App · iOS & Android (on process)

Product Snapshot

This case study documents a funnel redesign from a 6-step registration wall to value-first conversion. bbconmigo is an AI-powered pediatric support product for parents of newborns in Mexico. New parents were abandoning before sending a single message, not because the product was bad, but because they had to prove they wanted it before they could try it.

We moved registration to after the value. Conversion went from ~20% to 50%.

50%
Conversion at registration gate
From ~20% estimated baseline
−4
Screens before first value
From 6 pre-value to 2 post-value
2 wks
Time to measurable lift
Zero infrastructure changes

Context

AI-powered pediatric support product

Parents of newborns in Mexico using Julieta for real-time guidance.

Problem

Six registration screens before help

Parents had to complete account creation before sending a first message.

Solution

Value-first conversion flow

Let parents experience Julieta first, then ask them to save the conversation.

Role

Co-founder · Lead Designer

Funnel diagnosis, hypothesis, UX redesign, conversion copy, and measurement framing.

Team

Engineering collaboration

Worked with engineering on implementation.

Outcome

50% conversion at gate

Measured 2 weeks post-launch.

Scope

Funnel redesign · Onboarding · Conversion copy

Sequence change and copy reframe, without infrastructure changes.

Context

bbconmigo connects parents of newborns with AI-assisted pediatric guidance, available at any hour. The target user is a first-time parent in Mexico, often awake at 2am, anxious about something their baby is doing or not doing.

The product has a conversational AI agent named Julieta who responds in real time, asks clarifying questions, and guides parents through common newborn symptoms.

The funnel redesign came out of a direct observation: we had a product that worked, but users weren't reaching it.

The Problem

Six screens of registration before a single word of help.

The original funnel required parents to create an account: phone number, SMS code, their name, their age, the baby's name, the baby's age in months, and two legal checkboxes. Before they could type their first message.

Every screen was a decision point. Every decision point was a place to leave.

The user's state when they open the app is fear and urgency. Every registration screen is friction between them and the answer they need. We were solving for our data collection before solving for their problem.

Core issue

The core issue wasn't the product, it was the sequence. Parents who reached the chat were satisfied. The drop-off happened entirely before Julieta said a word. That told us the product wasn't the bottleneck. The gate was.

Funnel Comparison

Before — register to access

Landing → open screen 1/6 of onboarding

CTA on landing page goes directly to onboarding.

Create account: phone number

Mexican +52 number required. Google Sign-in available as alternative.

OTP verification

6-digit SMS code, separate screen.

Parent profile: name + age

Full name and age in years. Step 1 of 3.

Baby profile: name + age in month

"Cada mes importa en esta etapa." Step 2 of 3.

Plan confirmation + legal terms

$0 for 7 days, two checkboxes required, CTA disabled until both checked.

Chat opens

First message possible only after all 6 steps complete. No value delivered until now.

After — access to register

Landing → chat opens immediately

CTA on landing page goes directly to Julieta's chat interface. No account required to start.

Parent describes the symptom

Free text, voice input, or quick-tap chips. One tap to start.

Julieta responds and follows up

Empathetic response, structured multiple-choice follow-ups, audio playback.

Screen softly gates: the save moment

"Julieta already knows what's happening with your baby." Registration is framed as saving progress.

Minimum data: name + phone only

CTA: "See what Julieta found." Two fields. The parent is already invested.

OTP → conversation continues

Chat resumes exactly where it stopped. Continuity preserved.

BeforeAfter
Estimated drop-off across 6 screens in a high-anxiety moment: ~80% of visitors never sent a message.Value delivered at step 3. Registration completed only after the parent has experienced the product.

The Design Decision

We didn't reduce friction. We moved it.

The tempting fix was to shorten the registration form fewer fields, fewer screens. We rejected it. A shorter form before value is still a form before value. The real problem wasn't length; it was sequence.

Before — registration as entry costAfter — registration as progress save
"Create an account to continue" → you must prove intent before receiving anything. The user is in a loss frame from the start."Save this conversation to keep going" → you've already received something. Registering protects what you have. The user is in a gain frame.

Behavioral principle

Once a parent has received Julieta's response with real and specific guidance about their baby, they assign value to that conversation. The cost of not completing registration is now perceived as losing something, not as the price of admission. That asymmetry is what drives 50% conversion at the gate.

Chosen

Move registration after first product value

Rejected

Shorten the registration form before first value

Rationale

A shorter form before value is still a form before value. The constraint was sequence, not only length.

Tradeoff

The product gives value before capturing all desired account data.

Implication

Registration becomes a save moment, not an entry cost.

What We Deliberately Didn't Change

The AI quality, Julieta's tone, the underlying data model, and the subscription offer all stayed the same. The hypothesis was that the funnel was the constraint, not the product.

ObservationWhat it proved
Users who reached the chat were already satisfied with the product. Qualitative feedback was positive.The problem wasn't what happened inside the product — it was what happened before it.
The quick-tap symptom chips reduced the activation energy of the first message.Instead of a blank text input under stress, parents could tap a word to begin. One tap to start is meaningfully different from a blank field.

Theory of constraints in practice

The bottleneck wasn't engagement quality or product fit. It was the gate before engagement. Fixing the right constraint produced fast, measurable results without touching the product itself.

Outcome

5 out of 10 parents who start a conversation complete registration.

The baseline was estimated at ~20% completion across the 6-screen pre-value flow. Post-redesign, 50% of users who reach the save gate complete it. No infrastructure changes. No new features. Only a sequence change and a copy reframe.

~1–2 / 10
Estimated completion before
6 required screens before value
5 / 10
Confirmed conversion after
2 screens after value delivered
50%
Registration gate conversion
2 weeks post-launch

These are early signals from a live product results. The directional signal is strong. The next measurement phase covers cohort retention and LTV correlation, whether users who convert after experiencing value retain better than those who registered cold.

Role

Product designer 0->1 of bbconmigo. Responsible for the original funnel diagnosis, hypothesis definition, UX redesign, conversion copy, and measurement framing. Worked in close collaboration with the engineering team on implementation.

Funnel audit
Conversion copywriting
Save gate pattern
Behavior design
First-principles thinking
0→1 product
AI product design
HealthTech · Mexico

Recommended