§ 01 · Overview
Philo's trial sign-up had a critical leak: TV and mobile-web users abandoned at the payment step. I redesigned the trial-to-paid funnel, and only the steps where people bailed. I found the friction and shipped targeted fixes. Within 6 weeks of launch, trial completion and day-8 conversions rose, and support tickets about "unexpected charges" fell.
§ 02 · Goals
Pinpoint the leak
Use session replays, funnel analytics, and support transcripts to locate where and why users quit, device by device.Design lean, targeted fixes
Attack only the highest-friction moments: identity on TV, trial terms clarity, and payment friction. No bloated redesign.Prove impact in 6 weeks
Ship fast behind flags. Success = higher trial completion, higher day-8 conversion, shorter signup time, fewer billing complaints.
§ 03 · My role
Led the Signup & Trial Conversion workstream. Owned research, flows, UI, copy, experiment design, and KPI definition. Partnered with PM on scope and with Eng on rollout and instrumentation.
§ 04 · Team
1 PM, 3 Engineers (TV, Web, Platform), 1 Data Analyst. I led design; Eng implemented; CX fed qualitative signals.
§ 05 · Problem
Problem
Data showed a clear bottleneck:
55% abandonment on TV at the "email + payment" step
31% abandonment on mobile web at the same step
Support tickets spiked around "When will I be charged?" and "I thought this was free."
Root causes:
Identity friction on TV (typing emails with a remote; code-email mismatches).
Trial terms buried; users didn't see first-charge date or cancel-anytime up front.
Payment friction on mobile web (card-first, weak wallet surfacing; ambiguous auth/hold messaging).
I narrowed scope to three interventions that would move numbers without ballooning build time.

§ 06 · Research
Research
Methods
Session Replays (TV + mWeb): mapped rage-clicks, backtracks, and exits at field-level.
Funnel Analytics: segmented by device and payment method; tracked time-to-complete.
Support Transcripts: coded "billing expectations" complaints; pulled top phrases.
Rapid Interviews (n=12): recent trial starters who failed vs. succeeded.
Payment Outcome Audit: compared auth success and retries by method.
What I found
TV identity was the choke point. Users dreaded email entry with a remote; device codes expired mid-flow.
Expectation gap on money. Users didn't see trial length, first-charge date, or hold behavior until late.
Wrong payment default. Card-first flow on mobile web ignored available wallets; higher friction and slightly worse auth.
Language problem. "Authorization" sounded like "I'm being charged now."
Personas: research snapshot
| Persona | Description | Pain Points (Observed) | Design Moves |
|---|---|---|---|
| Remote-Averse TV Starter | Starts trial on TV with a remote; low tolerance for form entry. | Email typing on TV; device codes expiring; backtracks at identity step. | QR → mobile handoff; 10-min pairing window + refresh; state save on fail. |
| Wallet-First Mobile User | Starts on mobile web; prefers Apple/Google Pay/Cash App. | Card-first flow; extra fields; higher auth failures on small screens. | Wallets surfaced first; real-time validation; clearer error states. |
| Skeptical Trialer | Price-sensitive; wary of hidden charges. | Trial length/first-charge date buried; "authorization" wording read as "charge now." | Up-front trial panel (length, charge date, price); plain-language hold copy; sticky summary. |
| Household Manager | Sets up service for family; may bounce between devices. | Loses progress switching TV ↔ phone; repeats steps after timeouts. | Cross-device state continuity; progressive disclosure; longer code TTL. |
| Browsing Channel Checker | Exploring lineup before commitment. | Perceives signup as long; uncertain value until after paywall. | Faster trial start with minimal fields; lineup preview retained during flow. |
Friction Map → Evidence → Fix
| Step | What We Saw | Why It Mattered | Targeted Fix |
|---|---|---|---|
| TV: Identity | 55% abandonment at email/payment; code expiry spikes exits. | Remote text entry is high-friction; time pressure amplifies quits. | QR handoff to mobile; extend/refresh codes; countdown feedback. |
| mWeb: Payment | 31% abandonment; wallet availability ignored; auth errors cluster on cards. | Extra fields + poorer auth = leaks right before conversion. | Wallet-first order; adaptive retries; clearer inline validation. |
| Money Expectations | Support tickets: "I thought this was free" / "When is the charge?" | Surprise costs drive churn, chargebacks, and mistrust. | Top-of-flow trial panel (dates, price); plain English hold explainer; sticky recap. |
| Error Recovery | Users who hit one error were 2.3× more likely to exit. | Single failure often ends the session on TV/mobile. | Save/restore state; focused error copy with one-tap retry paths. |

§ 07 · Design + Development
Design + development
A) Cross-Device QR Handoff (TV)
Replaced email entry on TV with QR → mobile handoff: scan code, complete identity + payment on phone, auto-pair back to TV.
10-minute pairing window with live countdown and one-tap "refresh code."
State save across devices; no restart if a step fails.
Why it works: kills remote typing and timeouts; users finish on the device that's fastest for forms.
B) Trial Terms, Up Front (All Devices)
New top-of-flow money panel:
"Free for 7 days"
"First charge on [DATE] at $X"
"Cancel anytime. No fees."
Inline tooltip explaining auth holds vs charges in plain English.
Sticky summary persists through checkout; no surprises.
Why it works: sets expectations before commitment; reduces last-second fear.
C) Wallet-First Payments (mWeb + Hand-off Flow)
Reordered methods: Apple Pay / Google Pay / Cash App Pay surfaced first when available; card remains a secondary option.
Real-time validation and clearer error states.
Soft-retry + card updater enabled server-side (Eng) with transparent copy client-side.
Why it works: fewer fields, higher auth, less friction on small screens.
D) Plain-Language Microcopy + Trust
Rewrote money copy: "We'll place a small temporary hold to confirm your card. You won't be charged until [DATE]."
Added network and compliance badges only where anxiety spiked (payment step), not everywhere.
Why it works: confidence without clutter.
E) Instrumentation & Experiment Design
New event taxonomy (
trial_viewed,trial_charge_date_seen,payment_method_surface,auth_result,handoff_start/success).Device-specific funnels to isolate TV vs mWeb.
Shipped behind flags; 50/50 audience split, powered for +5pp conversion detection.

§ 08 · Results
Results
6 weeks after launch:
- Trial completion +17 pts
- TV signup abandonment −27 pts
- Day-8 trial to paid +12 pts
- Average TV signup time −67%
- Payment auth success (mobile web) +3.8 pts
- Billing-expectation tickets −37%
- Chargebacks per 1k trials −19%
