Role (simulated)
Product Manager: Growth & Marketplace
Goal
Close the payment loop: protect group LTV
North Star
+12–18% group 30-day reorder rate
Guardrail
Split completion rate ≥80%
Timeline to v1
~6 weeks to A/B-testable
One person pays.
Three people forget.
One occasion doesn't repeat.
Three people forget.
One occasion doesn't repeat.
This is not a UX improvement: it's a retention problem. Uber Eats built the infrastructure for group ordering but left the payment loop wide open. Every open loop erodes the loyalty of the one person most likely to organise next time: the initiator.
1
Marketplace Thinking: Why This Is a Platform Problem
Uber Eats is a tri-sided marketplace. The payment loop failure doesn't just affect customer satisfaction: it damages marketplace health at every level.
🏪
Restaurants
Want larger, frequent group orders: their highest-value tickets.
Avg group: £57
→
📱
Uber Eats Platform
~30% commission. Group orders are 3.4× more valuable. Payment loop is wide open.
Loop OPEN: gap
→
👥
Customers
Initiator bears all financial risk. Post-meal payment friction suppresses reordering.
~23% lower reorders
The marketplace insight
Uber Eats solved ordering coordination in 2022 (Group Order link). It left payment coordination entirely to the customer: specifically to the initiator most likely to drive repeat group occasions. This is a marketplace health problem, not a product convenience feature.
2
Unit Economics: Why Group Reorder Rate Matters
Solo order LTV (12 months)
£204
Avg solo basket £17 × ~12 orders/year. Platform take ~30% = £61 revenue per user per year.
Group initiator LTV (12 months)
£684
Avg group basket £57 × ~12 occasions/year if retained. Platform take = £205/year: 3.4× the solo value.
Impact of 15% reorder uplift
+£30
Per group initiator per year in additional platform revenue. Across hundreds of thousands of initiators, this is meaningful growth from a single feature.
3
Why This Problem First
Business impact
Group initiators generate 3.4× the LTV of solo users. Even a 15% reorder uplift in this cohort has outsized revenue impact vs any solo-side improvement.
User pain severity
100% of bill splits happen outside the app today. The failure is total, predictable, and happens every single group occasion: no partial fix exists.
Build confidence
Group Order cart APIs have existed since 2022. This is incremental build on proven infrastructure: high confidence, medium effort.
4
Problem Quantification
~23%
Lower reorder rate: group vs solo. This gap is the business problem. Payment friction is the primary hypothesis for why it exists.
Hypothesis from user research
100%
of bill splits happen outside the app today. Zero in-app payment resolution exists: confirmed by feature audit.
Product gap: confirmed
~8 min
Average time wasted per group order on post-meal payment resolution. 8 minutes of friction associated with Uber Eats every occasion.
Estimated from user interviews
5
Hypothesis
🧪
Core Hypothesis
If we close the payment loop inside Uber Eats before the order places, group payment friction will be eliminated and group 30-day reorder rate will increase by 12–18%. The initiator's experience improves enough that they organise the next occasion willingly.
6
RICE Prioritisation: Why This Feature, Not Others
| Option | Reach | Impact | Confidence | Effort | Score | Chosen? |
|---|---|---|---|---|---|---|
| Pre-order split bill (this) | ~30% of orders | High: closes root cause | High | Medium (2–3 sprints) | High | ✓ Yes |
| Post-order payment request | Same | Low: doesn't close loop | Low | Low (1 sprint) | Low | No |
| In-app payment wallet | All users | High but diffuse | Low: FCA complexity | Very High | Mid | Phase 3 |
| Better group notifications | All group orders | Low: doesn't fix root cause | High | Low | Low | No |
| Saved groups / social graph | Repeat groups | Medium: convenience layer | Medium | Medium | Mid | Phase 3 |
7
The Solution: Pre-order Split Bill
🚧 Current State
- One person fronts £40–80+ with no in-app recovery
- Bill split outsourced to Monzo / Revolut / WhatsApp: every time
- No audit trail: disputes resurface days later
- "Organiser tax" suppresses willingness to arrange next occasion
✦ Pre-order Split Bill
- Native split flow at checkout: before the order places
- 3 modes: Even Split: By Item: Custom %
- Payees pay via web link: no app install required
- Delivery fee + service charge split proportionally
- 30-min window + initiator override for non-payers
ubereats
Your Order
🍔 Smash Burger£12.99
🌮 Tacos × 2£16.50
🍟 Fries£4.49
🥤 Drinks × 3£8.97
Total£57.41
Place Order
😰 No split option: payment leak every time
Before
✨ SPLIT BILL
4 people splitting
Each pays their own share
Y
J
P
S
By Item
Even
%
You
Burger + Fries
£14.98
Jack
Tacos + Drinks
£20.50
Priya
Cheesecake + fees
£21.93
✓ Send Payment Requests
Split screen
Payment Status
Confirms once all 4 pay: 29 min left
You: £14.98
Paid ✓
PAID
Jack: £20.50
Awaiting
Nudge
Priya: £21.93
Paid ✓
PAID
Sam: £0
Paid ✓
PAID
3 of 4 paid
Waiting for Jack
Tracking
8
Tradeoffs: The Decisions That Actually Mattered
| Decision | Rejected | Chosen | The honest tradeoff |
|---|---|---|---|
| Split timing | Post-order request | ✓ Pre-order split | Post-order doesn't close the loop: someone still fronts the bill. The harder path is the only right path. Cost: ~1 extra sprint; 30-min window edge cases where initiator must decide to cover or cancel. |
| Payee experience | App install required | ✓ Web payment link | Forcing a download to pay £14.98 kills completion rate. Completion rate is the guardrail: we don't compromise it. Cost: we lose the app acquisition opportunity. Acceptable trade. |
| v1 scope | All 3 modes at launch | ✓ Even Split first | Even Split validates the core hypothesis cheaply. If completion ≥80% and reorder delta is positive, By Item is justified. Build cheap, learn fast. Cost: office teams get even split; friend groups wait for By Item. |
| Non-payment handling | Block order indefinitely | ✓ 30-min + initiator override | Indefinite blocking punishes the whole group. Time window creates urgency without cruelty. Cost: initiator carries residual risk: acceptable because it's their active choice. |
9
What I Would NOT Build in v1
🚫 In-app payment wallet
FCA regulation, fraud infrastructure, trust build: a multi-quarter programme. Web link delivers the same job at 10% the complexity. Wallet is Phase 3 after validating that split checkout moves reorder rate.
🚫 Post-order split request
The "easy" version that doesn't solve the problem. Someone still fronts the bill. Building this is optimising for effort over impact: the classic PM mistake.
🚫 Saved groups / social graph
Only worth building after the retention hypothesis is confirmed. Social features are expensive and hard to undo. Validate the mechanism first.
10
Execution Reality
Sprint Plan: 2 engineers + 1 designer + PM
Sprint 1 (2 weeks)
Even Split + web payment link + payment tracking screen. Ship to 5% of group order users. Track: completion rate + 30-day reorder delta.
Sprint 2 (2 weeks)
By Item mode + item assignment UI: only if Sprint 1 completion rate ≥80%. Nudge: "3 people in your order: split it?"
Critical dependencies
FCA legal review on payment facilitation. Delivery partner payout timing (can't wait for all 4). Payment edge cases: partial completion, chargebacks.
Engineering
Payment state machine
Biggest risk: payment edge cases. What if 3 of 4 pay and 1 fails at T+29min? Explicit error states must be defined before launch.
Legal / Compliance
FCA payment facilitation
Web link approach likely keeps us outside e-money regulation vs in-app wallet. Must be confirmed before Sprint 1 ships.
Finance / Ops
Revenue recognition
How does Uber Eats recognise revenue when 4 people pay at different times? Delivery partner payouts can't wait: needs a defined float mechanism.
11
Metrics: Connected to Revenue
North Star
+12–18%
Group 30-day reorder rate delta for split completers vs matched non-split control. Directly maps to group initiator LTV.
If flat: payment friction isn't the root cause. Investigate before doubling down.
Guardrail
≥80%
Split sessions where all participants pay within 30 mins. Below 65% = web link trust too low or window too short. Immediate investigation trigger.
Most critical operational metric in first 4 weeks.
Leading Indicator
+8–15%
Average group order value: if bill anxiety suppressed larger orders, removing it should unlock them within 4–6 weeks.
Early signal that behaviour has changed.
Case Study 01 of 03: Flagship: Marketplace Thinking