arrow_back Back Next →
Checkout UX

ML Checkout Flow

A full redesign of Mercado Libre's purchase flow reducing a 7-step checkout to 3 intelligent steps, with smart defaults that cut the time from cart to confirmation by more than half.

Role

Senior UX Designer

Year

2024

Duration

14 weeks

Deliverables

UX Design, A/B Framework, Handoff

Tools

Figma, Maze

ML Checkout Flow

ML Checkout 3-step purchase flow with smart address and payment defaults

01 Overview

Mercado Libre processes over 12 million transactions per day across 18 Latin American countries. Even a small improvement to checkout conversion at that scale represents hundreds of millions of dollars in additional GMV per year. The existing checkout flow had been layered incrementally over seven years and had grown to 7 steps, with users required to re-confirm address, shipping method, payment method, and installment plan on separate screens.

Funnel analysis showed that 68% of users who entered the checkout abandoned before completing the purchase. Exit surveys pointed to three repeated friction points: "too many steps," "I had to enter my address again," and "I didn't understand the installment options." The redesign brief was to eliminate all three.

02 The Problem

Session recording analysis of 10,000 checkout sessions identified the specific points where drop-off spiked. The largest single drop-off 34% happened on the payment method selection screen, specifically when users were presented with installment options. The existing design showed 12 installment configurations as a flat list, each with a different monthly amount and interest rate. Users were expected to mentally compare them and select the right one. Most gave up.

A secondary problem was that returning users who represent over 60% of Mercado Libre purchases were being asked to re-select their default address and payment method on every purchase despite having set preferences in their account. The system wasn't using the information it already had.

Funnel analysis

Funnel analysis 68% abandonment rate with drop-off spikes mapped by step

03 Process

The redesign process began with a checkout teardown: mapping every input field, decision point, and screen in the existing flow and labeling each as "necessary," "redundant," or "can be pre-filled." Of the 23 distinct user inputs in the existing flow, 14 could be pre-filled from account data or smart defaults without any loss of user control.

The installment problem required a different approach: instead of showing all options, the redesigned flow shows the single best option for the user's account calculated from their Mercado Pago credit limit, the item price, and their past installment preferences with a secondary "See all options" path for users who want to compare.

04 Solution

The new ML Checkout is a 3-step flow: Review (items, seller, estimated delivery), Delivery (pre-filled address, editable with one tap), and Payment (best installment option pre-selected, full order summary visible). Returning users can complete checkout in under 90 seconds without changing any defaults.

The installment display was rebuilt as a single recommendation card that shows the monthly amount, the total amount with interest, and a plain-language explanation of the plan "6 payments with no interest" rather than a table of numbers. The "See more options" link is present but de-emphasized, because most users choosing the recommended option was the desired behavior.

3-step flow

3-step flow Review, Delivery, Payment with pre-filled smart defaults

Payment recommendation

Payment step single recommended installment card with plain-language explanation

05 Results

−31%

Checkout abandonment rate

−57%

Time to purchase completion

+18%

Installment adoption rate