Purchase Flow • Alfa-Market

1. About the project

Summary

Designed the in-app purchase flow for the Podeli app from scratch — from adding items from different sellers to the cart through order checkout and management.

Using CJM, competitive analysis, and UX testing, I found a clear structure for the multi-seller cart, accounted for the technical constraint of separate checkouts, and prepared the MVP for launch across two channels — Podeli and the Alfa-Bank app.

The new flow let the product move from a clickout model to a full marketplace that earns commission on completed orders.

Success metrics

Context

Alfa-Bank is Russia's largest private bank, building not just banking products but its own digital ecosystem — payments, installments, travel, and other services. Alfa-Bank's official website

Alfa-Bank's strategy included a goal of claiming a place in the e-commerce market to diversify revenue streams. Since the food segment within the Alfa Group ecosystem was already covered by X5 Group, the new marketplace focused on clothing, electronics, cosmetics, and home goods.

About the B2C and B2B tracks

The bank took a comprehensive approach. In 2025, a B2B track was created to build solutions for sellers:

• managing major marketplaces — Wildberries, Ozon, Lamoda, and others — from a single dashboard;
• building the seller's own online store.

For this ecosystem, an independent marketplace — Alfa-Market — was built to let sellers avoid the steep commissions charged by major marketplaces.

Business Challenge

The bank already had a product called Podeli, which worked on a clickout model: users found a product in the app but completed the purchase on the partner's website.

The business decided to turn Podeli into a full marketplace and change the monetization model — moving from being paid for referral clicks to earning commission on completed orders. The marketplace also needed to become available in a WebView inside the Alfa-Bank app.

The shared value proposition across both channels: buy in installments with no markup and earn cashback through Alfa-Bank's loyalty program.

Launching in-app purchases required designing checkout and order-management flows from scratch. I owned this workstream.

User specifics

The Podeli audience

Women aged 25–45 who shop on marketplaces regularly. Analytics revealed three segments:

• Shopper — 55% — browse and buy products.
• BNPL — 34% — manage installments and payments.
• Mixed — 11% — do both.

The cart was built primarily for Shoppers, but it couldn't break the familiar BNPL flows.

The Alfa-Bank audience

Already familiar with how cashback works and expected transparent terms for earning it. In the WebView, we emphasized the discount and the ability to make a good deal right inside the banking app they already used.

Constraints

To shorten time to market, I designed the MVP and the target version in parallel. The main technical constraint early on: items from different sellers couldn't be checked out as a single order.

2. Solving the problem

Research before the redesign

Research with 11 respondents

We conducted interviews with 11 respondents:

• Age — 6 people aged 25–34, 5 people aged 35–45.
• Marketplace — Wildberries (5), Ozon (4), Yandex Market (2).

All shop online regularly, live in cities of different sizes, and don't work at Alfa-Bank.

High-priority problems found in the Podeli app

The solution

CJM

Based on the research, NN/g recommendations, and technical constraints, I mapped the target purchase journey.

The CJM helped treat the cart not as a standalone screen but as the connecting step between choosing a product and checking out with different sellers. It organized the known and potential barriers that the design needed to address and that UX testing needed to validate.

Competitive analysis

Solution options

Because of the technical constraint, each seller's items had to be checked out as a separate order. At the sketch stage, I worked through two options for the multi-seller cart: one split items across store tabs, the other displayed them on a single page in separate blocks, each with its own checkout button.

Both approaches had their own advantages and risks that needed to be validated with users.

Option 1 — Tabs

Each seller's items live in their own tab. The user switches between stores and checks out the current order through a single primary button.

Option 2 — Single screen

Items are grouped into vertical blocks by seller. Each store has its own checkout button.

Research to choose an option

We ran moderated UX testing of both multi-seller cart options. Participants were asked to build a cart and check out; on entering the cart, they saw the two concepts — seller tabs and a single-page layout — in a randomized order.

We observed how easily users found items they'd added, changed quantities, understood the seller split, and moved on to checkout.

Key result: 100% of comparison-test participants preferred the unified list split by seller. The tabs option hid part of the cart and caused people to lose items they'd added.

Read the full research

Working through additional states

On the product card, I accounted for

• links to the product and the seller,
• showing unit price when quantity changes,
• a low-stock warning.

For the empty cart, I designed a way forward: the user can return to shopping or see personalized recommendations. This keeps the screen from becoming a dead end and helps them get back to browsing.

Removing items

I analyzed how competitors handle removing items and saving them to favorites. For quick actions, I added swipe — a familiar mobile gesture.

To let users undo an accidental removal, I considered three solutions:

1. An "Undo" button with a timer keeps the removed item's context visible and clearly shows how much time is left to cancel.
2. A snackbar with an "Undo" action doesn't block the interface, but the pattern itself often reads as a system message or error.
3. A confirmation modal prevents accidental removal but interrupts the flow and requires an extra action every time.

In the end I went with the "Undo" button with a countdown. It doesn't slow down intentional removal, stays visible in the item's context, and gives users a predictable way to fix a mistake.

Purchases and payments in one section

Previously, the "Purchases" section only showed installment orders. With the marketplace launch, I needed to add delivery information while keeping the familiar payment-tracking flow.

I considered splitting it into two separate items — "Orders" and "Payments" — but that complicated navigation and broke the familiar flow. So I merged deliveries and payments inside "Purchases".

I placed deliveries in a horizontal scroll above payments. Users immediately see the status of current orders without it crowding out upcoming payment info. Order details include a pickup QR code, pickup point address, order number, payment method, and any required actions.

4. Results after launch

Metrics we're tracking:

• increased conversion into add-to-cart, checkout start, and completed purchase.
• MAU growth;
• GMV and commission revenue growth;

We're continuing to gather data to assess the new flow's impact on user activity and business metrics.

Key Learnings & Reflection

What worked

• Found the right cart structure. UX testing showed the unified list was clearer than tabs: users saw all items and understood the seller split better.
• Accounted for real user behavior. Since the cart is often used to store items, I added partial selection — users can check out the items they want and keep the rest.

What's outside the MVP scope

• A single checkout across sellers. Due to technical constraints, each order was checked out separately. Users understood the mechanic but experienced it as extra friction.
• Part of the target flow. To shorten time to market, the first version shipped the core functionality, leaving the rest of the improvements in the backlog.

What I'll carry into future projects

• Test information architecture against real content: a familiar pattern may not fit a different user context.
• Sync with engineering on technical constraints earlier, and split the MVP from the target flow from the start.

Next steps

• Merge checkout across different sellers into a single flow.
• More actively communicate the loyalty program's benefits to Alfa-Bank customers in the marketplace.
• Measure how changes affect conversion, repeat purchases, MAU, GMV, and product revenue.