Designing the purchase decision

Situation

The cart showed signs of distrust

This project originated in the professional practice workshop at CASA UX, a training and mentoring space led by Ludmila Disa and Lucre Gandolfo. I worked with Bárbara Montes, UX/UI Designer, and Melany Moglia, UX Researcher, on a fictional Amazon case study.

The starting point was a PRD describing a business and experience problem: cart abandonment was rising and more people were returning from the cart to the product page before completing the purchase. That regression suggested the flow was not solving a critical need at the most sensitive moment in the decision.

• 62% of people built a cart and did not buy.
• 41% returned from the cart to the product page.
• 9% was the increase in abandonment rate at the final stage of the journey.

As a team, we decided to focus on the mobile experience and review three screens end to end: product page, add-to-cart screen, and cart. The goal was to understand what users need in order to decide whether to buy and propose improvements that reduce friction.

Task

Translate the problem into actionable research questions

Before thinking about solutions, we worked on problem framing: breaking down assumptions, identifying biases, and turning the problem into concrete design and research questions. The task was to identify which signals users needed to validate their purchase confidently and how to reorganize the flow to reduce cognitive load without distracting from the main objective.

From the initial review, we formulated a central hypothesis: the current flow was not providing critical information at the right moment, which created insecurity, doubts, and a lack of confidence to move forward to checkout.

That hypothesis became a very concrete research goal: to understand what information, validations, and signals people need during cart building and review in order to decide whether to complete or abandon the purchase.

• Why were people returning to the product page after adding an item?
• What information did they consider essential to make the final decision?
• Was the friction caused by missing information, by hierarchy, or by the moment in which it appeared?

Action

Heuristic analysis, research, and full-flow redesign

The work progressed in layers: first, we analyzed the current flow through Nielsen's heuristics, UX laws, information hierarchy, and language; then we conducted desk research and benchmarking and added evidence through a survey with real users; finally, we redesigned the screens and validated the proposal in a functional prototype.

1. Diagnosing the current flow

The heuristic review showed that the problem was a combination of visual density, ambiguous signals, poorly prioritized decisions, and lack of context at critical moments in the journey.

• On the product page, the title had weak hierarchy, there was too much text about discounts, payments, and shipping, and the “Add to cart” and “Buy now” CTAs competed with very similar visual weight.
• On the add-to-cart screen, basic information such as title, quantity, or product price was missing; in addition, the warranty cross-sell introduced new concepts at a critical moment and distracted from the main action, which was moving forward with the purchase.
• In the cart, the content did not prioritize final validation: there was too much text to communicate simple benefits, and the CTA “Proceed to order” could be interpreted as immediate payment rather than as a step toward checkout.

←

→

We found issues with heuristic 4 (Consistency and Standards) when the flow changed terminology between screens to talk about warranty or protection; with heuristic 8 (Aesthetic and Minimalist Design) because of text overload and the difficulty of isolating what mattered; and with heuristic 1 (Visibility of System Status) when people could not clearly see what had been added, how much it cost, or what the next step was.

We also leaned on Miller's law to reduce mental load by grouping related information into blocks; the affordance principle to review which elements actually read as actions; and the relationship between visual hierarchy, proximity, and recognition over recall to make more obvious what people needed to confirm before buying.

2. Desk research and market benchmark

Before proposing improvements, we conducted desk research and benchmarking to understand how other marketplaces handled this same moment in the journey. We focused especially on the mobile flows of Mercado Libre, eBay, SHEIN, Walmart, and TikTok Shop.

That analysis helped us compare add-to-cart patterns, purchase review, price and discount hierarchy, shipping treatment, presence of reviews, cross-sell, and CTA clarity. The goal was to understand which conventions were already established in the market and where there were opportunities to reduce friction.

3. User research to validate the problem

To gain insights from the live product, we ran a survey with real Amazon users. The data reinforced something key: the doubts were not so much about purchase intent as about the need to validate information before moving forward.

• 100% of respondents had abandoned a cart at least once.
• 70% identified delivery times as key information for deciding.
• 50% highlighted shipping cost as critical information to complete the purchase.

The survey revealed very clear findings: people needed to validate costs and delivery times, they returned to the product page to recheck variants and details, visual density made it harder to identify what mattered, cross-sell competed with the main task, and reviews worked as a trust signal even within the cart context.

4. Narrative structure of the flow

Beyond information architecture, we worked on the narrative structure of the experience: what the user needs to see, when, in what order, and with what priority in order to move forward confidently.

We identified that the current cart was functioning mostly as a transactional instance, when in reality people were still deciding. So we redefined the approach: the cart should support validation, comparison, and trust-building before checkout.

• Is this the right product? Variant, photos, details, and rating.
• How much will I really pay? Final price, shipping, taxes, and discounts.
• When will it arrive? Delivery date, speed, and availability.
• Can I trust it? Reputation, returns, clarity, and consistency.
• Am I ready to complete the purchase? Clear CTA, summary, and an obvious next step.

5. Redesigning the three screens

With that evidence, we redesigned the architecture and information hierarchy of the full flow. Each screen needed to answer the question the user was asking at that moment, without forcing them to go back.

• Product page: we reordered information to prioritize photo, price, title, rating, discounts, free shipping, variants, and delivery date. Reviews became more prominent because they emerged as a trust input in the survey.
• Add-to-cart screen: we made it visible which product had been added, how many units, and what the next CTA was; we also moved promotions and discounts to a more appropriate point in the flow.
• Cart: we reorganized content to highlight the screen title, key information for each product, related offers, and a more legible order summary with clearer subtotals.

We also worked on better use of white space to highlight what mattered most according to the information hierarchy analysis, along with clearer color semantics and typographic size and weight to guide reading. For example, free shipping was highlighted in a light but noticeable green, while the discount percentage used a red accent to convey urgency and support conversion without competing with the product's core information.

Several heuristics were resolved more clearly in our proposal: visibility of system status by showing what was added and how to proceed; consistency by unifying copy and criteria across screens; the match between system and real world by making signals like discounts and benefits more recognizable; and flexibility and efficiency of use by allowing product details to be reviewed without leaving the cart.

←

→

6. Two new features to reduce regressions and add value

In addition to the base redesign, we proposed two specific features to address behaviors observed in the flow and mentioned in the PRD.

• Product detail modal inside the cart: allows people to review characteristics without going back to the full product page, which can reduce regressions to that screen.
• Delete-product dialog: written so title, body, and CTA worked together, offering control and an option to save the product for later.

In both features, we also applied heuristic criteria and UX principles: user control and freedom, visibility of system status, and stronger affordance so the interface expressed more clearly what would happen after each action.

7. Functional prototype and usability test

We took the proposal into a functional Figma prototype and tested it to understand whether it really made purchase validation easier and increased confidence to continue to checkout.

The hypotheses we wanted to validate were:

• The new cart makes it possible to validate the purchase with greater clarity and lower cognitive load.
• Users better understand costs, discounts, and shipping conditions.
• Quick access to relevant information reduces the need to return to the product page.
• The suggestion to save products before deleting them adds value without creating friction.

Result

A more consistent experience and concrete learnings

Because this was a workshop, the outcome was not a production implementation but a tested prototype that allowed us to validate part of the proposal while also identifying new opportunities for improvement.

The main advance was having turned the challenge proposed by Product in the PRD into a system of design decisions backed by evidence: going back to the product page was not a user whim, but a symptom of insufficient or poorly prioritized information in the cart.

The test surfaced very useful learnings for the next iteration:

• The image change when switching colors needed better visual feedback.
• The price summary still created confusion around discounts and shipping.
• Access to product details from the cart was well received.
• The “save” action was not always interpreted as clickable.
• The delete modal could likely be resolved better with a lighter inline alternative.