Skip to content

Rebuilding the app’s purchase funnel

The app sold park extras through one generic form that didn't explain what you were buying. I led its redesign into a funnel that explains each product, and the components we built for it became the start of the Skyloop Design System.

Three phones showing the new funnel: product detail, the Express catalogue and confirmation

PortAventura World

Client
PortAventura World
Role
Product designer, project lead
Team
2 product designers, 1 design lead
Period
May 2024 – February 2025
Reading time
6 min
The problem
Express passes, show seats, food and rentals were all sold through the same generic form, which didn't explain what you were buying.
What I did
I led the redesign end to end: a month of research with guests, a product page and catalogue that explain each product, and a flow built for the web so park QR codes could open it.
The outcome
Purchases reaching confirmation rose from 44% to 62%, and the components became the start of the Skyloop Design System.

The PortAventura World app sells extras for your day in the park: Express passes to skip the queue, Prime Seat for the front rows at the shows, and a long tail of in-park products. In 2024 all of them went through one generic form, and fewer than half of the purchases people started reached the end. I led the redesign from the first interview to the shipped screens, with another product designer beside me and the design lead approving each flow.

The problem

The old funnel was a generic form. Date, park, pass: the same numbered steps for every product, and the product was assembled from whatever you ticked. Nothing explained what you were buying, and nothing looked like PortAventura.

The old Express purchase: date and park preset, options as long radio labels
The old Express purchase: date and park preset, options as long radio labels
The same screen with the selection sheet open
The same screen with the selection sheet open
Booking a seat for a show in the old flow
Booking a seat for a show in the old flow

That one form sold Express passes, Prime Seat, horror passages, food ordered ahead, table bookings, virtual reality and stroller rentals. Each has its own rules, and the form explained none of them. Two passes could differ by two words in a long radio label, with the detail behind an info icon and an asterisk, and changing the date, the park or the show meant finding a small text link.

The price breakdown lived in a sheet at the bottom of the screen. It grew to a third of the screen’s height, and people didn’t realise they could open or close it. In the last months of the old funnel, 44 of every 100 purchases started reached confirmation.

What the research changed

I ran the whole double diamond, and the first half took a month: interviews with guests in the park, user journeys and affinity maps, and a benchmark of the apps people already buy from on their phones (Uber Eats, Just Eat, Airbnb) and of ticket sellers like Atrápalo. The other designer reviewed each round with me, and the design lead played devil’s advocate on the high-fidelity screens.

User journey map for a family persona, from awareness to loyalty: steps, actions, goals, feelings, pain points and opportunities
User journey map for a family persona, from awareness to loyalty: steps, actions, goals, feelings, pain points and opportunities

Three findings became the brief. Products had to be explained properly, because guests couldn’t tell what one included or how it differed from the next. The purchase had to follow patterns people already know from the apps they order food and book trips with, so nothing needed learning on a park day. And it had to look and feel like PortAventura, which the form never did.

Before launch we tested the new flow with guests: 45-minute guided remote sessions where each one completed a full Express purchase, then compared two versions of the shopping page and two ways of grouping food and drink.

Shopping page, version A: every food option at the top level
Shopping page, version A: every food option at the top level
Shopping page, version B: one Food & Beverage entry
Shopping page, version B: one Food & Beverage entry
Grouping, version A: each product type on its own page
Grouping, version A: each product type on its own page
Grouping, version B: all options on one page
Grouping, version B: all options on one page

The purchase steps were clear and quick enough for a day at the park. Food and drink wasn’t: people wanted the category first and the products after, each type of product on its own page, and before picking a restaurant they wanted to know the park area, the type of food, whether there was somewhere to sit and the price. A few missed the info icon, and a few asked for Apple Pay. That became six recommendations for the next iteration, starting with a single Food & Beverage entry.

Constraints

The funnel had to fit many product types into one concept, each with its own rules: time slots, quantities, show times, rentals. We had one to two months to deliver, few screens and a lot of cases. The app runs in four languages across three parks, over a third of its sessions aren’t in Spanish, and each park sells its own products. And the purchase had to work outside the app too, because guests would reach it from QR codes on the park’s signs and screens.

Decisions

Build the flow once, for the web

Everything we had was React Native, inside the app, and guests would also open a purchase from a QR code on a sign. I sized how far the flow would have to scale and we chose to build it in React for the web: one flow that opens inside the app and in a browser in the park. None of the app’s components existed there, so we designed them from scratch, and those components became the first pieces of the Skyloop Design System.

Explain each product on the first screen people see

The research kept pointing at the same gap: people couldn’t tell what a product included. The team wanted products selected only from a detail page, where there was room to explain them; PortAventura wanted them added straight from the catalogue. I worried a one-line card couldn’t carry a product with this many conditions, so we designed for both.

The catalogue card had to be clear enough to buy from, and the detail page had to answer every question: what’s included, the height requirement, whether there’s a photo ride, and quantity and time-slot selectors that show the price of each slot and warn you when it’s sold out or nearly gone. Multi-selection with time slots inside a catalogue made it the hardest part of the MVP.

Then the data came in, and we were wrong: selecting from the catalogue worked well for these products. We changed our minds.

Product detail: what it is, requirements, quantity and time slots with availability
Product detail: what it is, requirements, quantity and time slots with availability
The summary: every product with its image, price and quantity
The summary: every product with its image, price and quantity
Confirmation: next steps and instructions for each product
Confirmation: next steps and instructions for each product

Give the summary and the confirmation real content

In the old flow the price breakdown hid in a sheet nobody knew how to open, and the purchase ended with a total and a button. In the new one, the summary shows each product with its image, the price per person and day and a quantity you can still change, plus a line about secure payment and where your tickets will be saved. The confirmation lists your next steps, gives instructions for each product you bought, and tells you where to find your tickets later.

Result

The new funnel in the app: from the home screen to the catalogue, a product’s detail and the cart

We shipped in February 2025, nine months after the first interview. Before, 44 of every 100 people who started a purchase reached confirmation. In the months after launch, 62 did. After all those months working out how to explain a stroller rental, it was good to see.

Since 2024 the app’s revenue has grown about 2.6 times, as the old purchase paths were replaced by this funnel and the ones that came after it. That’s the team’s result, and the funnel was one part of it.

What I’m proudest of is what the project left behind. The components we built in a hurry became the first pieces of the Skyloop Design System we use today, and the catalogue pattern later made its way into the ticket funnel.

What I’d do differently

I’d start the design system with its foundations. With one or two months to deliver, we built the complex components first, like the product card and the quantity selector, without defining the smaller pieces they’re made of. That created debt, and we laid the foundations later, under components that already existed. Next time I’d define the atoms first, even with that deadline.

I’d also test a pattern again before reusing it for a different product. Catalogue selection worked for extras, so the ticket funnel adopted it to stay consistent. Tickets have far more to explain: two-day tickets, two parks, which parks each ticket includes, whether you can bring food in, and how many consecutive days you have to use them. A single line on a card can’t hold all that, and the interviews we’re running with guests now back this up.

  • 44% → 62% reached confirmation

    Share of started purchases that made it to the confirmation screen, before and after launch.

  • ~2.6× app revenue

    Growth from 2024 to today. A team result: the funnel was one part of it.

  • 1 month of research

    With guests in the park, before anyone drew the first wireframe.