Penfolds: Navigation and Checkout Redesign

This was my capstone project as a UX design student at Swinburne University, run with Penfolds, part of Treasury Wine Estates, where I had previously worked. It ran for about 12 to 13 weeks.

It was not a paid engagement, and I want to be straight about that. What it was: a real brief, on a real commercial site, with the Penfolds team sharing their Google Analytics data and sitting through the presentation at the end. The constraints were real even if the contract wasn't.

The scope was three connected surfaces: the global navigation, the mini-cart, and the checkout.

Penfolds sells a large, deep catalogue: wines by range, by region, by variety, by vintage, plus cellar doors, restaurants, events, clubs and a century of brand history. All of it had found its way into the global navigation.

The result was a menu where categories expanded into more categories, which expanded again. Every item that went in made the menu more complete and every item made it harder to use.

Four problems compounded: cognitive overload from too many simultaneous options; a cluttered presentation working against a brand built on restraint; a structure that fought the mobile devices most people were arriving on; and reduced discoverability, where important pages were harder to find precisely because so much surrounded them.

The mini-cart and checkout had the second half of the problem. A multi-step checkout, limited payment options, no visible signals that the transaction was secure, and mandatory account creation before purchase. Each one is a reason to leave a cart behind.

I ran interviews and surveys with people from a range of backgrounds, looking at how they browse and buy online generally, not only on Penfolds. Alongside that: competitive analysis, published case studies, and the Google Analytics data Penfolds provided for their own site.

The analytics gave me the other half: which parts of the site people actually used. That turned navigation from a matter of taste into a matter of evidence.

I re-mapped and re-prioritised the menu against the analytics rather than against the org chart. Five top-level items survived: Grange, Our Wines, Experiences, Our Story, and Login.

The footer stopped being a dumping ground and became the second half of the information architecture: the place for depth, once the menu was handling breadth.

The existing cart used an X to close the panel and an X to remove an item. Two identical controls, inches apart, with very different consequences. One dismisses the cart, the other deletes something you chose to buy.

I replaced the destructive one with the word Remove. The X now means only one thing. It is a small change and it is the one I would defend hardest: the cost of getting it wrong is a customer losing a selection they had already committed to.

The significant product decision was removing the requirement to create an account before buying.

The business case for forced registration is real: an account captures an email address, supports marketing, and builds a customer profile. It is not an irrational thing to want. But published research on checkout abandonment puts the cost of compelling it at roughly a quarter of would-be purchasers, and a customer who leaves gives you neither the sale nor the email address.

So the account became optional, and the newsletter became a checkbox inside the flow. The business still gets the address from people happy to give it, and stops trading sales for the ones who are not.

Mobile was not a narrower version of the desktop design. Two decisions diverged deliberately.

Search moved to the top of the menu. Interviews were unambiguous that search carries more weight on a phone, so opening the hamburger puts search first, expanding into a dedicated field and button rather than sending people to a separate page.

The mini-cart overlay disappeared entirely. On a small screen, an overlay summarising a cart before a separate checkout is a step that costs more than it gives. Tapping the cart goes straight to checkout, with the order summary collapsed behind a show-full-summary control for anyone who wants to verify it. Same information, one less stop.

The footer restructured from five columns to two, and the nav headings took extra weight to separate parent from child in a stacked list.

Accessibility was set as a project goal rather than a final pass: colour contrast, touch targets sized for real thumbs, correct labelling, and an information order that still makes sense read aloud by a screen reader.

Consistency between desktop and mobile did related work: the same elements in the same relative places, so moving between devices mid-purchase does not mean relearning the interface.

Honestly: this was specified and designed for, not audited. I did not run it against assistive technology, and I would not claim it as verified.

I built interactive prototypes for desktop and mobile in Figma, matching the existing Penfolds design language, and presented the work back to their team with the prototypes shared for them to use.

Then it stopped. The designs were never implemented.

Validation testing was the next step and the project ended before it ran, so I have no usability findings and no conversion numbers. I would rather say that plainly than dress up a prediction as a result.

As for why none of it shipped: it was not a verdict on the work. Changes to a site like this move through commercial, brand, legal, technical and regional approval gates, and a redesign of the global navigation touches every one of them. The work was sound and it had no route through the organisation.

That is the part I actually took with me. I spent eight years in change management before this, and I had watched good proposals die in approval chains from the other side of the table. Penfolds was the first time it happened to something I had designed myself. A design that cannot be adopted is not finished, and adoption is a design problem. It needs a route mapped through the organisation, and stakeholders brought along early enough to have a stake in it.

If I ran this again, I would spend less time perfecting the prototype and more time on the smallest shippable slice, the mini-cart X, say, and on who would need to say yes to it.