Context
Team: I was the only designer, working with the CEO, CTO, a product manager and front-end and back-end developers.
My part: the research, the problem framing, every design decision across desktop, tablet and mobile, presenting the work to the team and the handoff to engineering.
Timeline: 2 weeks in 2024, during my contract at Artfinder.
Image ratios: the 1:1 and 2:1 labels on the artwork images mark the fixed image ratio I set for each layout, as a spec for engineering.
The problem
A leaky basket: in Google Analytics, about 70% of people who added an artwork to their basket didn’t go on to start checkout.
More removing than buying: on the basket page, removing an item was the third most common event, ahead of starting checkout.
Mostly on phones: over 60% of people viewing artworks were on mobile.
How I found the problems
Analytics: Google Analytics events on the basket page, and the funnel from product page to purchase.
Session recordings: Hotjar recordings of visits to the basket, which I linked as evidence for each problem.
Heuristics: I checked each problem against usability heuristics such as visibility of system status, recognition rather than recall, and aesthetic and minimalist design.
Market analysis: I reviewed basket pages from art marketplaces such as Saatchi Art, Singulart and Rise Art, and from large retailers such as IKEA, John Lewis, ASOS and Amazon. For several problems, our direct competitors didn’t follow good practice, but the large retailers did.
Problem cards: each of the nine problems got its own card: what’s wrong, the evidence, the heuristic it breaks, a problem statement written from the user’s side, and the measure the fix should move (completion, time on task, errors or satisfaction).
The nine problems
Four of them shaped the three decisions below; the other five are under ‘Other changes’.
| Problem | Covered in | |
|---|---|---|
| 1 | Artworks that couldn’t ship, or whose shop was closed, had to be removed before checkout | Decision 1 |
| 2 | Warnings that repeated or didn’t apply | Decision 2 |
| 3 | Guarantees at the very bottom of the page | Decision 3 |
| 4 | No tax in the order summary | Decision 3 |
| 5 | A ‘Shipping to’ dropdown for every seller | Other changes |
| 6 | No item count | Other changes |
| 7 | Card details that all looked the same | Other changes |
| 8 | The summary before the items on mobile | Other changes |
| 9 | Large baskets that froze the page | Other changes |
Decision 1: ‘Save for later’ instead of ‘remove to continue’
What I saw: when a seller’s shop was temporarily closed, or an artwork couldn’t be shipped or imported to the buyer’s country, the basket told people to remove it before they could continue. On an art marketplace, people often spend a long time finding a piece they connect with, so being told to delete it is a real loss.
What I did: added ‘Save for later’, so people can set an artwork aside and still check out the rest.
Why it mattered: removing an item was already the third most common action on the basket page. Some removals may have been forced by shipping restrictions or closed shops, which made this worth addressing.
Decision 2: One clear message instead of a wall of warnings
What I saw: red, yellow and blue information blocks stacked up at the top of the page and inside each seller’s section. Some repeated the same message; others didn’t apply in that situation. A page full of warnings makes people feel something is wrong, and every extra question adds time before they act.
What I did: I defined rules for when each message appears. If a shop is closed, or the art can’t be imported to the buyer’s country, the page shows only that message, hides the rest, and tells people what to do next.
Decision 3: Trust and the full price, where people decide to pay
What I saw: the guarantees that answer people’s biggest worries about buying art online (secure payments, a money-back guarantee and dedicated support) sat at the very bottom of the page, just above the footer. And the order summary never mentioned tax, so the price could jump when tax appeared at checkout.
What I did: moved the guarantees up next to the order summary and checkout button, and added a tax line to the order summary with a short explanation in context.
Why: this is the part of the basket closest to payment, so it’s where reassurance and price clarity matter most.
Other changes
One shipping location: each seller’s section had its own ‘Shipping to’ dropdown. It looked as if each could be set separately, but changing one changed them all. I replaced them with a single shipping location at the top, as large retailers such as IKEA and Crate & Barrel do.
Item count: the page didn’t say how many items were in the basket, so people scrolled to count them. The new heading shows the number.
Clearer cards: title, artist, medium, size, price and packaging all looked the same. I gave them a hierarchy so people can scan them.
Mobile order: on mobile, the order summary came before the items, but people check their items first. I moved the full summary below the items and kept a compact total and checkout button at the top.
Large baskets: the team reported that very large baskets froze the page while every item and price loaded at once. I designed the basket to show the first items with a ‘load more’ control.
Results
What shipped: the team shipped all nine changes after my contract ended, and the new basket is still live on artfinder.com nearly two years later.
Measurement plan: each change came with a success measure (completion, time on task, errors or satisfaction) for the team to track.
Limitation: my contract ended before launch, so I don’t have post-launch numbers.
What I’d do differently
Agree the measures before launch: I’d set up the before-and-after comparison with the team and stay long enough to read it.
Test the new design with users: my evidence came from analytics, recordings and heuristics. I’d add moderated usability tests on the new design before it shipped.