Galerie
Prints

No hesitation.

The assignment was deliberately open: choose a real e-commerce brand — ideally local, ideally one you cared about, with an experience that needed work — and redesign it. No big brands with bulletproof information architecture. Somewhere I could actually walk in and recruit people to interview.

As soon as I read the brief, I didn't have a moment's hesitation. As a keen lover of photography — and a street photographer as a hobby — there was only one destination for me. It was a place I had often window-shopped at, and now here was my chance for a proper introduction. Their shop, in person, was absolutely stunning.

Now, back to the brief. Focus, Anil. The question still remained: could I improve on their branding and web design? Well — this is UX, so I'd have to put it through its paces with real users and find out. I was very intrigued, indeed.

Challenge embraced.

Galerie Prints sits in Wimbledon Park, selling everything from open re-prints to rare, signed, limited editions, with a bespoke framing and paper service behind every sale. The inventory is gorgeous; the prices run from a hundred pounds to a hundred thousand. The stakes per purchase are unusually high — and, as I'd soon learn, the website wasn't helping.

First, I had to get permission to run the study. I emailed the owner, who graciously and enthusiastically gave me his full blessing to go ahead. He was away on holiday, so I'd later meet his shop assistant, Josh, in person instead — but with permission secured, I could begin. The first job: recruit testers, and put the website through its paces.

Before any theories, I watched the site fail.

I didn't start with assumptions. I started by watching real people try to use the live site. The brief I gave them was a little story: you've had a Christmas bonus, you're moving into a new place, and you've set aside £800 for two prints — a portrait for above the living-room fireplace, a landscape for above the bed. Get them both to checkout without paying, then find the gallery's phone number to arrange a discreet delivery. Ten minutes on the clock.

First up was Sam — a former architect, now a designer, and exactly the kind of confident, spatially-literate user a site shouldn't trip up. And here's the twist: he passed. He got both prints into the basket with three minutes to spare. On paper, a success.

But passing the task and having a good time are very different things — and watching him, the cracks showed. The pages were slow to load. "Is this normal that it takes this long?" he asked, waiting on a category. "It takes ages to go to the second page, doesn't it?"

In fairness to Galerie, this was the flip side of a genuine strength: theirs is an enormous, image-rich catalogue — thousands of high-resolution prints, each offered in multiple sizes and finishes. All that gorgeous imagery is what makes the gallery special, and at the time it was also a database-and-loading issue weighing the pages down. Not carelessness — the honest cost of a catalogue that had simply outgrown its setup. (To the owner's credit, this was later addressed and optimised on his end.) A good problem to have — but, back then, it was still costing buyers at the worst possible moment.

That tape measure became the emblem of the whole project. Think about it: a former architect — someone trained to read space and scale for a living — still couldn't judge how big a print would be from the page, and reached for a physical ruler to cope. If the site defeats him on size, it defeats everyone. It had failed at the one thing this shop needed most — helping someone believe in a piece they can't hold.

It wasn't only size. Sam was stung by the pricing too: "It says from £165, from £295… that's not a good indicator," he told me, after picking a piece he loved only to find the size he wanted cost far more. "A bit of a waste of time."

And the cruellest detail? A filter did exist on that page — Sam just never saw it. He walked straight past it and did everything the hard way. A feature nobody finds is a feature that isn't there. By the end he was practically designing the fix with me: "You need a quick-view thing… filter by price range, filter by landscape… that's really clear."

Sam could have been a one-off. So I ran the same test with Will — a UX design coursemate based in Japan, testing remotely over a screen-share on a one-gigabit connection (blisteringly fast for 2022). I wanted to rule out the obvious explanation: was it just slow broadband? If the heavy pages still dragged on the fastest line I could find, half a world away, then the cause sat with the catalogue and the database behind it, not the user's connection. And that's exactly what I saw — even on a gigabit line, the image-dense pages took their time. Confirmation that this was a real, technical, server-side issue — the kind that's very fixable, and which the owner did indeed optimise later.

But Will handed me the bigger discovery. When he reached the paper-type selector, he simply stopped — not because he was deciding, but because he had no idea it was a decision at all:

Two testers, two confident passes — the site worked, but it made them fight for everything: size, price, paper, patience. Then came Stipica, and the site stopped working altogether.

Stipica is a Senior Product Manager. He navigates websites for a living — if anyone was going to sail through, it was him. Same task, same clock. But under the weight of all that imagery, the experience strained badly — and for him, an action as basic as adding to the basket simply wouldn't take.

He added a print to the basket and it vanished. He tried again. Nothing. He sat there, watching a page that wouldn't resolve, convinced something was downloading. By nine minutes in, he'd managed to look at a grand total of one item.

Then he said the quiet part out loud:

And then, the most human moment of the whole study. After the timer stopped, having not completed a single part of the task, Stipica looked up and asked — completely sincerely:

It was. Spectacularly so. A failed task is a failed task — but as a piece of evidence, Stipica's session was the most successful of the three. Three professionals, each an expert in exactly what the site fumbles — scale, speed, navigation — and the site beat all three. That's not bad luck. That's a brief.

I had my symptoms now. To understand the cause, I needed the people who watch buyers every day — so my next stop was the gallery itself.

Then Josh said the thing that changed the design.

I almost treated the interview with Galerie's own shop assistant as a formality — context, nothing more. I was wrong. Josh had watched three years of real buyers come through that door, and he understood something they couldn't articulate about themselves.

That reframed everything. The home page couldn't be a catalogue — it had to be a gateway. Lead with the names and the feeling; let discovery do the selling.

But it also exposed a deeper problem hiding inside Josh's insight. Slim Aarons works as a gateway because he's the one name people happen to recognise. Most buyers don't know the artists at all. This is exactly what I'd later hear from the buyers themselves. What they share isn't knowledge — it's instinct. They can't name what they want, but they know it the instant they see it.

So how do you build a gateway for people who don't know a single name to walk through?

I parked that question — it would become the most important feature in the redesign — and went to see how the rest of the market handled it. If anyone had cracked discovery, the big art platforms would have.

The same doubt, in every voice.

Three buyers, three completely different wallets — Martina's careful hundred, Rohan's easy impulse, Alan's considered thousand. I expected them to disagree about everything. Instead, the same quiet anxiety kept surfacing, no matter who was talking.

They couldn't tell how big the art was. Not really. Frame, mount, the actual dimensions on a real wall — it was always a guess. And when people guess on something they're paying for, they hesitate. Rohan put it most plainly:

Then there was the second doubt — friction. Both Rohan and Alan told me, independently, that a slow or fiddly site simply loses them. They don't push through. They leave.

And the third — trust. The moment a price climbs past about £500, the whole posture changes. Alan researches, reads reviews, phones the gallery. The easy-going buyer becomes a careful one. A physical shop behind the website, they all agreed, made the risk feel survivable.

Two kinds of buyer, pulling opposite ways.

As I mapped them, the buyers split into two camps — and the split was the design problem in miniature. Whatever I built had to serve both without compromising either.

Discovery for one, trust for the other, and a fix for the size problem they all shared. That was the brief the people had written for me — far more useful than the one on the page.

So I sketched the gateway.

With the people in my head — Martina's size anxiety, Rohan's impatience, Alan's caution, Josh's gateway — I sketched the spine. A home page built around discovery, and a shop page where filtering by price, size and orientation finally existed. My own margin note, scrawled at the time, said it best: "mix up everything."

Then I made two of them come back.

Here's where it gets satisfying. For the prototype test, I deliberately re-recruited Sam and Stipica — the architect who'd reached for a tape measure, and the Senior Product Manager the heavy pages had ground to a halt on. The same task that had defeated them before: find a Beatles print under £500, choose a paper, reach checkout. I wanted to know if I'd actually fixed their problems, not just problems in the abstract. And I added Anna — fresh eyes who’d never seen the original site at all.

The result in one line: the testers who failed the live site passed the redesign. Sam in two minutes, Anna in two, and Stipica straight through the basket that had collapsed on him a week earlier.

But the pass rate isn't the story. The story is what they noticed — because the thing they all poked at was the exact fix I'd built for them.

Remember the paper-type off-ramp? On the live site, Will picked his paper blind and Stipica had to open a separate tab he called "a hurdle." So in the prototype I pulled the paper types onto the product page. And the moment Sam hit it, he caught it:

Stipica caught it too. The Senior Product Manager who'd called the live-site detour "a hurdle" reached the same spot on the prototype and clocked it straight away: "Because it's on the same page, right?" The two testers who'd been most stung by the original off-ramp both recognised, unprompted, that it was gone.

Even better was the scale fix — because Sam had personally inspired it. On the live site he'd reached for a tape measure. So when I built the redesign, I did exactly what he'd done: I grabbed a measuring tape, found a chair in one of the room shots, worked out it was about three foot, and used it to size the print accurately on a real wall. Then I rendered the same print at different sizes in real, furnished rooms — so a buyer could finally judge how big it lands without owning a ruler.

The same print, three sizes, real rooms — the chair Sam measured became the scale key

When I showed Sam, he laughed — "that looks good… that's great." The buyer's coping mechanism had become the design solution. The very thing he'd improvised to survive the old site was now built into the new one.

Then they did what good testers do — they found the next problem. All three converged on the same refinement, completely independently: the "learn more about paper" prompt needed a clearer way in.

Stipica wanted it as a pop-up overlay so he'd never lose his place — "keep my interest and my focus there." Anna wanted an obvious click affordance. Sam, when I described the pop-out idea, agreed it was probably the move. Three people, three angles, one shared verdict: the inline paper info was right, but the way in needed to be louder. This was the "one thing" my presentation flagged as unresolved — and I left it that way on purpose. When three smart testers solve the same detail three different ways, the honest move isn't to pick one and pretend it's validated; it's to name it as the next iteration and say so out loud. Knowing when not to force a fix is its own kind of design judgement.

And that's the difference between the two rounds in a single image. On the live site, my testers fought the basics — pages straining under all that imagery, filters they couldn’t find, a tape measure at the desk. On the prototype, they'd sailed past all of that and were arguing about whether a paper link should be a button or an overlay. The floor had moved up. That's what a redesign is supposed to do.

Ten interviews. Ten days. One very tired designer.