2018 — 2020 • Web
Photobox’s photo book editor was outdated, fragmented, and at risk with the end of Flash. I helped redesign and rebuild it, integrating AI to simplify creation, reducing time-to-completion, and increasing order value. We scaled it across brands, regions, and multiple languages.
Intro
Any company that has been around for nearly 20 years comes with its own set of legacy challenges. At Photobox, this showed up in design, technology, and even internal processes. But change was coming. Investors were pushing for modernisation, with a clear focus on putting Photobox products at the forefront. And they are wonderful products. A new photo book editor was already in development—one that would give users a more flexible and intuitive creation experience.
With this shift, Photobox was building its first permanent in-house product team, moving away from an agency-led model. I was one of the first permanent product designers to join, moving over from my time in agency to give design a stronger voice within their products and product design processes.
The problem: outdated technology, fragmented UX
Photobooks were and probably still are Photobox’s biggest revenue driver, bringing in whopping ~£24 million annually. The A4 Personalised Photobook alone accounted for 17.9% of total sales. But the platform users relied on to create these books was completely outdated.
Most of the book range was still built on a Flash-based editor. This was a ticking time bomb—Adobe had already announced they were ending support for Flash, and Chrome 76 was about to turn it off completely by default. If Photobox didn’t act fast, their biggest revenue stream would be at monumental risk.
On top of that, the existing editor system was a mess.
Over the years, Photobox had attempted to build new editors, leading to a fragmented experience:
Two separate editors (one in Flash, one in HTML5)
18 different versions—each tied to a specific product range (I know, wow.)
No data-driven decision-making, making it impossible to track performance or improve the experience
Every time they introduced a new product, a team would duplicate the existing editors code, creating inconsistent fonts, backgrounds, themes, and UX. It was an unsustainable ****system, and with the Flash deadline approaching, the priority was clear… protect revenue by rebuilding the book editor first.
What I worked on & My impact
With a tight deadline, we had to move fast. Instead of aiming for feature parity with the old editor, we focused on building a better experience using data and user insights to guide us.
One major decision was whether to invest in mobile. The agency-led team had been focused purely on desktop, believing that complex products like photo books were too difficult to create on mobile. But during that Christmas—Photobox’s busiest period—mobile traffic spiked significantly, leading to high bounce rates in the mobile studio because of bad the platform was at the time.
I worked closely with Chris, my UX counterpart, to build a business case for stronger mobile investment. The numbers spoke for themselves to be honest, and we quickly secured funding to expand the team, hiring a dedicated designer and bringing in more developers to help build the mobile editor at pace. To ensure we were making the right decisions, we integrated data tracking and user feedback loops directly into the editor.
Our product owner and data team connected Looker to pull daily metrics into Slack, tracking:
Average order value (AOV)
Daily visitor count
Creations started & Creations completed (looking specifically at conversion rates, creation time and drop-off rate within the funnel.)
We also embedded Usabilla for in-editor feedback, allowing users to comment and report issues at any stage. On top of that, we worked with analysts to build a conversion funnel, helping us identify exactly where users were dropping off, and more importantly, why.
Introducing AI to the creation experience
Alongside the editor rebuild, a separate team in France was working on an AI-powered tool called Smartfill. It had been partially integrated into the existing HTML5 studio, but its potential hadn’t been fully realised.
Smartfill automatically sorted and placed users’ photos into books, either chronologically or by filename, drastically reducing the time and effort needed to create a book. However, our research revealed a clear user divide:
Some users wanted full creative control, customising layouts and styles
Others just wanted to finish quickly, preferring an auto-assisted approach(especially when there was an impending celebration date coming up fast!)
To cater to both, we repositioned Smartfill as EasyCreate™—a personalised creation journey that didn’t just autofill photos but guided users step by step. Our most important insight? Users were far more likely to complete their book if they didn’t start from a blank page.
With this in mind, we designed a stronger AI-assisted onboarding flow, ensuring users saw their photos instantly placed into a book and getting them invested in the process from the get go.
To give EasyCreate™ the attention it needed, we also hired a dedicated product designer to focus solely on the AI experience.
Navigating the flash deadline & Business constraints
The biggest concern internally was how users would react to the Flash editor being disabled. If they couldn’t find their way into the new editor, there was a risk they would leave for a competitor.
When Chrome 76 launched, we closely monitored:
A/B test results comparing Flash and the new editor
Customer support call volumes
Browser usage trends
To the company’s surprise, conversion rates remained stable—our users found a way around the issue by switching browsers or reaching out to customer support. While this wasn’t ideal, it reassured the business that we could phase out Flash more gradually without an immediate loss in revenue.
Scaling across brands & markets
Beyond Photobox, the business also had to support it's other brands:
Hoffmann (Spain)
PosterXXL (Germany)
Rest of Europe (ROE)
This meant adapting the new editor for three brands, across eight languages, while ensuring content was maintained and updated across all local markets. The design team was responsible for making sure the product was supported in all regions, which added a layer of complexity we had to work through.
The challenge of scaling without a roadmap
For some reason, the new operating model strictly prohibited roadmaps—the belief was that they 'constrained teams' and limited the ability to pivot. In theory, this encouraged agility, but in reality, it created chaos. With so many moving parts, including physical product supply chains, having no structured plan often led to rushed releases, last-minute fixes, and little time for refinement.
There were times when we put significant effort into researching and designing solutions, only to have them shipped too quickly without the necessary polish or iteration. But we understood the pressure. We had a tight deadline and had to prioritise what mattered most for the business. Our focus became reducing customer service calls, refining the core user experience, and making sure users had a smooth, frustration-free journey to the best of our ability at the time.
Key outcomes & Impact
Despite the complexities and constraints, we delivered a fully responsive, AI-enhanced editor that drove real impact:
225,000+ books sold in the first year of real testing
99.99% renderer success rate, ensuring stability during peak periods
5.5% higher conversion rate than the old Flash editor
50% reduction in average creation time (users completed books four hours faster than before)
30% increase in AOV (+£10.44 per creation), driven by a more seamless upsell journey
