Vodafone.com

Vodafone.com was the company's global corporate website — the destination for investors, journalists, partners, prospective employees, and anyone trying to understand Vodafone as a global organisation rather than a local consumer telco.

It had two problems by the time the re-platform was commissioned.

The platform was failing structurally. The legacy CMS was difficult to maintain, slow to update, and didn't support the kind of modular content reuse that Vodafone needed to keep dozens of country properties and content streams consistent.

The information architecture was working against its visitors. Around 93% of vodafone.com's traffic was looking for a local market site — vodafone.co.uk, vodafone.de, vodafone.it, and 30+ others — not the corporate site. Visitors bounced. Or wandered. Or never came back to either property.

The brief was to redesign the corporate site, re-platform it onto Adobe Experience Manager, and solve the global-to-local navigation problem in the process.

I joined as Lead User Experience Consultant. This was my first project owning a full UX workstream end-to-end.

That distinction is worth naming. By 2015 I had three years of senior consulting behind me at BAE — but in a particular shape. I'd led interaction design on EE's digital transformation in 2013. I'd led interaction design on Vodafone Enterprise in 2014. I'd held a Lead IC role on internal BAE projects, owning interaction design while reporting into other people's UX leads. I was good at the IC craft. I'd led specific workstreams within bigger ones. I hadn't yet had the role where I owned the entire UX function on a project.

Vodafone.com was the step up. I owned the IA. I owned the interaction design. I owned the personalisation and content strategies — the strategic layer that translated Vodafone's existing research into a system. I owned stakeholder engagement with Vodafone's product, brand, and engineering leadership. And I mentored the junior UX designers on the team.

A lot of the senior leadership work I've done since — Lloyd's Register, Taskize, Santander — drew on the muscle I built here.

Vodafone arrived with research and personas already in place. Their team had done the customer work — six personas covering enterprise customers, consumers arriving from the wrong door, investors and regulators, jobseekers, media and press, and not-for-profit partners.

My contribution wasn't to redo this work. It was to build the strategic layer above it — taking Vodafone's existing personas, journeys, and customer research and turning them into an operational personalisation system that could decide, for any visitor in any session, what content the site should surface.

This is the senior craft I want to call out specifically. Generating personas in a workshop is something many designers can do. Taking existing customer research from a client and architecting a personalisation system that scales across a global corporate site — that's a different and rarer move. It's the move that earns you the next bigger project.

I redesigned the corporate site's IA from the ground up. The old structure had grown organically and didn't map to how any of the six audiences actually thought about Vodafone. The new IA was built around six clear top-level areas — Who we are, What we do, Investors, Media, Careers, and the routing path to local market sites — each with a clean hierarchy underneath.

The IA decision was deliberately conservative at the structural level and ambitious at the layer above: the structure would be stable and obvious, and the personalisation system would do the work of surfacing the right content to the right person within that structure.

That division of labour — stability where it matters, intelligence where it pays back — is one of the most useful design principles I've taken out of this project. It's the reason the architecture is still standing a decade later.

The 93% problem was solved through a structured decision flow that respected both prior user choice and inferred intent.

• Returning users who'd already chosen vodafone.com kept their choice. No re-prompt on every visit.
• First-time visitors with detectable location were offered their local market — softly, with the option to stay.
• First-time visitors with no detected location got vodafone.com as fallback, with country selection always available.
• Full OpCo countries got direct links to that operating company's site.
• Partner countries got a clear message explaining the partner relationship.
• No-presence countries got an honest message rather than a broken redirect.
• The Americas got their own routing, because Vodafone treated the region's brand presence differently.
• Device-experience mismatches were caught at the end, with a clear desktop-fallback option.

This wasn't a clever piece of personalisation. It was careful UX engineering — respecting users, ageing well, and not surprising anyone when traffic patterns shifted years later.

The country redirect was one piece of a much larger system. The full personalisation architecture decided, for any visitor in any session, what content the site should surface — on the homepage, in the navigation, in modules, and through targeted campaigns. Eleven signals fed into AEM's Experience Manager:

• Customer segment · Browsing history · Referrer link · Recently visited content
• Location (HTML5) · Roaming flag · Brand affinity (Orange, T-Mobile, EE)
• Device / OS / browser · Time and date · Account details (signed-in via MyEE / future MyVF)
• Planned social graph integration via Facebook's open graph (specced for a future phase).

The Content Pyramid decided, for any given module on any given page, what kind of content to surface. P1 (up to 5%) — campaigns every visitor should see. P2 (up to 15%) — content personalised by both location and role. P3 (up to 25%) — personalised by location or role. P4 (>50%, default) — general content when no rule matched.

This was the system that made the IA work. The architecture stayed simple and obvious; the personalisation made it feel relevant.

AEM is a modular content management platform. To make the personalisation architecture deliverable at scale, the interaction design had to be modular too — each piece of content reusable, each module compatible with multiple content types and aspect ratios, each interaction documented at production specification.

Each module was specified to depth. Required versus optional metadata. Behaviour when content was missing. Aspect ratio variants. Notes for engineering. The patterns held together as a coherent system rather than a collection of one-off screens.

The interaction design also covered the choreography of the experience — animations, transitions, responsive behaviour across five breakpoints. I prototyped these in Axure and tested motion choices in Adobe Edge and After Effects before they shipped to development.

The work didn't happen in isolation. I led multiple workshops and show-and-tells across the project — bringing Vodafone stakeholders, BAE engineering, and the wider design team into the same room (or wall, more often) to align on the IA, the personalisation rules, and the content strategy.

These sessions did three jobs at once: they built shared understanding across teams that wouldn't otherwise have had it, they surfaced disagreements early when they were cheap to resolve, and they gave Vodafone stakeholders ownership of the decisions they'd be defending internally after we'd gone.

I mentored the junior UX designers on the team throughout. The mentorship pattern I used — explain the reasoning, ask sharp questions before stepping in, let the work be theirs where it could be — was the one I'd been mentored with myself. It's the same pattern every junior I've worked with since has been mentored under.

Vodafone.com was the project where I stepped from "Lead IC on a workstream" to "Lead on the whole UX function." That's a bigger jump than it sounds.

The best architecture decisions are the boring ones. I made the IA deliberately conservative — six clear top-level areas, an obvious hierarchy — and put the cleverness in the personalisation layer above it. A decade later, the IA still works. The personalisation has evolved. That's the right division of labour: stability where stability matters, intelligence where intelligence pays back.

Durability is a design value, not a side effect. The work that lasts ten years isn't the work that was most exciting on launch day. It's the work where the structural decisions were good enough that the next team could extend the system rather than start over. That's the bar I now hold myself to — not "will this work today" but "will this still be working when I'm long gone?"