Dyson Support Home — Case Study

00 Overview

The Moment & The Brief

Dyson owners entered support through 8+ disconnected steps and dead-ended constantly. The brief said "design a Support Home page for Australia and New Zealand." What shipped was bigger: a behaviour-first entry system, validated with owners, that became the standard for every M2 market.

Help existed. Owners couldn't find it.

Troubleshooting, spare parts and service pages all existed — but with no front door connecting them, owners bounced between flows or gave up and called support.

A Support Home that owners actually use

Two months post-launch: bounce −12.3% (desktop), −15% (mobile), exit −25.7% (mobile) — and a tested template adopted by SG, KR, IN and TR.

Lead Product Designer

Accountable for the end-to-end experience across 6 global markets — from research and information architecture to delivery and measurement.

Key Responsibilities

Led end-to-end UX — research → concept → testing → delivery
Defined product direction across markets
Collaborated with content strategists and regional teams
Worked within a hard constraint: no new backend capability

01 Context

Self-Service Isn't a Nice-to-Have. It's the Brand Promise.

Premium owners expect premium care — and every failed self-serve attempt becomes a phone call. ANZ, the smallest market on the M2 platform, was the right test bed to learn fast before rolling out across APAC.

Owner trust

A confusing support flow undercuts the product story Dyson tells pre-purchase.

Cost to serve

Reducing avoidable contacts protects the cost-to-serve ratio across APAC.

Roadmap leverage

Learn fast in the smallest M2 market, then scale the pattern to SG, KR, IN and TR.

Six Markets. Six Different Support Setups.

Before designing anything, I audited what each M2 market actually offered. Every region had reinvented support locally — and the gap was stark: only the UK had a Support Home and a Guides path. Every APAC market on M2 was missing both.

Audit matrix of six markets showing Support Home and Guides & Manuals missing in every APAC market

Working Within Real Constraints

Can't

Build new functionality on M2

No new backend capability, no new components. The platform was what it was.

Can

Reuse existing support pages

Troubleshooting, Spare Parts and Service Centres already existed. The win was the connective tissue — not new pages.

02 Clarify Problems

Reframing the Brief, Not the Product.

The brief arrived page-shaped. But ANZ wasn't just a market problem — it was a system gap. Treating it as a layout exercise would have shipped a prettier dead end.

What was briefed

"Design a Support Home page for Australia and New Zealand"

A page-shaped problem, scoped as a content layout exercise.

What I reframed it as

"How do owners enter self-serve — and where are we losing them?"

A behaviour-shaped problem — solvable inside the platform constraints by re-organising existing modules and copy.

03 Shape Concept

Three Bets on the Owner's Mental Model

Each option made a different bet on how owners think when something goes wrong — and each came with an honest trade-off.

How might we create a precision-based support framework that scales globally — without requiring a backend overhaul?

Option A mockups — service-first support home

Option A · Service first

Pick the type of help, then the product

Buildable inside existing M2 modules
No new components, no engineering blockers
Doesn't solve the deeper parts-findability problem

Option B mockups — product-first range level navigation

Option B · Product first, range

Pick the product family, then the help

Reuses the existing product-page IA
Forces a category mental model onto service tasks
Broke when owners weren't sure of their category

Option C mockups — SKU-level support hub

Option C · Product first, SKU

Pick the exact machine, see all support in one hub

Best long-term experience for owners
Highest task completion in testing (9/10)
Requires deep changes to core flows

04 Test & Ship

Owners Don't Think in Services. They Think in Products.

Usability testing made the mental model unmistakable — SKU-level scored highest (SUS 84.75). But the decision wasn't the winner-take-all kind: the lightest model unlocked the heaviest one.

SUS scores: Service-first 80.25, Range 80.45, SKU-level 84.75 — All three passed the usability bar — SKU-level clearly led.

Impact versus effort quadrant showing Service-first as low effort high impact, SKU-level as high effort high impact — The call: ship Service-first now within constraints; make SKU-level the roadmap destination.

Tightening the Journey Before Launch

Two assumptions didn't survive contact with owners. Both fixes were retested clean before launch.

Iteration one: machine registration moved lower — owners ignored it on first scroll; registration matters after they find their product

Iteration two: Owner's Guide versus User's Manual merged — owners couldn't decode the difference and gave up choosing

The Shipped Support Home

Final shipped designs: Support Home and Guides & Manuals on desktop and mobile

05 Impact

Two Months Post-Launch — Every Metric Moving the Right Way

−12.3%

Bounce rate · Desktop

−4.8%

Exit rate · Desktop

−15%

Bounce rate · Mobile

−25.7%

Exit rate · Mobile

Post-launch attention analysis of the shipped Support Home pages — Post-launch attention analysis: owners engaging down the full page instead of bouncing at the top.

06 Scale

From One Market Page to an APAC System

What we shipped wasn't just a market page — it was a tested template every M2 market could adopt. AU/NZ became the blueprint for Singapore, Korea, India and Turkey.

Diagram of the AU/NZ Support Home template propagating to SG, KR, IN and TR

Pattern 01

Information hierarchy

Product support first, other services second — one order, every market.

Pattern 02

Component library

Service card, product picker, owner block — reusable without engineering.

Pattern 03

Copy & wayfinding rules

Specific verbs, no dead-ends — codified so every localisation stays navigable.

LEVERAGE UNDER CONSTRAINTS

Instead of rebuilding the platform, we targeted the moments with the most user pain and the clearest opportunity to ship. Designed once, tested once, adopted five times — and the SKU-level model the tests validated became the north star for the next platform generation.