Research & strategy

Double Diamond

Discover, Define, Develop, Deliver: explore the problem before you solve it.

⏱ 12 minBeginnerBy UK Design CouncilSince 2004

Use it when

Planning a project from a vague brief to a shipped solution.

You'll walk away with

A problem statement, How-might-we questions and a phase plan.

Step 1 · The idea

Design the right thing, then design the thing right

The Double Diamond was developed by the UK's Design Council in 2004, based on its research into how designers actually work. It shows the design process as two diamonds.

The first diamond is about the problem: go wide to understand it (Discover), then narrow down to one clear problem (Define). The second diamond is about the solution: go wide with ideas (Develop), then narrow down to what works (Deliver).

The shape shows the key habit: diverge, then converge — twice.

Most teams skip the first diamond. Someone says “add a live chat button”, and they start designing. The Double Diamond makes you ask: is that even the real problem?

Step 2 · Play with it

Tap a phase

Each half of a diamond is a phase. See what you do there, which methods to use, and how you know you're done.

Problem space · Diverge

Discover

What's really going on?

Understand the problem from the user's side, without jumping to solutions.

Methods

User interviews
Observation / contextual inquiry
Analytics and support tickets
Stakeholder interviews
Competitor review
Diary studies

Output

Research notes, quotes, data and early insights.

You're done when

You keep hearing the same things from new users (saturation).

Step 3 · Worked example

Cutting order cancellations at a food delivery app

A fictional team at “Thali”, a food delivery app, walks through both diamonds.

From: Product head

Too many customers cancel orders after paying. Can we add a live chat button by next sprint?

Monday, 10:02 AM

Step 1

The brief arrives

The brief already contains a solution: “add live chat”. Before designing it, the team checks what's really causing cancellations.

Before Discover

Research wall

“The rider was assigned, then no updates.”

“It said 25 min. It took an hour.”

“The restaurant was still cooking.”

“I didn't want to chat, just know when.”

Order data: most cancels follow a missed ETA

Step 2

Discover: talk to customers

They interview 12 customers and look at order data. Going wide shows several reasons — not just “nobody to talk to”.

Discover · diverge

Problem statement

Customers see a hopeful ETA at checkout and hear nothing when it slips, so they give up and cancel.

How might we…

…keep people honestly informed until the food arrives?

Step 3

Define: one clear problem

Affinity mapping groups the notes. The biggest cluster is about wrong or silent ETAs, not a lack of chat. That changes the design.

Define · converge

Crazy 8s

⏱️ Honest ETA range at checkout

🔔 Alert the moment an order runs late

🍛 “Restaurant is still cooking” status

🛵 Rider progress on the map

Step 4

Develop: many ideas

Now they go wide on solutions and make rough prototypes of the best three.

Develop · diverge

Friday · 8:40 PM

🛵 Thali

Running 10 min late: the restaurant is busy. New arrival 9:05–9:10 PM. We'll update you again →

Tested with 8 customers: clearest of the three

Step 5

Deliver: test and ship

Testing shows an honest ETA plus a proactive delay alert is clearest. They ship it first and track the cancellation rate.

Deliver · converge

After launch

Cancels after a delay↓ down

New insight

Rain nights still break every ETA

Back to Discover →

Step 6

Loop back

It worked, and it revealed a new problem: rainy nights. That goes back into Discover. The process loops.

Iterate

Fictional worked example for teaching. Thali is not a real company, and the research findings and numbers are illustrative.

Step 4 · Try it

“How might we” generator

The bridge between Define and Develop. Describe your problem and get several ways to frame it. Good HMWs are not too broad and not too narrow.

Who is the user?

What are they trying to do?

What gets in the way?

Pick the one or two that make your team go “ooh”. Those are good starting points for brainstorming.

1 · Direct
How might we help customers waiting for a late order know when their food will really arrive?
2 · Remove the barrier
How might we take away an ETA that goes silent after checkout for customers waiting for a late order?
3 · Question the assumption
How might we help customers waiting for a late order know when their food will really arrive without an ETA that goes silent after checkout?
4 · Make it effortless
How might we make it effortless for customers waiting for a late order to know when their food will really arrive?
5 · Flip it
How might we make an ETA that goes silent after checkout something customers waiting for a late order never have to think about?
6 · Borrow an idea
How might a hotel concierge help customers waiting for a late order know when their food will really arrive?

Behind the diamonds

Four principles

In 2019 the Design Council updated the model into a wider “Framework for Innovation” with four principles.

Put people first

Start with the needs, strengths and hopes of the people using the service.

Communicate visually and inclusively

Help everyone understand the problem and ideas: sketches, maps and prototypes, not only documents.

Collaborate and co-create

Work with users, engineers and stakeholders, and take inspiration from others.

Iterate, iterate, iterate

Test early, learn from failure and loop back. The diamonds aren't a one-way line.

Watch out

Common mistakes

✓ Do

Treat the brief as a hypothesis to check, not an order
Time-box each phase so research doesn't go on forever
Write down the problem statement and get agreement
Loop back when testing reveals something new
Make it visible: research walls, maps, sketches

✗ Avoid

Skipping the first diamond (“solutionising”)
Treating it as a strict waterfall
Converging too early, with only one idea
Research that never changes the design
Shipping and never measuring

Cheat sheet

Double Diamond in one screen

   PROBLEM SPACE              SOLUTION SPACE
   <> Discover  > Define      <> Develop   > Deliver
   diverge       converge     diverge      converge

DISCOVER  What's really going on?        interviews, observation, data
DEFINE    Which problem do we solve?     affinity map, personas, problem statement, HMW
DEVELOP   What could the solutions be?   brainstorm, Crazy 8s, low-fi prototypes
DELIVER   Which works? Ship it.          usability tests, A/B tests, handoff, measure

PRINCIPLES  people first · communicate visually · co-create · iterate
HMW FORMULA  How might we help [user] [goal] without [barrier]?

Use it at work

Figma template

A ready-made canvas for running this framework with your team. Free, editable, and it works in Figma and FigJam.

Figma & FigJam · free

Double Diamond project board

Four phase columns
Method checklists
Problem statement

1Download the SVG, or copy it to your clipboard.
2In Figma or FigJam, drag the file onto the canvas — or paste with ⌘V / Ctrl V.
3Everything arrives as editable layers: type in the boxes, add stickies, duplicate rows.

Download SVGOpen in Figma ↗

Check yourself