AgentSkillsCN

Design Sprint

在面临高风险、高 stakes的问题时，将决策、设计与验证压缩至短短五天内完成——在投入工程资源之前，先以客户反馈为导向，快速验证产品构想，避免资源浪费

SKILL.md
--- frontmatter
id: design-sprint
name: Design Sprint
tagline: Five-day structured process to solve big problems and test ideas with customers
description: Compress decision-making, design, and validation into five focused days when facing high-stakes problems requiring customer feedback before committing engineering resources
category: product-development
subcategory: product-discovery
similar_to:
  - continuous-discovery-habits
  - lean-product-playbook
  - mom-test
prerequisite_skills:
  - opportunity-solution-trees
  - product-market-fit-survey
link: https://www.gv.com/sprint/

Core Principle

The Design Sprint is a time-boxed five-day process created at Google Ventures for solving critical business problems through prototyping and testing ideas with real customers. Instead of months of debate, meetings, and incremental progress, teams compress decision-making, design, and validation into one focused week.

Created by Jake Knapp in 2010 at Google, refined at GV with contributions from Braden Kowitz (story-centered design) and Michael Margolis (one-day user research), the sprint follows a structured daily rhythm from problem mapping to customer validation.

When to Use

Deploy Design Sprints when you need to:

  • •De-risk big decisions before committing significant engineering resources
  • •Break deadlocks when teams have competing visions or endless debates
  • •Validate new product concepts before building anything real
  • •Explore new markets or pivots with customer feedback
  • •Align stakeholders around a shared vision through collaborative process

Ideal for high-stakes, high-uncertainty problems where the cost of being wrong is high (new features, rebrands, strategic pivots). NOT for minor iterations or problems with obvious solutions.

How It Works

The Five-Day Structure

Monday - Map (Define the Challenge)

Set a long-term goal and map the problem space.

  • •Morning: Start at the end - define what success looks like 6-12 months out
  • •Afternoon: Map the customer journey from start to finish
  • •Evening: Pick a target - one specific, ambitious piece to tackle this week
  • •Output: Long-term goal, map of the problem, and focused target

Key Activities:

  1. •Set audacious but achievable long-term goal (2-3 years out)
  2. •List "How Might We" questions from team and experts
  3. •Create customer journey map showing all steps and actors
  4. •Vote on most critical target area for the sprint
  5. •Choose a "Decider" who has final vote on key decisions

Tuesday - Sketch (Generate Solutions)

Individually sketch competing solutions inspired by existing ideas.

  • •Morning: Lightning demos - review existing solutions for inspiration
  • •Afternoon: Four-step sketch process emphasizing ideas over art
  • •Output: Detailed solution sketches from each team member

Sketch Process (Individual, Silent Work):

  1. •Notes (20 min): Walk around, take notes on key ideas
  2. •Ideas (20 min): Doodle, make lists, mind-map rough concepts
  3. •Crazy 8s (8 min): Rapidly sketch 8 variations in 8 minutes
  4. •Solution Sketch (30-90 min): Detailed 3-panel storyboard of your best idea

Wednesday - Decide (Pick the Best Solution)

Critique solutions and decide what to prototype.

  • •Morning: Structured critique - sticky note voting on solution sketches
  • •Afternoon: Supervote by Decider, stitch winning ideas into storyboard
  • •Output: Storyboard ready for prototyping

Decision Protocol:

  1. •Art Museum - Display all sketches anonymously on wall
  2. •Heat Map - Team silently places dot stickers on interesting parts
  3. •Speed Critique - 3 minutes per sketch, note standout ideas
  4. •Straw Poll - Each person votes for one solution
  5. •Supervote - Decider makes final call (gets 3 votes)
  6. •Storyboard - Turn winning scenes into 10-15 panel customer journey

Thursday - Prototype (Build a Realistic Facade)

Create a realistic-looking prototype that customers can react to.

  • •All Day: Build "Goldilocks quality" prototype - just real enough to get honest reactions
  • •"Fake It" Philosophy: Simulate the end experience without building real infrastructure
  • •Output: Testable prototype (landing page, clickable mockup, physical model, etc.)

Prototype Principles:

  • •You can prototype anything in one day (mindset shift is critical)
  • •Focus on customer-facing surface, not real back-end
  • •Divide and conquer - assign roles (Maker, Stitcher, Writer, Asset Collector, Interviewer)
  • •Just enough quality to trigger real reactions (too polished wastes time, too rough gets fake feedback)

Friday - Test (Validate with Real Customers)

Interview 5 customers, watch them interact with prototype, learn what to do next.

  • •Morning: Conduct five 1-hour customer interviews with prototype
  • •Afternoon: Debrief patterns, decide next steps
  • •Output: Clear learning on what works, what doesn't, what to do Monday

Testing Protocol:

  1. •Recruit 5 target customers (scheduled earlier in week)
  2. •Five-Act Interview - Friendly welcome → Context questions → Prototype → Debrief → Quick quiz
  3. •Whole team watches from observation room, takes notes
  4. •Look for patterns across all 5 interviews
  5. •Decide - Which ideas worked? What needs to change? Build, iterate, or kill?

Team Composition

Required Roles:

  • •Decider - Has authority to make final calls (founder, VP, whoever owns the outcome)
  • •Facilitator - Guides sprint process, manages time, keeps energy high
  • •Designer - Creates prototype (often 2 designers for complex products)
  • •Subject experts - Engineering, marketing, customer support (anyone with critical knowledge)

Team Size: 7 people or fewer (more creates logistical chaos)

Time Commitment: Full-time for all five days - no "checking email," no half-attendance. Sprint only works with complete focus.

Practical Examples

Example 1: Slack Pre-Launch Pivot Context: Stewart Butterfield's gaming company dying, team had built internal chat tool

  • •Monday: Mapped problem - "How might we make internal communication a product?"
  • •Tuesday: Sketched competing visions for standalone chat product
  • •Wednesday: Decided on searchable, integrations-first approach
  • •Thursday: Built prototype landing page + demo video showing key interactions
  • •Friday: Tested with 5 technical teams, validated "searchable" as killer feature
  • •Outcome: Decided to pivot entire company to chat product (became Slack)

Example 2: Blue Bottle Coffee Retail Experience Context: Expanding from online to physical retail, needed to define in-store experience

  • •Monday: Goal - "Premium coffee experience that converts first-timers to regulars"
  • •Tuesday: Sketched 8 different cafe layouts and service models
  • •Wednesday: Chose barista-forward, educational experience (vs. fast self-serve)
  • •Thursday: Built physical mockup of counter layout, printed menu, scripted interactions
  • •Friday: Tested with 5 coffee drinkers, discovered educational approach resonated
  • •Outcome: Defined retail blueprint scaled to 70+ locations

Example 3: Foundation Medicine (Genomic Testing Startup) Context: Needed to explain complex cancer genomics to oncologists in sales process

  • •Sprint: Tested 3 different report formats and explanation frameworks
  • •Discovery: Doctors wanted actionable treatment recommendations, not academic detail
  • •Pivot: Redesigned reports to lead with therapies, not gene mutations
  • •Impact: Improved close rate, eventually sold to Roche for $2.4B

Expected Outcomes

After a successful sprint, you'll have:

  • •Validated or invalidated key assumptions with real customer feedback
  • •Aligned stakeholders on direction (everyone saw same customer reactions)
  • •Clear next steps - what to build, what to change, what to kill
  • •Momentum from shipping in days, not months
  • •Reduced risk by failing fast on bad ideas before heavy investment

Success metric: On Friday afternoon, you can confidently answer "What should we do Monday?"

Common Pitfalls

Skipping Days or Cherry-Picking Steps: The structure is designed as a system. Skipping sketch day or testing only 2 people breaks the process.

Wrong Problem: If you map a tactical feature on Monday instead of strategic question, you'll get incremental answers. Go bigger.

Half-Time Participation: If Decider keeps leaving for meetings or team checks Slack, you lose focus and momentum.

Perfect Prototype: Teams waste Thursday polishing details customers don't notice. "Goldilocks quality" is sufficient.

Ignoring Negative Feedback: Friday's goal is learning, not validation. Pay attention to confused faces and hesitation.

Complementary Practices

Combine Design Sprint with:

  • •Continuous Discovery Habits for ongoing customer touchpoints after sprint
  • •Lean Product Playbook for post-sprint build-measure-learn iteration
  • •Opportunity Solution Trees to map decisions and assumptions made during sprint
  • •Product-Market Fit Survey to quantify validation after launching sprint outcomes

Concept Origin: Jake Knapp (Google Ventures, 2010) Key Resources:

  • •Sprint: How to Solve Big Problems and Test New Ideas in Just Five Days (Knapp, Zeratsky, Kowitz, 2016)
  • •The Sprint Book website (thesprintbook.com)
  • •GV Sprint methodology (gv.com/sprint)