Testing & Refining Ideas: Stage 4 of Design Thinking – What Works?
Stage 4: What Works?
After selecting the best ideas (What Wows?), it’s time to test them in the real world with actual users. This stage helps businesses understand if an idea truly works before investing heavily in a full-scale launch.
Key Focus:
✅ Testing ideas with real users through experiments.
✅ Quick feedback loops to improve prototypes.
✅ Fail fast and learn early to avoid costly mistakes.
Tools Used in Stage 4: "What Works?"
Tool 9: Customer Co-Creation
This involves directly involving customers in product development. Instead of guessing what they want, we give them a prototype, observe their reactions, and improve accordingly.
How It Works:
- Create a simple prototype of your product/service.
- Let customers use it & gather feedback (what they like/dislike).
- Refine & improve the prototype based on their feedback.
- Repeat the process until the product meets customer expectations.
Example:
- LEGO Ideas allows fans to submit new product ideas. The best ones are developed into actual LEGO sets.
- Nike co-created sneakers with athletes by testing early prototypes on them before launching.
Tool 10: Learning Launch
This is a small-scale test in a real market to see how customers actually behave, not just what they say.
How It Works:
- Launch a limited version of the product/service (e.g., in one city).
- Observe customer behavior – do they actually buy it?
- Analyze insights – what works and what needs improvement?
- Refine & scale up only if the product proves successful.
Example:
- Zappos (Online Shoe Retailer) – Before fully launching, the founder tested demand by listing shoes online, buying them from stores, and shipping them manually. The positive response led to a full-fledged company.
- Starbucks tested Mobile Orders in select locations before rolling it out globally.
Comparison: Customer Co-Creation vs. Learning Launch
| Feature | Customer Co-Creation | Learning Launch |
|---|---|---|
| Purpose | Improve prototypes with customer feedback | Test real-world customer behavior |
| Focus | Iterative design & idea refinement | Measuring actual market demand |
| Process | Customer workshops, user testing, iterative development | Small-scale release, monitoring purchases & reactions |
| Example | Nike testing sneakers with athletes | Zappos listing shoes online before launching fully |
Stage 4: "What Works?"
1. Airbnb – Testing the Home Rental Model
📌 Problem: Would people rent homes from strangers instead of hotels?
📌 Customer Co-Creation: Early users gave feedback on pricing, photos, and trust issues.
📌 Learning Launch: Airbnb founders rented out their own apartment to test the idea before scaling.
📌 Result: The test showed strong demand, leading to Airbnb’s success.
2. Dropbox – Testing Demand with a Simple Video
📌 Problem: Would customers understand and want cloud storage?
📌 Customer Co-Creation: Users suggested features they needed.
📌 Learning Launch: Instead of building the product first, Dropbox made a simple explainer video. The massive interest led to a waitlist of 75,000 users!
📌 Result: The demand was validated, so Dropbox built the product knowing customers would use it.
Key Takeaways:
- Stage 4 (What Works?) ensures an idea is tested before full launch.
- Customer Co-Creation helps refine a product with user feedback.
- Learning Launch tests whether people actually buy the product.
- Companies like LEGO, Nike, Airbnb, and Dropbox use these methods to reduce risk and increase success.
Here are additional case studies from automobile, travel, and education industries showcasing Stage 4: What Works? in Design Thinking using Customer Co-Creation and Learning Launch.
Case Study 1: Automobiles – Tesla’s Over-the-Air Software Updates
Problem:
Traditional car companies required physical recalls for software updates, which was expensive and inconvenient for customers.
Customer Co-Creation:
📌 Tesla tested beta versions of its self-driving and battery management software with real drivers.
📌 Early adopters provided feedback on autopilot, energy consumption, and safety features.
Learning Launch:
📌 Instead of recalling cars, Tesla pushed updates remotely and tracked how drivers responded.
📌 If customers faced issues, Tesla quickly rolled out another update based on real-world feedback.
Result:
🚀 Tesla’s over-the-air updates became a game-changer, reducing recall costs and keeping cars updated remotely.
Case Study 2: Travel – Airbnb’s Early Market Experiment in New York
Problem:
Would people trust renting homes from strangers rather than staying in hotels?
Customer Co-Creation:
📌 Airbnb founders stayed in their hosts' homes to understand customer pain points (trust, safety, pricing).
📌 They added better profile verification, review systems, and insurance based on feedback.
Learning Launch:
📌 Instead of launching globally, Airbnb tested its service in New York, tracking how hosts and guests interacted.
📌 When they saw positive demand, they scaled to other cities.
Result:
🏡 Today, Airbnb is a $100B+ company, proving the success of small, controlled tests before expansion.
Case Study 3: Education – Duolingo’s A/B Testing for Language Learning
Problem:
Would people stick with an online language-learning app instead of traditional classes?
Customer Co-Creation:
📌 Duolingo tested different lesson styles, gamification elements, and difficulty levels with early users.
📌 Feedback led them to add streaks, rewards, and bite-sized lessons to improve engagement.
Learning Launch:
📌 They released multiple versions of lessons and tested which format kept users engaged the longest.
📌 Instead of assuming what works, they tracked actual behavior (which lessons people completed vs. dropped).
Result:
📚 Duolingo became the world’s most popular language-learning app, proving that small experiments lead to better user retention.
Case Study 4: Tech Startups – Instagram’s Pivot from Burbn
Problem:
The founders of Instagram originally built Burbn, an app with check-ins, photo sharing, and social features. However, they noticed users only engaged with the photo-sharing feature.
Customer Co-Creation:
📌 The team studied user behavior and gathered feedback on why people preferred posting photos.
📌 Users wanted simple filters to enhance pictures before sharing them.
Learning Launch:
📌 Instead of improving all Burbn features, they stripped everything away except photo sharing and launched Instagram as a minimal app.
📌 Within 24 hours, Instagram gained 25,000+ users, proving the idea worked.
Result:
📸 Instagram became one of the fastest-growing social platforms, eventually acquired by Facebook for $1 billion.
Case Study 5: Healthcare – IBM Watson’s AI in Cancer Treatment
Problem:
Doctors struggled with analyzing massive amounts of medical data to recommend personalized treatments.
Customer Co-Creation:
📌 IBM partnered with doctors and hospitals to test how AI could assist in diagnosis and treatment recommendations.
📌 Feedback showed that doctors needed clearer insights, not just raw data.
Learning Launch:
📌 IBM piloted Watson AI in specific hospitals, analyzing its effectiveness in real patient cases.
📌 The system improved over time with real-world data, learning from actual doctor feedback.
Result:
🏥 IBM Watson became a trusted AI assistant in healthcare, helping doctors make faster and more accurate treatment decisions.
Case Study 6: Retail – Amazon’s "1-Click" Checkout
Problem:
Customers abandoned their shopping carts because the checkout process was too long and complicated.
Customer Co-Creation:
📌 Amazon tracked customer behavior and realized people wanted faster checkout options.
📌 They experimented with reducing the number of clicks required to purchase.
Learning Launch:
📌 Amazon launched 1-Click checkout as a small test for Prime users.
📌 Data showed a significant increase in conversions, leading to a full rollout.
Result:
🛒 1-Click checkout revolutionized e-commerce, increasing Amazon’s sales and making the buying experience seamless.
Comparison Table: Industry-Specific Testing Examples
| Industry | Problem | Customer Co-Creation | Learning Launch | Final Outcome |
|---|---|---|---|---|
| Automobile (Tesla) | Software updates required costly recalls. | Tested autopilot and battery updates with early users. | Released over-the-air updates and tracked real-world response. | Tesla reduced recall costs and improved customer experience. |
| Travel (Airbnb) | People were hesitant to rent homes from strangers. | Founders stayed in Airbnb homes and improved trust features. | Launched in NYC first, then expanded globally. | Airbnb became a $100B+ company. |
| Education (Duolingo) | Unclear if people would use an online language app. | Tested gamification and lesson structures with users. | A/B tested lesson styles and tracked engagement. | Duolingo became the top language-learning app. |
| Tech Startup (Instagram) | Users engaged only with photo-sharing, not check-ins. | Analyzed user behavior, found demand for photo filters. | Launched a simplified version with only photo-sharing. | Instagram grew rapidly and was acquired for $1B. |
| Healthcare (IBM Watson) | Doctors needed help analyzing medical data. | Tested AI with hospitals and refined based on doctor feedback. | Piloted Watson AI in real-world patient cases. | Watson AI became a valuable tool in healthcare. |
| Retail (Amazon) | Checkout process was too long, leading to abandoned carts. | Studied user behavior and tested faster checkout options. | Launched 1-Click as a small test before full release. | 1-Click checkout boosted Amazon’s conversions. |
Key Takeaways from These Case Studies
✅ Testing assumptions reduces risk before a full launch.
✅ Customer Co-Creation improves early product versions based on real user feedback.
✅ Learning Launch helps measure real-world demand before scaling.
✅ Successful brands (Tesla, Airbnb, Duolingo) use these methods to ensure long-term success.
✅ Tech Startups: Instagram focused on the most engaging feature and became a billion-dollar brand.
✅ Healthcare: IBM Watson improved medical AI with direct doctor feedback.
✅ Retail: Amazon’s 1-Click checkout streamlined e-commerce and increased conversions.
✅ All industries use small-scale testing to reduce risk and improve user experience.
