This content was translated from Korean to English using AI.

When building a product, I have always had the desire to systematically validate whether it is truly what the customer wants.

By chance, I came across Amazon’s Working Backwards method and decided to organize the content.

Jeff Bezos | Is the Working Backwards process optional?

Amazon systematically checks and validates from the customer’s perspective.

  • How? And through what lens and process do they check and validate?

Brief Overview

What is Working Backwards?

  • A method Amazon uses when designing products: “Define the customer’s desired outcome first, then design backwards from there.”

Purpose

  • To create experiences that people truly want
  • To identify what is needed first, rather than “what can be built”

How?

  • Examine the data, create the test question (the customer’s desired outcome) first, then build a plan to solve that question.

Specific Method

  1. Imagine the scene the customer would love most.
    • e.g., “They open the app and pick a travel destination within 1 minute.”
  2. Write that scene as if it were a one-line news headline.
    • This is called a PR (Press Release).
  3. Anticipate and answer tough questions people would ask in advance.
    • FAQ
  4. Based on these results and answers, work backwards to determine how to build it.

Applying the Method

Idea Summary (Input)

  • Product/Service: A service that quickly finds travel information, with user feedback instantly reflected in filtered search results
  • Target Customer: Individual travelers/couples aged 20-40 who need quick comparisons
  • Core Problem: Information overload, inaccurate filters, delayed review reflection
  • Validation Evidence: (Assumed) 12 interviews + simple prototype testing

A. Validation Results Summary (Input Example)

[Hypothesis ID] H-101

  • Customer Problem Hypothesis: There is plenty of travel information, but it is difficult to quickly find results that “match my conditions.”
  • Target Customer/Segment: Individual travelers/couples aged 20-40, high mobile search ratio
  • Success Metrics (Quantitative/Qualitative): Results meeting conditions within 1 minute, 70%+ revisit intention
  • Validation Method: 12 interviews + clickable prototype
  • Key Findings (Evidence Summary):
    • 10/12 experienced “inaccurate filters”
    • 8/12 wanted instant reflection of the latest reviews/feedback
  • Conclusion: Valid
  • Scope Impact: “Trust/reflection speed” elevated as greater value over “fast search”
  • Risks/Unresolved Questions:
    • Real-time feedback reflection may degrade quality
    • Need to address malicious/distorted reviews
  • Next Action: Experiment with minimum viable trust score/spam filtering

B. Roadmap Transition (Outcome to Roadmap Example)

[Roadmap Item] R-201

  • Purpose (Why): Provide trustworthy filtered results within 1 minute
  • Customer Value (Outcome): Quickly find results with the latest feedback reflected
  • KPI/Success Metrics:
    • Search completion time < 60 seconds
    • Feedback reflection satisfaction 4.2/5 or higher
    • Target Release: 2026 Q2 (MVP)
    • Launch Scope: 1-2 regions, 3 categories only
  • Prerequisites: Minimum feedback reflection rules/spam filter functionality
  • Validation Basis: H-101

C. Epic Transition (Roadmap to Epics Example)

[Epic] E-301

  • Linked Roadmap: R-201
  • Problem/Goal: Build core filter UI/UX for fast search
  • Key User Stories:
    • Start searching within 30 seconds using “budget/schedule/companion type”
    • Readjust conditions directly from results
  • Acceptance Criteria (AC):
    • Results displayed < 2 seconds after filter application
    • Results updated within 5 seconds on filter change
  • Technical Considerations: Caching strategy, search indexing
  • Risks/Assumptions: Need to ensure data freshness
  • Deliverables: Filter UI, Search API v1
  • Responsible Team/Owner: Search Team
  • Definition of Done (DoD): 8 out of 10 users in testing rate it “fast”

[Epic] E-302

  • Linked Roadmap: R-201
  • Problem/Goal: A structure where user feedback is quickly reflected in results
  • Key User Stories:
    • When a user leaves a review/tag, it is reflected immediately
    • Results sorted with priority on recent feedback
  • Acceptance Criteria (AC):
    • Reflected within 5 minutes of feedback submission
    • Minimum malicious/duplicate review filtering rules applied
  • Technical Considerations: Event processing, trust score
  • Risks/Assumptions: Spam countermeasures
  • Deliverables: Feedback pipeline, Trust Score v1
  • Responsible Team/Owner: Data/ML Team
  • Definition of Done (DoD): 70%+ of 100 beta users say “reflection is fast”

[Epic] E-303

  • Linked Roadmap: R-201
  • Problem/Goal: Freshness/trust-based result ranking
  • Key User Stories:
    • Latest information appears first
    • Low-trust information is pushed down
  • Acceptance Criteria (AC):
    • Weighting applied to data from the last 30 days
    • Bottom 10% trust results automatically demoted
  • Technical Considerations: Ranking parameter design
  • Risks/Assumptions: Data bias for specific regions
  • Deliverables: Ranking logic v1
  • Responsible Team/Owner: Search Team
  • Definition of Done (DoD): 10% CTR improvement in A/B testing

PR (Press Release) Draft

[Title]
“Find the Travel Information You Want in Under 1 Minute” — Launch of Real-Time Feedback-Powered Filtered Search Service

[Subtitle/Summary]
With search results that instantly reflect travelers’ latest feedback, we help you quickly find the best options even amid information overload.

[Seoul, April 15, 2026]
Today we announce a new filtered search service that helps you find travel information quickly and accurately. This service re-sorts results by rapidly reflecting user feedback, enabling decision-making based on the latest information.

[Customer Problem/Context]
The more search results travelers see, the harder it becomes to choose, and when reviews are reflected late, they are exposed to information that differs from actual experiences.

[Solution]

  • Start exploring within 1 minute with condition-based filters
  • Results sorted based on recent feedback
  • Low-trust information automatically demoted

[Customer Quote]
“The latest reviews being reflected right before my decision reduced uncertainty.”

[Internal Leader Quote]
“The core of travel is trustworthy, up-to-date information. We aim to make travelers’ decisions faster and more accurate by instantly reflecting customer feedback.”

[Details]
The service launches by city/theme, providing a mobile-first experience. The initial version starts with a limited scope of 1-2 regions and 3 categories.

[Pricing/Launch/Availability]
Beta launch planned for Q2 2026. The initial beta will be offered for free.

[Call to Action]
Beta signup: (link/contact channel)


FAQ Draft

[Customer FAQ]

  1. Who is this service for?
  • Individual travelers/couples aged 20-40 who need to compare travel options quickly.
  1. What is the biggest value?
  • You can quickly find results with the latest feedback reflected.
  1. How is it different from existing travel search?
  • Users’ recent feedback is reflected immediately, resulting in higher result reliability.
  1. How fast is feedback reflected?
  • In the beta, the goal is reflection within 5 minutes of submission.
  1. How are malicious/distorted reviews handled?
  • They are demoted via duplicate/spam pattern filtering and trust score-based processing.
  1. Which regions/categories will launch first?
  • Initially offered in a limited scope of 1-2 regions and 3 categories.
  1. What is the pricing?
  • Free during the beta period; a pricing model will be announced at official launch.

[Internal FAQ]

  1. Why now?
  • With travel demand recovery, the importance of up-to-date information has increased, and this is an area with recurring customer complaints.
  1. What are the success metrics?
  • Search completion time < 60 seconds, feedback reflection satisfaction 4.2/5 or higher.
  1. Why is the MVP scope this size?
  • The scope was reduced to validate the core value (fast search + fast feedback reflection).
  1. What are the main risks?
  • Spam/malicious reviews, data bias, quality degradation from real-time reflection.
  1. What is the mitigation plan?
  • Trust scores and minimum filtering rules are included in the beta, with adjustments via A/B testing.