Vetora logo
🎯Interview Toolkit

The Four-Step System Design Framework

Master Vetora's recommended four-step approach to system design interviews: Clarify Requirements, High-Level Design, Deep Dive, and Wrap Up. Learn how to time-box each phase for a 45-minute interview and compare with other popular frameworks.

Overview

The four-step system design framework is a structured approach that transforms an open-ended design question into a manageable, time-boxed conversation. In a 45-minute interview, every minute counts, and candidates who jump straight into drawing boxes on the whiteboard without clarifying requirements or who spend 30 minutes on a single component consistently underperform. The framework provides guardrails that ensure you demonstrate the full range of skills interviewers evaluate: requirements gathering, architectural thinking, technical depth, and operational awareness.

The first step, Clarify Requirements, takes approximately 5 minutes. You ask targeted questions to pin down functional requirements (what the system does), non-functional requirements (latency, throughput, availability, consistency), scale estimates (DAU, QPS, storage), and constraints (budget, existing infrastructure, regulatory). This step signals that you think before you build -- a trait that separates senior engineers from juniors. The second step, High-Level Design, takes about 15 minutes. You sketch the major components (clients, load balancers, API servers, databases, caches, message queues), define the data flow between them, and outline API contracts. The goal is a complete end-to-end architecture at a high level, not a detailed design of any single component.

The third step, Deep Dive, is where you spend the bulk of your time -- roughly 20 minutes. You pick 2-3 areas to elaborate based on what is most interesting or challenging about the problem. This might be the database schema and sharding strategy, the caching layer and invalidation policy, or the message queue design for async processing. The interviewer may guide you toward specific areas or let you choose. Either way, this is where you demonstrate depth: discuss trade-offs, reference specific technologies, and reason about failure modes. The fourth step, Wrap Up, takes the final 5 minutes. You proactively discuss bottlenecks in your design, propose a scaling plan for 10x growth, mention monitoring and alerting strategies, and summarize key trade-offs. This step demonstrates operational maturity.

Compare this with other popular frameworks: ByteByteGo's RESHADED (Requirements, Estimation, Storage schema, High-level design, API design, Detailed design, Evaluation, Distinctive component) covers similar ground but with more explicit sub-steps. DesignGurus' framework emphasizes a separate API design phase. The four-step framework deliberately keeps the structure simple -- four phases are easy to remember under pressure, and the interviewer can see your structure without you narrating a checklist. The key insight is that all these frameworks converge on the same principle: breadth first, then depth, with explicit time for requirements and wrap-up.

Key Points
  • 1The 5/15/20/5 minute split ensures balanced coverage. Spending too long on requirements leaves no time for design; skipping the wrap-up misses the chance to demonstrate operational thinking and self-awareness about your design's limitations.
  • 2Step 1 (Clarify Requirements) is where you establish scope. A URL shortener with 100 users and one with 100 million users are fundamentally different systems. Clarifying scale early determines every subsequent design decision.
  • 3Step 2 (High-Level Design) should produce a complete end-to-end architecture. Every request should have a visible path from client to response. Resist the urge to detail any single component at this stage.
  • 4Step 3 (Deep Dive) is your chance to show senior-level depth. Pick areas where the problem is hardest -- the hot partition, the consistency challenge, the real-time constraint -- and reason through trade-offs with specific technologies and numbers.
  • 5Step 4 (Wrap Up) is where many candidates leave points on the table. Proactively naming bottlenecks, proposing monitoring, and discussing a 10x scaling plan signals operational maturity that distinguishes L5/E5 candidates from L4/E4.
  • 6Time-boxing is non-negotiable in a 45-minute interview. Set a mental timer for each phase and announce transitions explicitly: 'I think we have a solid high-level design. Let me now deep-dive into the database layer.' This shows the interviewer you are in control of the conversation.
Simple Example

The Architecture Tour Analogy

Think of a system design interview like giving a tour of a house you are building. Step 1 (Clarify Requirements) is asking the client how many bedrooms they need, their budget, and whether they want a pool -- you cannot design what you do not understand. Step 2 (High-Level Design) is showing the floor plan: rooms, hallways, plumbing routes, electrical layout. Step 3 (Deep Dive) is walking into 2-3 rooms and discussing the materials, fixtures, and engineering details. Step 4 (Wrap Up) is stepping back outside, pointing out that the foundation can support a second story if they want to expand, and mentioning where you would add smoke detectors. The client wants to see that you understand both the big picture and the fine details.

Real-World Examples

Google

Google L5 (Senior) system design interviews explicitly evaluate structured thinking. Interviewers use a rubric that checks whether the candidate clarified requirements, proposed a high-level design before diving into details, and addressed scalability and reliability. Candidates who follow a clear four-step framework consistently score higher on the 'problem-solving approach' axis because the interviewer can see their thought process rather than trying to reconstruct it from scattered whiteboard notes.

Meta

Meta E5 (Senior) design interviews generate strong signals from structure. The feedback form includes sections for 'scoping and requirements,' 'architecture,' 'deep technical discussion,' and 'operational considerations' -- mapping directly to the four steps. Meta interviewers report that candidates who self-direct through a structured framework require fewer nudges, which is itself a positive signal for the 'independence' dimension.

Amazon

Amazon SDE-3 system design interviews layer Leadership Principles onto the technical framework. 'Customer Obsession' maps to Step 1 (clarifying what the user actually needs), 'Dive Deep' maps to Step 3 (technical depth), and 'Think Big' maps to Step 4 (scaling plan). Candidates who structure their approach find it easier to weave in LP examples because each phase naturally aligns with specific principles.

Trade-Offs
AspectDescription
Structure vs FlexibilityA rigid framework ensures you cover all bases but may feel formulaic. The best candidates internalize the structure so it becomes invisible -- they follow the steps naturally while adapting to the interviewer's interests and the problem's unique challenges.
Breadth vs DepthThe framework allocates 15 minutes to breadth (high-level design) and 20 to depth (deep dive). Some problems demand more depth (complex consistency requirements) while others need more breadth (multi-service architectures). Adjust the split by plus or minus 5 minutes based on the problem.
Candidate-Led vs Interviewer-LedThe framework assumes you drive the conversation. Some interviewers prefer to direct the discussion with specific questions. In that case, use the framework as a mental checklist rather than a rigid agenda, and smoothly adapt to the interviewer's flow.
Simplicity vs CompletenessFour steps are easy to remember under pressure but compress some important activities (API design, schema design) into broader phases. More granular frameworks like RESHADED make these explicit, which can help less experienced candidates remember to cover them.
Case Study

Applying the Framework to a Chat System Design

Scenario

A candidate is asked to design a WhatsApp-like messaging system in a 45-minute Google L5 interview. The problem is broad, with many possible directions: 1:1 messaging, group chats, media sharing, read receipts, presence indicators, and end-to-end encryption. Without a framework, candidates commonly spend 25 minutes on the message delivery pipeline and run out of time before discussing storage, scaling, or offline delivery.

Solution

Using the four-step framework, the candidate spends the first 5 minutes clarifying: 1:1 and group messaging (up to 500 members), text and image support, read receipts, 500M DAU, 100B messages/day, p99 delivery latency under 500ms. The next 15 minutes produce a high-level design: mobile clients, WebSocket gateway, message service, group service, media service, message store (Cassandra), user presence cache (Redis), notification service, and object storage for media. The 20-minute deep dive covers the WebSocket connection management (how to route messages to the right server), Cassandra schema design for message storage (partition by conversation ID, cluster by timestamp), and group message fan-out strategy (write-time vs read-time fan-out trade-off). The final 5 minutes address bottlenecks (hot groups with thousands of members), a scaling plan (shard WebSocket gateways by user ID hash), and monitoring (message delivery latency histogram, undelivered message queue depth).

Outcome

The candidate received a 'Strong Hire' rating. The interviewer feedback specifically noted the clear structure, the efficient requirements gathering that scoped the problem quickly, and the depth of the deep-dive on message fan-out. The wrap-up discussion on monitoring and scaling demonstrated the operational maturity expected at L5. The same candidate had failed a previous interview by spending 35 minutes on database schema without a high-level design.

Common Mistakes
  • Skipping requirements clarification and jumping straight into design. This leads to building the wrong system -- a URL shortener for 100 users does not need sharding, and one for 1 billion users cannot rely on a single database. The first 5 minutes prevent 40 minutes of wasted effort.
  • Spending too long on the high-level design trying to make it perfect. The high-level design is a skeleton, not a finished product. Get the major components on the board in 15 minutes and refine during the deep dive. Perfectionism at this stage steals time from the depth that earns strong ratings.
  • Deep-diving into the wrong areas. Choosing to elaborate on a well-understood commodity component (standard REST API) instead of the hard parts of the problem (consistency model, real-time delivery, conflict resolution) misses the opportunity to demonstrate senior-level thinking.
  • Forgetting the wrap-up. Many candidates use all remaining time on the deep dive and end abruptly when time is called. The wrap-up is your final impression -- proactively discussing bottlenecks and scaling shows you can think beyond the immediate problem.
Related Concepts

See The Four-Step System Design Framework in action

Explore system design templates that use the four-step system design framework and run traffic simulations to see how these concepts perform under real load.

Browse Templates

Practice the 4-step framework on a URL shortener design

Metrics to watch
throughput_rpsp99_latency_msstorage_growth_gbavailability_pct
Run Simulation
Test Your Understanding

1In the four-step framework, approximately how much time should be allocated to the Deep Dive phase in a 45-minute interview?

2Why is the Clarify Requirements step critical even when the problem seems straightforward?

3What should you do during the Wrap Up phase?

Deeper Reading