Part 5 of 9 · 2 min read
Step 3: Go deep
The longest step: choosing which parts to dig into and managing time.
Think of a surgeon after the first overview. The surgeon does not examine every organ equally. The surgeon concentrates where the problem is.
Step 3 takes about 10 to 25 minutes, the longest of the four. By now both sides have agreed on the goal and the blueprint, and have some idea where the hard parts are.
The designer and the listener choose which components to investigate. The choice depends on the problem:
- For a URL shortener, an interesting part is how a long address becomes a short code.
- For a chat system, interesting parts are keeping delay low and showing who is online.
- For a feed, an interesting part is how posts reach followers.
- For a senior candidate, the listener may ask about bottlenecks and cost.
Two rules guide this step.
- Go deep where it matters most. Start with the most critical or riskiest part, not the part that is most fun.
- Watch the clock. It is easy to get lost in a detail that proves nothing about design skill. Explaining the full inner workings of a ranking formula, for example, takes precious minutes and does not show that the designer can build a scalable system.
A good deep dive follows a pattern:
- State the problem in the part: "The short code must be unique and hard to guess."
- Give two or three options.
- Compare them on the driving qualities.
- Choose one and say what is given up.
- Check the choice against the numbers from step 2.
Interview angle
Answer in your head first, then open it.
Why offer two options during a deep dive?
Offering two approaches and comparing them is worth more than offering one. It shows that the designer sees the space of options, and that the final choice was a decision and not luck.
Check yourself:
- Why start with the riskiest part?
- Name the five parts of a good deep dive.