AI InterviewGet started

Design Search Autocomplete: System Design Interview Practice

Design Search Autocomplete as a system design interview question: As a user types into a search box, suggest the top completions for what they have typed so…

How do you answer "Design Search Autocomplete" in a system design interview?

Store popular queries in a trie and keep, at every prefix node, the top few queries below it ranked by frequency. A keystroke walks one node down and returns that stored list. An offline pipeline aggregates query logs, rebuilds the trie periodically and swaps it onto replicated, cached servers, sharded by prefix when it outgrows one machine.

  • Level: Intermediate
  • Scored out of: 100
  • Cost: Free to read; a free account gets the review

As a user types into a search box, suggest the top completions for what they have typed so far.

The scale

10 million daily users, a suggestion request on every keystroke, answers within 100 ms.

What a strong answer covers

  1. Requirements (20 points). States the functional requirements in scope and the non-functional ones that drive the design (scale, latency, availability, consistency), and leaves out what is out of scope.
  2. Scale estimates (15 points). Turns the stated scale into requests per second (average and peak), storage and bandwidth, with the arithmetic shown, and uses the numbers to make a decision.
  3. API and data model (20 points). Defines the main API calls with their inputs and outputs, and a data model whose keys and store fit the access patterns.
  4. Architecture (25 points). Chooses components that fit the requirements and explains how a request moves through them, for the main read and write paths.
  5. Bottlenecks and trade-offs (20 points). Names the real bottlenecks and failure points at this scale and how the design handles them, weighing the alternatives (consistency vs availability, push vs pull, cost).

How the practice works

Write your design in sections — functional and non-functional requirements, scale estimates, the API, the data model, the request flow and a deep dive on the bottlenecks — and pick the components to draw the architecture. A review scores it on the five areas above in under a minute, then you can defend it in an AI mock interview.

The reference design

Design Search Autocomplete (Typeahead): the walkthrough, with animated figures.

Common questions

How do you answer "Design Search Autocomplete" in a system design interview?

Store popular queries in a trie and keep, at every prefix node, the top few queries below it ranked by frequency. A keystroke walks one node down and returns that stored list. An offline pipeline aggregates query logs, rebuilds the trie periodically and swaps it onto replicated, cached servers, sharded by prefix when it outgrows one machine.

What scale is "Design Search Autocomplete" designed for here?

10 million daily users, a suggestion request on every keystroke, answers within 100 ms. Your estimates turn that into requests per second, storage and bandwidth before you pick components.

How is a design scored?

Out of 100 on a fixed rubric: requirements (20), scale estimates (15), api and data model (20), architecture (25), bottlenecks and trade-offs (20). The review names what the design covered and what it missed, then points to the reference design.

Is "Design Search Autocomplete" a good first system design problem?

It is rated intermediate. Start with the beginner problems — the URL shortener and the rate limiter — and the System Design notes on load balancing, caching and sharding first.

Do I need an account?

Not to read the problem or the reference design. Getting your design reviewed needs a free account.

More design problems