# Hear how every React applicant debugs a real problem

> React interview questions with follow-ups and strong and weak answers, plus how Tracy asks them of every React applicant and returns the evidence.

- URL: https://tracyinterview.ai/blog/react-interview-questions
- Published: 2026-11-20
- Author: Louis

> **In short** Fifteen React questions, each with a follow-up and what a strong answer sounds like, test rendering, effects, performance, structure and judgment in a first conversation. Tracy AI Interviewer asks them of every applicant, can add a coding task to the same interview and follows up on the approach and its costs. Your engineers read the recordings and decide who gets the hands-on round.

Every React CV lists hooks, state management and performance. A CV says hooks. The follow-up shows whether the developer has ever opened the profiler before reaching for memoisation, or wraps everything in it. Under each question sits a follow-up that asks for the reasoning behind the habit.

The questions are for engineering managers and recruiters hiring mid-level and senior React developers, and fit a first conversation lasting 20 to 30 minutes. Pairing or live coding belongs later, with your own engineers. Candidates can use the follow-ups as practice; a developer with ten years of experience should answer each with a case from real work.

Tracy AI Interviewer takes that first conversation with every applicant. For neighbouring roles, see [frontend developer interview questions](/blog/frontend-developer-interview-questions) and [system design interview questions](/blog/system-design-interview-questions), and for the wider process, [technical interviews](/use-cases/technical-interviews).

## What makes a good React interview question?

A good React question asks for a case from real work and has a follow-up beside it. A written list does that well: every applicant gets the same questions, and the Listen for column says what a strong answer sounds like before anyone joins.

The list cannot ask what to do next. That depends on what the developer just said, and it is the part Tracy takes over.

## Fifteen React interview questions, grouped by what they test

One question from each group will do for a first pass; open with the same ones for everyone. Save each follow-up for after the first answer, and write down what a strong answer sounds like first.

### Group 1: rendering and state

*Table 1: Rendering and state, with follow-ups and what a strong answer contains.*

| Ask | Follow up with | Listen for |
| --- | --- | --- |
| 1. A component re-renders more often than you expect. How do you find out why? | What would you check before reaching for memoisation? | Uses the profiler, finds what changed (props, state, context or the parent) and fixes the cause. Memoisation comes late. |
| 2. Where should a piece of state live: in the component, lifted up, in context or in a store? | Pick a case from your last project. | As close to its use as possible, lifted when siblings need it, context for rarely changing values. Asks how often it changes. |
| 3. What is derived state, and why do people get it wrong? | How do you fix a component that copies props into state? | Computes from props and state during render because a copy goes stale. Uses a key to reset when identity changes. |

### Group 2: effects and data

*Table 2: Effects and data, with follow-ups and what a strong answer contains.*

| Ask | Follow up with | Listen for |
| --- | --- | --- |
| 4. When do you use an effect, and when not? | What bug does a missing dependency cause? | Effects sync with something outside React. Not for values computed during render or for responding to a click. Knows stale closures and cleanup. |
| 5. A search list shows stale results when the user types fast. What is going on? | How would you fix it? | Requests return out of order. Cancels the old request or ignores any response that is not the latest. |
| 6. How do you load and cache server data? | When would you not use a data library? | Server state differs from interface state: caching, retries, invalidation. A small app may not need a library. |

### Group 3: performance

*Table 3: Performance, with follow-ups and what a strong answer contains.*

| Ask | Follow up with | Listen for |
| --- | --- | --- |
| 7. A page gets slow after a long list is added. Where do you start? | What would you do for ten thousand rows? | Measures first. Virtualises long lists, uses stable keys, avoids creating work in render, and checks network and bundle too. |
| 8. When do useMemo and useCallback help, and when are they noise? | How do you tell in a real app? | Helps when a memoised child depends on the value or the computation is costly. Otherwise adds cost. Measures. |
| 9. How do you keep a bundle small as the app grows? | What do you split first? | Route-level splitting and lazy loading, a dependency audit, a bundle analyser. Does not split before measuring. |

### Group 4: structure and testing

*Table 4: Structure and testing, with follow-ups and what a strong answer contains.*

| Ask | Follow up with | Listen for |
| --- | --- | --- |
| 10. How do you structure a large React app so a new person can find things? | What breaks first as the team grows? | Folders by feature, clear lines between interface, data and logic, shared components with owners, conventions written in the repo. |
| 11. What do you test in a component, and what do you leave out? | Show me a test you would delete. | Behaviour as the user sees it after an action, mocks at the network edge. Avoids tests of implementation details and snapshot-only tests. |
| 12. How do you handle errors in a React app? | What does the user see when one widget fails? | Error boundaries, loading and empty states, reporting. One failing widget does not blank the page. |

### Group 5: judgment and experience

*Table 5: Judgment and experience, with follow-ups and what a strong answer contains.*

| Ask | Follow up with | Listen for |
| --- | --- | --- |
| 13. Tell me about a hard bug you fixed in a React app. | How did you know it was fixed for good? | A real bug, how they reproduced it, the root cause, and a test or guard they added. |
| 14. What have you changed your mind about in React over the years? | What do you still do the old way, and why? | A specific shift, such as classes to hooks or global stores to server-state libraries, with reasons from their own work. |
| 15. How do you build a custom dropdown so it is accessible? | What would a keyboard user hit first? | Prefers a native element or an established headless library. Focus handling, roles, arrow keys. Tests with a keyboard and a screen reader. |

## What a developer who has debugged this says

The pairs below are written for questions 1, 4 and 13. They are not quotes from real candidates.

*Table 6: Weak and strong React answers to questions 1, 4 and 13.*

| | Weak answer | Strong answer |
| --- | --- | --- |
| Question 1, the extra renders | "I would wrap it in React.memo." | "I would open the profiler and see why it rendered: the parent, state, props or context. Often the parent creates a new object on every render. Memo comes last." |
| Question 4, effects | "useEffect runs when the component loads, so I use it for anything that needs to run." | "For syncing with something outside React, such as a subscription, a timer or a fetch with cleanup. Not for values I can compute during render." |
| Question 13, the hard bug | "I cannot think of a specific one." | "A form lost input after the parent re-rendered with a changed key. I logged mounts to find it, fixed the key and added a test that types, re-renders and checks the value." |

## A stale search list, followed up until the fix is proven

The candidate here is made up, as is every line. The chain starts at question 5 and each follow-up picks up from the previous answer.

> **Tracy:** A search list shows stale results when the user types fast. What is going on?
>
> **Candidate (fictional):** Requests come back out of order. The slow one for "re" lands after the one for "react".
>
> **Tracy:** How would you fix it?
>
> **Candidate:** Cancel the old request when a new one starts, or ignore any response that is not the latest.
>
> **Tracy:** How would you prove the fix works?
>
> **Candidate:** A test with two requests where the first resolves last, and check the screen shows the second.

The last question separates a candidate who proves a fix from one who trusts it.

![Transcript card: stale search results, candidate says requests return out of order, Tracy asks how they would fix it.](/blog/images/react-interview-questions-1.png)

*Tracy picks up the out-of-order answer and asks how they would fix it, so your engineer hears whether the developer proves a fix. The exchange is invented. Illustration with sample data*

The follow-up shows whether the candidate proves a fix.

## What Tracy adds to React screening

01

### Reach every applicant

Each applicant can start on click, at any hour, with no scheduling and no recruiter on the call. Developers join when it suits them, and nobody waits for a calendar slot.

02

### Ask the next question

Tracy asks in the person's own field, follows what they said, and waits for the answer. "Requests come back out of order" is followed by how the fix would be proved.

03

### Keep the evidence

Every result includes a recording, a timestamped transcript, and a written assessment per question with its reasoning. An engineer can check the profiler answer and the fix before the live round.

![Four steps: write the questions, applicants click, Tracy interviews, engineers review.](/blog/images/react-interview-questions-3.png)

*Tracy runs the spoken screen and the coding follow-ups; your engineers read the evidence and choose who goes on.*

From question list to evidence.

## What a hundred first calls cost your team

At 20 to 30 minutes a conversation, a hundred React applicants is 33 to 50 hours of live calls.

- Want a coding task beside the spoken questions? Tracy's follow-ups cover the approach and the trade-offs, and your engineers read the answers.
- Tracy turns your job description into draft questions, and you edit them. See the [product page](/product).
- Applicants need a browser and a link, not an account or a calendar slot. The join screen says up front that it is recorded and transcribed; see [the applicant view](/participant-experience).
- Each result carries the recording, the transcript and an ATS note, so a manager can check the profiler answer and the fix. [Sample report](/blog/ai-interview-reports).
- In a video interview your reviewer sees face detection, tab changes, gaze, external screens, delay before answers and fullscreen exits, with the evidence for each. Read [how those signals are reviewed](/blog/ai-interview-cheating-detection) before treating one as meaningful.

Candidates rate the Tracy interview experience 92 on NPS. A credit is $1 and pays for 20 minutes of interview, counted per started block once a result exists, so a 25-minute interview uses two. Your first three results are free and need no card, and engineers who read reports need no seats. See [pricing](/pricing).

![Report for a fictional React developer: recording, transcript and the answers to three questions.](/blog/images/react-interview-questions-2.png)

*Each result opens to a recording and timestamped transcript, so your engineer reads the applicant's own words before the live round. Invented content; not a screenshot. Illustration with sample data*

Your engineer reads each React answer and opens the recording and transcript behind it.

## Three decisions your engineers keep

1. The engineering manager reads each report and decides who gets the hands-on round.
2. Calibrate: ask one senior developer to read five transcripts cold. If the transcripts tell a different story from the report, reword those questions before the next hundred applicants.
3. The hands-on round: pairing, a take-home or a code review stays with your team. See [coding interview follow-up questions](/blog/coding-interview-follow-up-questions) for what to ask there.

For three hires, a quick call with your lead developer is the simple route (compare [AI interview vs human interview](/blog/ai-interview-vs-human-interview) first). A hundred applicants need [AI technical interviews](/blog/ai-technical-interviews) instead. For the backend side of a full-stack opening, see [system design interview questions](/blog/system-design-interview-questions).

## Common questions

### What are good React interview questions?

Ask about rendering and state, effects and data, performance, structure and testing, and past bugs. Put a follow-up under each question so the answer has to come from experience.

### What React interview questions suit experienced developers?

Go to groups 3 to 5 and ask for a case from real work: a performance problem, a test they would delete, a bug and how they knew it was fixed.

### What React interview questions suit a developer with 10 years of experience?

Add what they changed their mind about and how they structure a large app for a growing team. Ask for reasons from their own projects.

### How should I prepare for React interview questions?

Learn why each hook exists, not only its syntax. Be ready to describe a bug you fixed and how you proved the fix.

### Can Tracy run a coding task for React candidates?

Yes. A coding task can join the same interview, with follow-ups on how the candidate would approach it and why. The report gives a recording, a timestamped transcript and an ATS note.

### Does Tracy pick the React developers who move forward?

No. Tracy reports what each applicant said; your engineers read it and pick.

---

**Send one link to every React applicant** Add the React questions you picked, send one link and read the recordings and reports as they arrive. The first 3 results are free. [Try it free](https://tracyinterview.ai/connect?utm_source=blog&utm_medium=seo&utm_campaign=react-interview-questions&utm_content=try-free) [Book a demo](https://cal.com/tracy-ai/tracy-demo-web?utm_source=blog&utm_medium=seo&utm_campaign=react-interview-questions&utm_content=demo)

## Sources

- [Tracy AI Interviewer home page](https://tracyinterview.ai/), retrieved 14 Nov 2026.
- [Tracy AI Interviewer pricing](https://tracyinterview.ai/pricing), retrieved 14 Nov 2026.
- [Tracy AI Interviewer Connect page](https://tracyinterview.ai/connect), retrieved 14 Nov 2026.
