Articles

Free Interview Question Randomizer by Company, Role, and Round

Build a free interview question randomizer by company, role, round, and difficulty with no-repeat controls, spreadsheet formulas, and practice tracking.

This free interview question randomizer helps you choose a useful practice prompt without spending the session browsing. Set a target company, role, interview round, and optional difficulty; draw one unused question from the matching pool; then record the result before drawing again. You can run it manually from the tables below or build the spreadsheet version in a few minutes.

Random selection is most useful after you define a relevant pool. A backend engineer preparing for a system design round should not receive an unrelated product case simply because it is random.

Use the randomizer now

Choose one value in each row. Use Any only when you genuinely want a mixed session.

Filter	Focused example	Broad option	Why it matters
Company	Google	Any	Preserves company relevance or increases variety
Role	Software Engineer	Any	Keeps the question aligned with expected work
Round	Coding	Any	Matches the skill and time pressure of the next stage
Difficulty	Medium	Any	Prevents every session from becoming either trivial or exhausting
Used?	Exclude checked rows	Include all	Avoids accidental repetition

Then follow this five-step draw:

Filter your question bank to the selected values.
Remove questions already marked Used unless this is a scheduled retest.
Count the remaining rows as N.
Generate a random integer from 1 through N and take that row.
Freeze the result, complete the session, and record the outcome before drawing again.

If only one question remains, that is the result. If no question remains, broaden one filter or deliberately reset the used flag. Do not keep rerolling until a familiar prompt appears; that converts random practice into preference selection.

Why company, role, and round come before randomness

A randomizer is a selection mechanism, not a curriculum. The filters define what counts as eligible practice.

Company narrows the pool toward candidate-reported question records associated with the employer. Those records can improve relevance, but they are not a prediction of your exact interview. Teams, locations, levels, and recruiting cycles can differ.

Role protects against false similarity. Software engineering, data engineering, machine learning, product management, and quantitative roles may all include technical reasoning, but they emphasize different artifacts and decisions. A role filter keeps the task close to the work you may be asked to demonstrate.

Round determines the practice conditions. An online assessment usually rewards independent implementation under a timer. A technical screen adds explanation and follow-ups. System design tests scoping and trade-offs. Behavioral rounds require specific evidence from your own experience. Randomizing across rounds without changing the timer or scoring method makes the result hard to interpret.

This is consistent with structured-interview practice: the U.S. Office of Personnel Management describes questions as measures of job-related competencies and recommends selecting content areas from the tasks and competencies critical to the position. For candidates, the practical lesson is simple: build the eligible pool around the signal you need to rehearse.

Build the question-bank sheet

Create a tab named Question Bank with these columns:

Column	Field	Example
A	Company	Amazon
B	Role	Software Engineer
C	Round	Coding
D	Difficulty	Medium
E	Complete question title	Traverse All Reachable Pages With Parallel BFS
G	Used?	Checkbox
H	Last result	Partial - missed duplicate suppression
I	Last practiced	2026-08-28

Keep the complete title and URL together. Do not create a URL by guessing the title slug. Open the destination once when adding a row, and treat the stored round as evidence about that question record rather than a guarantee about a future employer process.

On a second tab named Randomizer, place the selected company in J2, role in J3, round in J4, and difficulty in J5. Add Any to each dropdown. Put the result formula in J7.

Google Sheets formula and Excel setup

This Google Sheets formula filters the bank, excludes checked questions, and returns one randomly selected title-and-URL row.

=IFERROR(
  LET(
    pool,FILTER('Question Bank'!E2:F,
      IF($J$2="Any",'Question Bank'!A2:A<>"",'Question Bank'!A2:A=$J$2),
      IF($J$3="Any",'Question Bank'!B2:B<>"",'Question Bank'!B2:B=$J$3),
      IF($J$4="Any",'Question Bank'!C2:C<>"",'Question Bank'!C2:C=$J$4),
      IF($J$5="Any",'Question Bank'!D2:D<>"",'Question Bank'!D2:D=$J$5),
      'Question Bank'!G2:G<>TRUE),
    INDEX(pool,RANDBETWEEN(1,ROWS(pool)),0)
  ),
  "No eligible unused question"
)

FILTER creates the eligible pool. RANDBETWEEN(1, ROWS(pool)) generates an inclusive row number, and INDEX returns that row. Both Google and Microsoft document RANDBETWEEN as a random integer function; Microsoft also notes that it can change when a workbook recalculates.

In Excel 365, convert the bank to a Table and use the same LET, FILTER, INDEX, and RANDBETWEEN sequence with structured table references. Open-ended ranges such as A2:A are Google Sheets syntax; use a Table or bounded ranges in Excel.

Therefore, freeze the draw before beginning. Copy the result and paste values into a Current Question area, or immediately mark the chosen row and work from its URL. Otherwise an unrelated edit may replace the prompt midway through the session.

Choose the right round label

Use round labels that change how you practice, not every name a company may use internally.

Round label	Recommended conditions	Evidence to capture
Online assessment	Independent, invitation-matched timer and allowed tools	Correctness, hidden-edge readiness, and completion time
Coding screen	Shared editor or verbal explanation with follow-ups	Framing, algorithm, implementation, tests, and communication
Repository or debugging	Existing codebase, tests, debugger, and written assumptions	Diagnosis, change scope, regression safety, and code reading
System design	45-60 minutes with explicit requirements and scale	APIs, data model, capacity, reliability, and trade-offs
Behavioral	Spoken answer plus two skeptical follow-ups	Specific situation, personal action, result, and reflection
Hiring manager	Project deep dive, role motivation, and judgment questions	Scope, ownership, decisions, collaboration, and lessons

Map company-specific names into one of these practice labels. For example, a “technical deep dive” may belong under hiring manager or system design depending on what the invitation says. Your recruiter message remains the source of truth.

Use three randomization modes

Target mode locks company, role, and round. Use it during the final week before a known interview. The pool is smaller, so exclude recently practiced questions and schedule retests deliberately.

Role mode locks role and round but sets company to Any. This is useful earlier in preparation because it exposes different prompt wording while preserving the skill being tested. Varying examples can make practice less dependent on recognizing one familiar presentation.

Full-loop mode locks company and role but allows several planned rounds. Do not draw uniformly from every question if the bank contains hundreds of coding records and only a few behavioral prompts. First randomize the round from a balanced schedule, then randomize a question inside that round.

A practical five-session full loop could contain two coding draws, one design or practical-engineering draw, one behavioral draw, and one hiring-manager or project draw. Adjust that mix to the actual process rather than treating it as an employer standard.

Prevent repetition without losing useful retests

Random systems can repeat. That is mathematically normal but often wasteful for interview preparation. Add three controls:

Mark a question Used after the session, not after merely opening it.
Store the date and outcome so a weak attempt becomes a scheduled retest instead of disappearing.
Reset a question only after a delay and require a fresh solution without notes.

Use an outcome vocabulary such as Solved independently, Solved with hint, Partial, and Could not start. The label matters less than consistency. Add one short error note: “missed empty input,” “chose an unscalable data model,” or “behavioral result had no evidence.” That note determines what to study before the retest.

Research on retrieval practice supports actively producing an answer instead of only rereading one. Research on varied examples also suggests that applying knowledge to different examples can improve transfer to new examples. That does not mean maximum randomness is always best. Novices may need a short blocked lesson before mixed practice; the randomizer should test a skill after you have learned its basic method.

Score the session before drawing again

Use one outcome card for every round:

Prompt: [complete title and URL]

Conditions: [round, timer, tools, evaluator]

Result: [independent / hint / partial / incomplete]

Evidence: [one observable strength and one observable gap]

Next action: [study, focused drill, or dated retest]

For coding, record correctness, complexity, tests, and explanation. For system design, record requirements, APIs, data model, scale, failure handling, and trade-offs. For behavioral questions, record whether the story included your responsibility, decisions, evidence, and reflection.

Do not use the randomizer as a slot machine for confidence. Finish the scorecard before the next draw. A smaller number of completed, reviewed questions creates more useful evidence than a long history of rerolls.

A seven-day random-practice loop

Day 1: build the bank, define round labels, and complete one untimed baseline. Day 2: run a role-mode coding draw and record every failed edge case. Day 3: run a focused drill on the lowest skill, without using the randomizer.

Day 4: draw a different prompt from the same role and round. Day 5: switch to behavioral or hiring-manager practice and add two follow-ups. Day 6: run a company-targeted full loop with realistic timing. Day 7: review the evidence log, schedule retests, and remove bad or mislabeled rows.

The alternating pattern matters. Randomized assessment tells you what transfers; focused drills repair what failed.

Frequently asked questions

Is a random interview question generator better than a study plan?

No. A study plan defines coverage, sequence, and review. The randomizer selects a prompt inside that plan and reduces choice bias. Use focused lessons for new material, random draws to test transfer, and scheduled retests for weak questions. Random selection alone can leave important skills underrepresented.

Should I randomize across companies?

Use company Any when you want varied wording and broader role practice. Lock the company when an interview is close and the question bank has enough relevant records. In both cases, treat candidate-reported company tags as preparation evidence rather than a guarantee of what you will receive.

How many questions should be in each pool?

There is no universal minimum, but a pool of one or two questions is predictable and a huge unfiltered pool is noisy. Start with enough relevant questions to prevent memorization, then watch coverage by round and skill. If one category dominates, balance the round first and randomize within it.

How do I stop Google Sheets from changing the selected question?

RANDBETWEEN is volatile and can update when the sheet recalculates. After drawing, copy the selected title and URL and paste values into a fixed Current Question area. Complete and score that prompt before recalculating or editing the filters.

Should solved questions be removed forever?

No. Exclude them from ordinary draws, but retain the row, last result, and practice date. Schedule a later closed-notes retest when the first attempt required help, exposed a recurring weakness, or tests an important pattern. Repetition is useful when it is spaced and intentional.

Final takeaway

A useful interview question randomizer does three things in order: defines a relevant pool, selects without preference, and turns the result into evidence. Filter by company, role, and round; freeze one unused draw; practice under the right conditions; and record the gap before you continue.

The goal is not to guess the exact interview question. It is to become reliable when the next relevant prompt is not the one you hoped to see.

Sources and Further Reading

The spreadsheet workflow is an original practice tool, not an employer selection system or prediction of a specific interview process.