Online Assessment Difficulty Predictor
Select a company, role, and level. See the real Easy/Medium/Hard split, top topics, and estimated prep time — calculated from actual candidate reports in our database.
Want the actual questions?
Unlock full question text, candidate solutions, and round-by-round breakdowns.
How the difficulty predictor works
This tool queries LeakCode's database of 23,721+ real interview questions and filters by the criteria you select: company, role type, seniority level, and year. It then counts how many questions are tagged Easy, Medium, or Hard, and returns that as a percentage breakdown along with bar charts rendered server-side.
The difficulty ratings come from the original reporting candidates, not from a model or algorithm. A "Medium" tag means the candidate who wrote it thought it was medium-difficulty in the context of their OA. This gives you a human-calibrated signal rather than an LeetCode-style static classification.
What the estimated prep time means
We apply a conservative heuristic: Easy problems take roughly 15 minutes each to solve cold, Medium take 30 minutes, Hard take 45 minutes. These are solve-from-scratch times, not recognition times. If you've already seen most of the common patterns, your actual time will be lower. If you're early in your prep, these estimates are appropriate for planning.
The estimated time is useful for structuring a prep schedule. If the predictor shows 12 hours of medium questions for Google SWE, that tells you to budget at least two to three full prep days just for the problem-solving component — before accounting for system design, behavioral, and review.
Frequently asked questions
How accurate is the difficulty predictor?
Accuracy depends on sample size. For top companies like Google, Amazon, and Meta we have hundreds of OA data points, making predictions quite reliable. For smaller or newer companies with fewer reports, treat the output as a directional signal rather than a precise forecast. Each result shows the underlying question count so you can judge confidence.
Does this cover 2026 OA rounds?
Yes. The database is updated daily with new candidate reports. Use the year filter to narrow results to 2025 or 2026 OAs specifically.
What companies have the most data?
Google, Amazon, Meta, Microsoft, and Apple consistently have the largest sample sizes, followed by Bloomberg, Uber, Stripe, Salesforce, and Bytedance/TikTok.