How the interview readiness diagnostic adapts and scores
This page explains how the free interview readiness diagnostic works: 55 seeded questions across eight QA categories, ten served per session, difficulty adapting after every answer, and a difficulty-weighted score. It publishes no results yet. As of 2026-09-13 the diagnostic has 38 sessions and 24 completed reports, too few to support claims about candidates.
How big is the sample, and why are there no results yet?
As of 2026-09-13: 38 sessions started, 273 answers recorded, 24 reports completed. That is enough to check that the mechanism works and nothing like enough to say which categories QA candidates are weakest in. A page that published category averages from 24 reports would be presenting noise as a finding, so this page publishes the method and will add results only when the sample supports them.
For what the larger dataset does show, see where QA candidates lose points, which is built on full mock interviews rather than this diagnostic.
What does the question bank cover?
The bank holds 55 seeded questions across the eight categories the rest of the product uses, each tagged with a difficulty from 1 to 5. Every session serves 10 questions.
| Category | Questions | Difficulty levels |
|---|---|---|
| Behavioral QA | Questions5 | Difficulty levels1, 2, 3, 4, 5 |
| Test Automation | Questions10 | Difficulty levels1, 2, 3, 4, 5 |
| Manual Testing | Questions10 | Difficulty levels1, 2, 3, 4, 5 |
| API Testing | Questions8 | Difficulty levels1, 2, 3, 4, 5 |
| CI/CD & DevOps | Questions6 | Difficulty levels1, 2, 3, 4, 5 |
| General QA Theory | Questions6 | Difficulty levels1, 2, 3, 4, 5 |
| Performance Testing | Questions5 | Difficulty levels1, 3, 4, 5 |
| Live Coding | Questions5 | Difficulty levels1, 2, 3, 4, 5 |
How does the diagnostic adapt?
The first 8 questions walk the categories in a fixed order, one each, so every session covers all eight. The remaining 2 slots go back to categories the candidate has already got wrong, oldest miss first, so the weakest-category call at the end is backed by two data points rather than one. With no misses, the extra slots go to the least-covered categories.
Difficulty starts at 3. A correct answer steps it up by one and an incorrect answer steps it down by one, clamped to the 1 to 5 range. Selection and scoring are pure arithmetic over the question bank; no model call sits anywhere on the path from serving a question to producing a score.
How is the score calculated?
The score is difficulty-weighted: each question is worth its difficulty, so a correct level-5 answer is worth 5 times a correct level-1 answer. The overall score is weight earned over weight offered, times 100, rounded to a whole number. Category scores use the same formula over that category's questions, and the weakest category is the lowest-scoring one that was actually served.
| Difficulty | Weight |
|---|---|
| Level 1 | Weight1 |
| Level 2 | Weight2 |
| Level 3 | Weight3 |
| Level 4 | Weight4 |
| Level 5 | Weight5 |
| Band | Label |
|---|---|
| 81 to 100 | LabelInterview ready |
| 61 to 80 | LabelNearly there |
| 41 to 60 | LabelGaps to close |
| 0 to 40 | LabelEarly days |
What are the known limits of this method?
Ten questions is a screen, not an assessment. A single wrong answer in a category changes that category's score sharply, which is why the extra slots revisit misses and why the report names one weakest category rather than ranking all eight. The weighting is linear on purpose so it can be explained in a sentence; it is not fitted to any outcome data.
Results will be published on this page when the completed-report count supports a per-category claim, with the same exclusions and the same snapshot method as the mock interview dataset. The diagnostic is free and needs no account: take it here.
Where does this data come from?
What should you read next?
See where you would lose points
Two questions, scored on the same four dimensions this page reports.
Join 500+ QA engineers already practicing with AssertHired.
A test passes locally but fails in CI about one run in five. Walk me through what you check first, and why.
Scored on the same four dimensions as the real thing: Technical accuracy · Coverage · Clarity · Best practices.
Rather skip ahead? Create a free account