Where QA candidates lose points in mock interviews

What do the four dimensions score, and which one is lowest?

Every full mock interview on AssertHired is scored on four dimensions from 0 to 100: technical accuracy, communication, concrete examples and depth. The overall score is the plain mean of the four. Across 498 reports the means are close together, but which dimension comes last in a given report is not evenly spread at all.

A report that ties on two dimensions counts under both, so the last column can sum past 100 percent.

Technical accuracy is the lowest dimension in 3 percent of reports. Concrete examples is the lowest in 46 percent and depth in 39 percent. Candidates mostly know the material. What they do not do is anchor an answer in something they actually did, or go past the first layer when the interviewer waits.

Does the pattern change by interview type?

It changes in which of the two non-technical dimensions comes last, not in the headline. Categories with at least 30 reports are shown individually; the rest are grouped, because a per-dimension claim on a dozen reports would be a guess with a table around it.

The sharpest result is in Test Automation rounds (65 reports from 44 candidates): concrete examples is the lowest dimension in 54 of them, with a mean of 47.5 against 62.5 for technical accuracy. Candidates can explain what a locator or an auto-wait is; they struggle to describe a specific suite they built and a specific failure they fixed. In Behavioral QA rounds the weakest dimension is depth, lowest in 142 of 283 reports: the story is there, but the answer stops before the trade-off, the measurement or what changed afterwards.

What does the score distribution look like?

Scores are deliberately calibrated against a senior rubric, so a 57 is not a school grade. 215 of 498 reports (43 percent) score under 60 overall. Read the bands as where a candidate sits against an interviewer who expects senior answers, not as a pass mark.

What should a candidate do with this?

Prepare examples before you prepare facts. For each tool or practice on your CV, have one specific story ready: the system, what you built or found, one number, and what you would do differently. That single habit targets the dimension that comes last in 46 percent of reports.

Then practise going one layer deeper than the question asked. When an interviewer asks how you handled flaky tests, the first layer is the fix; the second is how you measured the flake rate before and after and what you changed in the pipeline. Depth is the lowest dimension in 39 percent of reports because most answers stop at the first layer.

The rubric itself is tested in public; see how the scoring is evaluated.

How was this measured?

Snapshot taken 2026-09-13 from the production database by a read-only script whose only output is the data file this page renders. Included: every feedback report with all four dimension scores and an overall score, which only exist for completed full interviews. The no-account two-question flow and the free onboarding answer are scored separately and are not in this set.

Excluded: 16 reports from the operator's own accounts, test aliases and reserved domains, using the same rule the analytics use; 0 reports could not be matched to an interview category. A candidate is a distinct account; 247 candidates produced 498 reports, so the sample is weighted towards people who practised more than once.

The scoring rubric was recalibrated on 2026-09-04. 10 of the 498 reports were graded under the current rubric; the rest predate it and score lower on average. The dimension ordering above is about which dimension comes last within a report, which the recalibration was not designed to change, but the absolute means will move as the post-change sample grows. Aggregate only: no per-person data is published, and no report text is quoted.