HireVue Numerosity Practice Data: How Scores Change Over Repeated Runs
Candidate Falcon
Editorial Team
Published 18 November 2026. Data from 2 May to 16 November 2026.
Candidate Falcon runs a timed simulation of Numerosity, one of the games candidates report from HireVue game-based assessments. This page reports what happened to scores when the same candidates played the simulation more than once. The figures come from our own practice records, not from HireVue, and they say nothing about how HireVue scores you.
Key findings
- In 313 completed 3-minute runs of Candidate Falcon's Numerosity simulation by 64 candidates, the typical (median) first run scored 18 correct answers.
- The typical second run scored 23 correct answers, and the typical third run scored 25.
- Compared with their own first run, candidates scored a median 24% higher on their second run (45 candidates; 60% of them improved).
- On their third run, candidates scored a median 34% higher than on their first (30 candidates; 77% improved).
- After the third run the typical score stayed at about 25 correct answers: among the 20 candidates who played five or more runs, the fifth run was a median 18% above the first, and 55% had improved.
- Wrong answers stayed low at every stage, at a median of one or two per run, so the extra correct answers did not come with more errors.
- Most candidates stopped early: 19 of the 64 completed one run, 15 completed two, 10 completed three or four, and 20 completed five or more.
Run by run
| Run | Candidates | Median correct answers | Median change on own first run | Improved on own first run | Median wrong answers |
|---|---|---|---|---|---|
| 1st | 64 | 18 | n/a | n/a | 1 |
| 2nd | 45 | 23 | +24% | 60% | 1 |
| 3rd | 30 | 25 | +34% | 77% | 1 |
| 4th | 25 | 25 | +38% | 68% | 1 |
| 5th | 20 | 24.5 | +18% | 55% | 1 |
| 6th | 18 | 26.5 | +19% | 72% | 2 |
Each row covers only the candidates who reached that run, so each row is a slightly different group. That is why the change column does not rise steadily: the 25 candidates who played a fourth run are not the same people as the 20 who played a fifth.
What this suggests for your preparation
Two or three full runs captured most of the improvement we measured. Use them to learn the controls and the pace, then spend extra time on the operation that causes most of your wrong answers. Our Numerosity guide explains the format, and the Numerosity practice page has the timed simulation these figures come from.
Practice makes the format familiar. It cannot tell you how an employer reads your result, and a better score in a simulation is not a prediction of your result in a live assessment.
Sample and method
- Source: Candidate Falcon's own records of its Numerosity simulation, one record per run, queried on 18 November 2026.
- Window: runs between 2 May and 16 November 2026.
- Included: runs marked completed, meaning the candidate played the 3-minute run to the end. Unfinished and abandoned runs are left out.
- Excluded: Candidate Falcon's own internal account.
- Score: correct answers in our simulation. It is not a HireVue score; HireVue does not publish how Numerosity is scored.
- Run number: the order of each candidate's completed runs.
- Change on own first run: for each candidate, the score on run n minus the score on run 1, divided by the score on run 1. The table reports the median across the candidates who reached run n. "Improved" means a higher score than that candidate's own first run.
- Why medians: a few very high or very low runs would pull an average around; the median is the middle candidate.
- Earlier version: the first version of these figures, published on 2 November 2026 on the Numerosity practice page, covered 298 runs by 59 candidates from 2 May to 23 October 2026. The new runs left the first-, second- and third-run medians unchanged.
Limitations
- Candidates choose how much to practise. People who keep practising may differ from people who stop after one run, for example in motivation or in how hard they found the first run.
- There is no comparison group. This is a record of practice, not a controlled experiment, so part of the improvement may simply be getting used to our simulation's screen and controls.
- A simulation is not the live assessment. We do not see candidates' HireVue results, and HireVue does not publish its scoring.
- Later runs have small samples. Only 20 candidates played five or more runs.
- Numerosity only. These figures do not describe other games, and we do not extrapolate them.
How to cite this page
Candidate Falcon (2026). HireVue Numerosity Practice Data: How Scores Change Over Repeated Runs. /blog/hirevue-numerosity-practice-data/ (data from 2 May to 16 November 2026).
The address of this page will not change when the figures are updated; each update will state its date and data window. For questions about the data, use the contact page.
Candidate Falcon is independent and not affiliated with HireVue.
