A baseline for testing every CV rejection criterion

VCO Group proposes a simple test for any CV rejection criterion: compare it against rejecting at random. In data covering more than 21,000 Young Talent Program applications in Vietnam, rejecting 100 applications at random costs you close to a quarter of the candidates who would have passed the test round. A criterion is only worth keeping if it costs less than that baseline.

Why most rejection criteria have never been tested

A programme's list of CV rejection criteria usually forms very organically. Someone notices a signal, the signal seems reasonable, it goes into the scoring guidance, and it stays there for seasons without anyone questioning it.

The reason is not a lack of rigour but a lack of data. To know whether a rejection criterion is sound, you need to know what the candidates it rejected would have done had they progressed. Because they were rejected, nobody knows.

The only way to answer is to have a dataset in which candidates sit an aptitude test before or alongside the CV round, so every application carries an independent measurement. VCO Group operates both systems for a number of Young Talent Programmes in Vietnam, and therefore has that dataset.

The next requirement is a comparison point. Without one, every figure about good candidates lost sounds alarming, and there is no way to distinguish a weak criterion from a strong one.

The baseline: what random rejection costs

The most sensible comparison point is rejecting applications entirely at random. This is the worst acceptable level, because any criterion carrying information must beat a coin toss.

In the dataset, rejecting 100 applications at random during a Young Talent season costs a programme, on average, close to a quarter of the candidates who would have passed the aptitude round. That is the baseline.

The test is therefore compact: take any criterion in use, count how many of the 100 applications it rejects would have passed the test round, and compare that against the baseline. Lower, and the criterion is creating value. Higher, and the criterion is doing harm and should go, because random selection would do better.

The strength of this test is that it requires no debate about whether a criterion is theoretically justified. It asks one question, and that question has a numerical answer.

Applying the test to a fairly widespread criterion

One criterion appears in many programmes' scoring guidance: penalise or reject candidates who are applying to several programmes at once, on the argument that someone spreading applications around lacks direction and is less likely to stay.

The argument sounds reasonable, and it reflects a real anxiety: investing effort in a candidate through the whole process only to have the offer declined at the last moment.

Test result for the criterion "reject multi-programme applicants"

The result holds when the data is split by segment, so it is not driven by one programme or one candidate group.

The mechanism is easy to understand from the candidate's side. Strong candidates know they have options elsewhere, so they apply in several places. Less confident candidates apply once and wait. In the data, multi-programme applicants are roughly twice as likely to pass the test round as single-programme applicants.

In other words, applying widely is not a signal of a lack of seriousness. It is a signal that the candidate knows their value in the labour market.

Handling the offer-acceptance concern differently

The anxiety behind that criterion is real and should not be dismissed. What needs to change is the instrument used to address it.

Rejecting multi-programme applicants at the CV round is the wrong instrument because it solves the problem by removing your strongest candidates. The right instrument is speed and process experience - competing to win that group rather than avoiding them.

Three things have direct effect: shorten the interval between rounds, especially in the first two weeks after application; publish the full process timeline up front so candidates can plan around other opportunities; and return results from each round quickly, including rejections.

Market context supports this. Southeast Asian hiring analysis for 2026 records that the strongest candidates leave the market within days rather than weeks, so a process with four interview rounds spaced two weeks apart is removing itself from the competition. Indeed's regional talent report, based on a survey of more than 4,300 professionals, devotes a section to identifying how long candidates tolerate a slow process before abandoning it.

Applying the test across the whole scoring sheet

This test is not limited to one criterion. It is a review process that can be run across the entire scoring sheet after each season.

The most compact method is to take the most recent season's data, provided at least one group of candidates sat the test before being rejected at the CV round. For each criterion in the sheet, count the candidates who met the test threshold among those the criterion rejected, express it per 100 applications, and rank every criterion from best to worst against the baseline.

The result usually falls into three groups. Criteria clearly better than baseline: keep, and consider increasing their weight. Criteria around baseline: contributing nothing but doing no harm - keep at low weight or demote to reference information. Criteria worse than baseline: remove.

Methodologically this is exactly what selection system research recommends. In their 2023 paper, Sackett and colleagues stress that the weighting between components in a selection system matters more than the presence of any single component, and that variation between contexts is large enough that validity research on an organisation's own data retains distinct value - it cannot be replaced by an average figure from international literature.

Three things to do

1. Establish your own baseline by calculating how many threshold-meeting candidates are lost when 100 applications are rejected at random, then use that figure as the reference for every criterion.
2. Review the entire rejection criteria list against this test after each season, and drop any criterion that performs worse than baseline.
3. Move your response to offer-acceptance risk away from rejecting multi-programme applicants and towards shortening the interval between rounds and publishing the full process timeline up front.

Frequently asked questions

Should candidates applying to several programmes at once be rejected?

No. In VCO Group data this criterion costs more than a third of good candidates per 100 rejections - worse than random rejection, which costs close to a quarter. The reason is that multi-programme applicants are roughly twice as likely to pass the test round as single-programme applicants.

How do you know whether a CV screening criterion works?

Compare it against the baseline of random rejection. Count how many of the 100 applications the criterion rejects would have passed the aptitude round, and compare against the equivalent figure for random rejection. Any criterion that costs more than baseline should go.

How do you reduce the risk of candidates declining an offer late?

Through speed, not by rejecting candidates who have options. Three things have direct effect: shorten the gaps between rounds in the first two weeks, publish the full process timeline up front, and return each round's result quickly, including rejections.

About VCO Group

VCO Group operates the assessment infrastructure behind Young Talent Programmes and Management Trainee schemes in Vietnam, covering application intake, aptitude testing, automated interviewing and Assessment Centres. VCO Group serves young talent and candidates with under three years of experience, working with large enterprises and multinationals in Vietnam. Its assessment and candidate management platform is called Testalents. Head office in Ho Chi Minh City.

If you would like to benchmark your own selection data against the market, or discuss the Young Talent Program Insight Report, contact Mr. Dung on 0334 889 068 or visit vcogroup.com.vn.