Blog 17 October 2026 · 5 min
Scored emotional intelligence in 2026: the filter that decides for you
Your application can be rejected because of a score you never saw being calculated. In 2026, emotional intelligence is no longer a vague impression left on a human recruiter: it is formalised, tested, and sometimes inferred by an algorithm from your video or your answers to a questionnaire. According to stubby-ai.com, ignoring this scoring of soft skills means letting a filter decide for you, without giving you the means to influence it. In this article, you will learn which dimensions are really being tracked, why empty adjectives and "perfect" answers work against you, and how to turn each emotional intelligence trait into observable evidence aligned with the job advert.
A score you never saw being calculated
Imagine this: your application is rejected because of a scored emotional intelligence rating that nobody ever showed you. In 2026, this scenario is no longer science fiction.
Recruiters are increasingly formalising how they assess soft skills. Personality tests, role-play exercises, video interviews analysed by algorithms: all of this already exists.
The CNIL (France's data protection authority) confirms it in its recruitment guide: tools can measure stress resistance, autonomy or ease of working in a team. And a screening tool can reject a candidate automatically.
Let's be honest: I have no proof that a standard ATS calculates an EI score for every CV. But the mechanics of scoring soft skills are well documented. Let's look at how to avoid being at their mercy.
Scored EI in practice: four dimensions being tracked
Emotional intelligence is often broken down into four families: self-awareness, emotional regulation, empathy and relationship management.
A word of caution: this breakdown is a practical framework, not a universal standard. Tests vary, and each provider has its own model.
The study by Walker & MacCann published in Assessment, for example, uses different axes: emotionality, sociability, self-control and well-being.
The key point: these dimensions are "searched for" in your CV, your cover letter and your tests. But a machine does not read an emotion. It reads words and answers.
Trap no. 1: the empty adjective that proves nothing
"Empathetic, caring, a good listener."
These words are everywhere. And that is precisely their problem: they are hard to use and difficult to interpret automatically.
An adjective says nothing verifiable. It contains no context, no action, no result. For an algorithm, as for a busy human, it is empty.
By contrast, a detailed example provides observable elements. That is the whole subject of our article on turning soft skills into measurable evidence, which sets out the general method.
Here, we zoom in: EI is the hardest to prove, because it can never be reduced to a keyword.
Trap no. 2: self-report tests and social desirability
More and more recruiters ask you to fill in an EI questionnaire. And here, the natural reflex is disastrous: answering what you think they expect.
The Walker & MacCann study is unequivocal. When participants were asked to "present themselves in the best light", their scores shot up (a faking effect of d = 1.47 in self-report).
In other words: anyone can inflate their answers. But this artificial inflation mainly measures your ability to manage your image, not your real EI.
The counter-intuitive advice: don't aim for perfection. A 100% positive profile is less credible than a consistent, nuanced one. Own up to a weakness, and you gain credibility.
This ties in with what we explain about the "perfect" answers that get you rejected in AI interviews: too polished is suspicious.
The paradox: a human skill judged by a machine
Here is the real unsettling part of 2026. EI is deeply relational, and yet it is increasingly inferred by algorithms that analyse audio or video.
The study by Ingber & Andalibi presented at the ACM FAccT 2025 conference tested 456 participants in a simulated video interview.
The result: those assessed by an "emotion AI" found the process more unfair, and artificially adapted their behaviour to what they imagined was expected.
This is the artificial over-performance bias. The system ends up measuring how comfortable you are in front of the camera, not your empathy. The researchers also point out that the scientific validity of emotion inference remains highly contested.
The good news: the CNIL requires transparency. You must be informed of the methods used, and a score must remain relevant, proportionate and open to challenge. You have rights.
The method: turn each EI trait into a micro-story
Stop describing. Tell a story. Here is my five-step method, precise and action-oriented.
- Identify the target dimension. Self-awareness, regulation, empathy, relationships: which one does the role really value?
- Choose a real situation. A conflict handled within a team, tough feedback received and acted on, a decision made under emotional pressure.
- Structure it as a micro-story. Brief context, precise action, observable result. Two sentences are often enough.
- Quantify where possible. "Defused a client conflict that threatened a €40K contract." The figure anchors the evidence.
- Align your wording with the advert. Reuse the vocabulary of the role without copying and pasting: you speak the same language as the recruiter.
One example is worth a thousand adjectives. Compare:
❌ "I am a very good listener." ✅ "After negative feedback from my manager, I reworked my reporting method; the team's feedback went from tense to constructive within a quarter."
The second version is observable, time-bound and verifiable. It is exactly what a recruiter, or an algorithm, can use.
Where a specialised AI makes the difference
Not every role expects the same EI. A manager relies on relationship management, a salesperson on client empathy, a developer on regulation under pressure.
Working out by hand which dimension to highlight for each job advert is slow and imprecise. That is where a tool like Stubby comes in.
The method offered by Stubby cross-references your CV and the job advert to identify the EI dimensions that are really expected, then suggests where to place your evidence. To be clear: this is a feature of the tool, not a fact established by the studies cited.
The idea remains simple: to help you compete on equal terms against increasingly automated processes.
Key takeaways
Scored emotional intelligence is not a fad: it is the formalisation of a judgement that used to be purely subjective. You can no longer ignore it.
Avoid empty adjectives. Resist suspicious perfection in tests. Turn each trait into a quantified micro-story aligned with the job advert.
And keep a critical mind: these tools often measure how comfortable you are with them, not your real humanity. Prove it, don't just claim it.
« "An algorithm doesn't read an emotion: it reads words, examples, results. Prove your emotional intelligence, don't just claim it." »
| Section | Key messages |
|---|---|
| The 2026 picture | Soft skills are increasingly formalised and scored; the CNIL (France's data protection authority) confirms the use of tests, scoring and video analysis, but no standard ATS systematically rates EI. |
| The dimensions tracked | Self-awareness, regulation, empathy, relationship management: a useful but not universal framework. The machine reads words, not emotions. |
| Trap no. 1 | Empty adjectives (empathetic, a good listener) are hard to use and difficult to interpret automatically. |
| Trap no. 2 | Self-report questionnaires are easy to fake (Walker & MacCann study, d=1.47); aim for consistency rather than perfection. |
| The paradox | A human skill judged by a machine creates an artificial over-performance bias (FAccT 2025 study, 456 participants). |
| The method | Turn each EI trait into a situational micro-story: dimension, real situation, structure, figure, wording aligned with the advert. |
| Stubby's role | Cross-reference CV and job advert to identify the EI dimensions really valued for the role and place the evidence in the right spot. |
