Our method
The Keelworth model.
It is a proprietary work-style system of our own, built on the most validated foundation in personality science. The model turns a series of work situations into a structured, comprehensive read on how a person tends to work, measured through behavior rather than self-rating, scored by our own engine and calibrated as more people complete it.
What it measures
Work-style tendencies, and nothing more.
Keelworth describes how a person tends to approach work. It reads thirteen traits across five areas, each as a position between two everyday descriptions, for example self-starting or needing prompting, organized or unstructured.
What it measures
- Work-style tendencies across thirteen traits
- How a person tends to handle common work situations
- Where they fall between two ends of each trait
What it does not measure
- Skill or knowledge
- Intelligence
- Character or honesty
- Mental health
- A prediction of job performance
The Keelworth model
Ours, on a Five-Factor foundation.
The Keelworth model is ours end to end. The science underneath it is the Five-Factor Model, the most validated framework in personality research, replicated across cultures and languages over decades of peer-reviewed work. We took that foundation and built everything above it ourselves, and the model is proprietary: the work situations a person responds to, the rating system that scores them, the bank of language that turns a score into a description, and the analysis that assembles the report are all Keelworth's own.
The model is built for hiring rather than for a clinic or a research lab. It reads thirteen work-style traits, grouped into five areas and written in the plain terms a hiring manager uses. Each trait is a position between two ends, and neither end is better than the other. A low score is a different working style, not a worse one.
Assessment results carry real weight with employers. In SHRM's 2025 research, most HR professionals rated assessment results at least as highly as education or years of experience when judging candidates. Keelworth is built to give that kind of signal in a report you can act on.
Each bar is a spectrum. A person's report places them along it and describes where they tend to sit.
The method
Behavior over self-rating.
Most personality tests, including most built on the Five-Factor Model, ask people to rate themselves on a scale: agree or disagree, very much like me or not at all. Self-ratings are easy to bend toward whatever looks best, and the traits people most want to inflate are the ones a self-rating inflates most. Keelworth works differently. It uses situational judgment: each item is a work situation we wrote for the model, and the person chooses what they would most likely do.
Every option carries a concealed value, and the most flattering choice is rarely the obvious one, which makes the assessment far harder to steer than a questionnaire. The items for each trait are spread across the whole assessment rather than grouped, and the order of the options is randomized for every session, so a response cannot settle into a pattern. It targets the known weak point of self-report measurement, social desirability, and reduces it at the source. Two attention checks sit inside the assessment to confirm the person is reading.
Scoring
The same answers always produce the same report.
Our own engine turns the answers into a profile. Each trait score is the mean of the items that measure it, mapped to a descriptive band of lower, moderate, or higher. Scoring is fully deterministic, with no language model anywhere in the path, so the same answers always produce the same report, word for word. Every candidate is read by the same fixed logic, which is what makes one report fair to compare against another.
The band thresholds are calibrated on real response data and refined continuously as more people complete the assessment. Every completed assessment sharpens that calibration.
Response quality
Checks that catch a rushed or gamed response.
Every set of answers passes through several independent quality screens, and the report states plainly when a reading is less certain.
Attention checks
Two items inside the assessment name the answer to pick. A person who is reading passes them, and a failure marks the whole response as unreliable.
Response-set and straight-line detection
Because option order is randomized, a genuine profile cannot concentrate on one position. A response that does is flagged as inattentive rather than read.
Within-trait agreement
When the items for a trait disagree with each other, the reading carries lower confidence, and the report softens it accordingly.
Over-claiming screen
When a response rates almost every trait at the ceiling, the report flags it, since few real profiles are that uniformly high.
Forced-choice cross-check
A set of questions asks the person to pick what is most and least like them among options that all look good. Because not everything can be chosen, it is hard to claim every strength at once.
Favorable-responding flag
When someone rates a trait high in the situations but ranks it at the bottom in the forced-choice questions, the report raises a flag for a closer look rather than a verdict.
The science
Where the confidence comes from.
Built on validated foundations
The Keelworth model rests on foundations that are solid: the most validated model in personality science, and a measurement and scoring method built to resist gaming. It is calibrated on real responses, and every completed assessment sharpens it.
One thing no responsible instrument claims is that a score predicts how a particular person will perform in a particular job. The Keelworth model does not claim it. It tells you how someone tends to operate and leaves the hiring decision where it belongs, with you.
Keelworth describes work-style tendencies. It does not measure skill, intelligence, character, or mental health, it is not a medical examination, and it should not be the sole basis for any employment decision. You are responsible for using it lawfully alongside your own process.
Compared to other tests
How Keelworth is different from other tests
Personality-type quizzes like the MBTI sort people into a small number of boxes. You come out as one of sixteen types, and two people in the same box are treated as the same. The bigger problem for hiring is that these quizzes were not built to predict how someone works. They were made to help people think about themselves, and the type you get can change from one sitting to the next. A box is easy to remember and fun to talk about, though it flattens a person into a label and tells you little about how they actually handle the work in front of them.
DISC sits in a different place. It works mostly as a shared language for teams, a quick way for people who already work together to talk about styles and smooth out friction. That is a fair use. It just is not a hiring read. It was not designed to compare candidates or to give an employer a careful description of how someone tends to work, and using it that way asks it to do a job it was never meant for.
One-off personality quizzes, the kind you find online, have a separate weakness. Most of them ask you to rate yourself on a scale, agree or disagree, one to five. Those questions are easy to read and easy to inflate. A candidate who wants the job can see what each item is fishing for and answer the way they think you want. There is usually nothing in the test to catch that, and nothing tying the result to real work behaviour.
Keelworth reads how a person tends to work from how they handle real work situations. Rather than asking you to score yourself, it puts you in a situation and asks what you would do, with options written to look about equally reasonable so there is no obvious right answer to game. It reads thirteen tendencies across five areas as continuous bands, lower to higher, rather than sorting you into a type. It also carries honesty checks: forced-choice items where everything looks good and not everything can be picked, attention and consistency checks, and a flag when someone rates a trait high but ranks it last when forced to choose. No assessment is unfakeable, and someone set on shading their answers still can, which is one reason Keelworth is an aid and not the decision.
For a hiring input, a behaviour-based work-style read is the better fit because it describes the thing you actually care about: how this person tends to approach the work, not which label they land on. It gives you a fair, repeatable description to set next to a skills check and a structured interview. The same answers always produce the same read, two people are read the same way, and the result is meant to inform your judgment, never to make the call for you.
See it in a real report.
Read a full sample report to see how the scores become a detailed read on how a person tends to work.