Why do you have archetypes at all if Big Five is continuous?

Because a continuous five-dimensional vector is unreadable to a sixteen-year-old who took the test for fun. Archetypes are a communication layer — a memorable named handle that a user can hold in their head, share with a friend, and return to weeks later. Underneath the handle the continuous scores are preserved, accessible, and used for everything that matters: career matching, premium content generation, longitudinal tracking. The archetype is the cover; the trait profile is the book.

Are archetypes valid in the way Big Five is valid?

No, and we are explicit about that. Big Five has decades of validity evidence as a continuous structure. The moment you collapse continuous traits into discrete types you sacrifice information. Pittenger (2005) made this point sharply about MBTI, and the criticism applies to any type system derived from continuous data. We use the archetype as a presentation device, not as a measurement claim. The continuous scores remain the source of truth for any actual decision.

How are archetypes derived from the underlying assessments?

Each archetype is defined by a region in the joint space of Big Five and RIASEC scores plus, where applicable, cognitive composite and skill profile features. A user's assessed profile is mapped to the nearest archetype centroid using cosine similarity with the same disclosed-weights approach used in the Career Fit Score. We do not use unsupervised clustering on user data — the archetype regions are defined a priori, in consultation with vocational psychologists, so that the system is stable and interpretable rather than drifting with the user base.

What if I do not see myself in my archetype?

Two things to check, in order. First, look at the channel-level scores under the archetype. If your continuous Big Five and RIASEC scores are close to the boundary between two archetypes, the assigned one is essentially arbitrary; the boundary cases are real cases. Second, the archetype is a population summary — it captures common patterns, not the full shape of any individual. If you disagree with your archetype but agree with the underlying continuous scores, trust the scores; the archetype label is a communication device, not a judgment.

Do you use archetypes for career matching?

No. Career matching uses the continuous trait and interest scores via the weighted-similarity Career Fit Score. The archetype is downstream of those scores, not upstream of them. Two users with the same archetype can still get different top careers because the continuous scores under the archetype differ.

Can the archetype change over time?

Yes, in two senses. First, the Big Five itself shows modest but real change across the lifespan, especially through young adulthood (Roberts et al., 2007). A user whose underlying traits shift will see their archetype shift accordingly. Second, day-to-day measurement noise can flip users near an archetype boundary. We label the latter explicitly: the result page shows the second-nearest archetype and the boundary distance, so a user near a boundary sees the ambiguity rather than a misleading hard label.

JobCannon Archetypes

A communication layer. Continuous trait and interest scores stay the source of truth. Archetypes give a user a memorable handle for what those scores collectively mean. We are explicit about why the layer exists, what it costs, and where it does not apply.

Why have archetypes at all

The most rigorous representation of a Big Five profile is a continuous five-dimensional vector with percentile bands and confidence intervals. It is also, for most users, unreadable. A teenager who just spent twelve minutes answering forced-choice questions does not want a chart of latent dimensions. They want to know what kind of person the test thinks they are, in a phrase they can share with a friend.

We resolve that tension by giving the underlying continuous representation the structural role and the archetype the presentational role. The trait profile is what drives career matching, longitudinal tracking, and premium content generation. The archetype is the cover — the named handle a user actually carries away from the experience.

The cost of this design is real. Collapsing five continuous dimensions to one discrete label destroys information. We accept that cost in exchange for a memorable result, and we preserve the underlying information so that users (and counsellors, and coordinators) who want the precision can always reach it.

How an archetype is derived

Each archetype is defined as a region in a joint trait–interest–skills space, with a centroid vector and a similarity radius. The regions are defined a priori, by vocational psychologists working with the canonical literature on person–environment fit, before any user data is seen. This is a deliberate choice: we do not run unsupervised clustering on assessed profiles to discoverarchetypes, because that procedure produces labels that drift as the user base changes and that are hard to interpret without further work.

When a user completes the underlying assessments, their continuous profile vector is computed and the cosine similarity against each archetype centroid is calculated. The nearest archetype is assigned, with the second-nearest and boundary distance preserved for transparency. If the user's profile is closer than a threshold to two archetypes, the result page surfaces that ambiguity explicitly rather than committing to a misleadingly confident label.

What the archetype is and is not

It isa memorable summary of how a user's assessed trait and interest scores sit together. It is a vocabulary that lets a user, a counsellor, or a coach discuss the profile without naming all five Big Five dimensions every time. It is a tag that can be used to organize premium content, to surface relevant career narratives, and to retrieve the user's own past result.

It is not a measurement claim. We make no scientific claim that the archetype label is a valid construct in the way Big Five is. The underlying continuous scores carry the measurement burden. We make no claim that two users with the same archetype will succeed in the same careers — career matching uses the continuous scores, not the archetype.

The relationship to MBTI

Our archetype system is conceptually adjacent to MBTI but mechanically different. MBTI is a discrete sixteen-type system derived from four dichotomies (McCrae & Costa, 1989, showed those dichotomies map roughly onto four of the Big Five dimensions, leaving Neuroticism out). Our archetypes are derived from the full five Big Five dimensions plus RIASEC and downstream features, are not constrained to dichotomies, and are always shown alongside the continuous scores that produced them.

The single most important practical difference: we never use the archetype label as an input to another decision. MBTI's common failure mode in workplace use is to feed the four-letter code into hiring, team-staffing, or coaching decisions as if it were the trait data — and Pittenger's (2005) critique of MBTI rests heavily on this misuse. By construction, our archetype cannot enter that pipeline: the trait scores enter it instead.

When the archetype changes

Two situations move a user across an archetype boundary. The first is genuine change in the underlying traits, especially across young adulthood when Conscientiousness and Agreeableness typically rise (Roberts et al., 2007). The second is measurement noise near a boundary, where a small score shift on retest flips the label. We disclose both. The result page shows the second- nearest archetype and the boundary distance, so a near-boundary user sees the ambiguity rather than a misleadingly stable label.

Honest limitations

The biggest limitation is fundamental to any typology over continuous data — and we say it plainly. Collapsing five continuous trait dimensions into one categorical label sacrifices information that cannot be recovered from the label alone. Hogan & Roberts (1996) framed this as the fidelity– bandwidth trade-off: broader summaries communicate more easily but predict specific outcomes less accurately than narrow trait scores. Archetypes prioritize bandwidth.

The second limitation is calibration on a single population. Archetype centroids are derived from literature predominantly published on Western-educated samples. Cross-cultural users may find their continuous trait scores accurate but their archetype label misleading because the centroid was set against a different reference population. We are improving on this with localized archetype definitions per market, starting with our first non-English deployments.

Citations

McCrae, R. R., & Costa, P. T. (1989). Reinterpreting the Myers-Briggs Type Indicator from the perspective of the five-factor model of personality. Journal of Personality, 57(1), 17–40. link
Pittenger, D. J. (2005). Cautionary comments regarding the Myers-Briggs Type Indicator. Consulting Psychology Journal: Practice and Research, 57(3), 210–221. link
Hogan, J., & Roberts, B. W. (1996). Issues and non-issues in the fidelity–bandwidth trade-off. Journal of Organizational Behavior, 17(6), 627–637. link
Roberts, B. W., Kuncel, N. R., Shiner, R., Caspi, A., & Goldberg, L. R. (2007). The power of personality. Perspectives on Psychological Science, 2(4), 313–345. link

Because a continuous five-dimensional vector is unreadable to a sixteen-year-old who took the test for fun. Archetypes are a communication layer — a memorable named handle that a user can hold in their head, share with a friend, and return to weeks later. Underneath the handle the continuous scores are preserved, accessible, and used for everything that matters: career matching, premium content generation, longitudinal tracking. The archetype is the cover; the trait profile is the book.
No, and we are explicit about that. Big Five has decades of validity evidence as a continuous structure. The moment you collapse continuous traits into discrete types you sacrifice information. Pittenger (2005) made this point sharply about MBTI, and the criticism applies to any type system derived from continuous data. We use the archetype as a presentation device, not as a measurement claim. The continuous scores remain the source of truth for any actual decision.
Each archetype is defined by a region in the joint space of Big Five and RIASEC scores plus, where applicable, cognitive composite and skill profile features. A user's assessed profile is mapped to the nearest archetype centroid using cosine similarity with the same disclosed-weights approach used in the Career Fit Score. We do not use unsupervised clustering on user data — the archetype regions are defined a priori, in consultation with vocational psychologists, so that the system is stable and interpretable rather than drifting with the user base.
Two things to check, in order. First, look at the channel-level scores under the archetype. If your continuous Big Five and RIASEC scores are close to the boundary between two archetypes, the assigned one is essentially arbitrary; the boundary cases are real cases. Second, the archetype is a population summary — it captures common patterns, not the full shape of any individual. If you disagree with your archetype but agree with the underlying continuous scores, trust the scores; the archetype label is a communication device, not a judgment.
No. Career matching uses the continuous trait and interest scores via the weighted-similarity Career Fit Score. The archetype is downstream of those scores, not upstream of them. Two users with the same archetype can still get different top careers because the continuous scores under the archetype differ.
Conceptually adjacent, mechanically different. MBTI is a discrete sixteen-type system derived from four dichotomies that map roughly onto four Big Five dimensions (McCrae & Costa, 1989). Our archetypes are derived from the full five Big Five dimensions plus RIASEC and downstream features, are not constrained to dichotomies, and are presented alongside the continuous scores from which they are derived. The biggest practical difference is that we never use the archetype label as a decision input — only as a presentation device.
Yes, in two senses. First, the Big Five itself shows modest but real change across the lifespan, especially through young adulthood (Roberts et al., 2007). A user whose underlying traits shift will see their archetype shift accordingly. Second, day-to-day measurement noise can flip users near an archetype boundary. We label the latter explicitly: the result page shows the second-nearest archetype and the boundary distance, so a user near a boundary sees the ambiguity rather than a misleading hard label.

Related methodology pages

JobCannon Archetypes

Why have archetypes at all

How an archetype is derived

What the archetype is and is not

The relationship to MBTI

When the archetype changes

Honest limitations

Citations

McCrae, R. R., & Costa, P. T. (1989). Reinterpreting the Myers-Briggs Type Indicator from the perspective of the five-factor model of personality. Journal of Personality, 57(1), 17–40. link
Pittenger, D. J. (2005). Cautionary comments regarding the Myers-Briggs Type Indicator. Consulting Psychology Journal: Practice and Research, 57(3), 210–221. link
Hogan, J., & Roberts, B. W. (1996). Issues and non-issues in the fidelity–bandwidth trade-off. Journal of Organizational Behavior, 17(6), 627–637. link
Roberts, B. W., Kuncel, N. R., Shiner, R., Caspi, A., & Goldberg, L. R. (2007). The power of personality. Perspectives on Psychological Science, 2(4), 313–345. link

Because a continuous five-dimensional vector is unreadable to a sixteen-year-old who took the test for fun. Archetypes are a communication layer — a memorable named handle that a user can hold in their head, share with a friend, and return to weeks later. Underneath the handle the continuous scores are preserved, accessible, and used for everything that matters: career matching, premium content generation, longitudinal tracking. The archetype is the cover; the trait profile is the book.
No, and we are explicit about that. Big Five has decades of validity evidence as a continuous structure. The moment you collapse continuous traits into discrete types you sacrifice information. Pittenger (2005) made this point sharply about MBTI, and the criticism applies to any type system derived from continuous data. We use the archetype as a presentation device, not as a measurement claim. The continuous scores remain the source of truth for any actual decision.
Each archetype is defined by a region in the joint space of Big Five and RIASEC scores plus, where applicable, cognitive composite and skill profile features. A user's assessed profile is mapped to the nearest archetype centroid using cosine similarity with the same disclosed-weights approach used in the Career Fit Score. We do not use unsupervised clustering on user data — the archetype regions are defined a priori, in consultation with vocational psychologists, so that the system is stable and interpretable rather than drifting with the user base.
Two things to check, in order. First, look at the channel-level scores under the archetype. If your continuous Big Five and RIASEC scores are close to the boundary between two archetypes, the assigned one is essentially arbitrary; the boundary cases are real cases. Second, the archetype is a population summary — it captures common patterns, not the full shape of any individual. If you disagree with your archetype but agree with the underlying continuous scores, trust the scores; the archetype label is a communication device, not a judgment.
No. Career matching uses the continuous trait and interest scores via the weighted-similarity Career Fit Score. The archetype is downstream of those scores, not upstream of them. Two users with the same archetype can still get different top careers because the continuous scores under the archetype differ.
Conceptually adjacent, mechanically different. MBTI is a discrete sixteen-type system derived from four dichotomies that map roughly onto four Big Five dimensions (McCrae & Costa, 1989). Our archetypes are derived from the full five Big Five dimensions plus RIASEC and downstream features, are not constrained to dichotomies, and are presented alongside the continuous scores from which they are derived. The biggest practical difference is that we never use the archetype label as a decision input — only as a presentation device.
Yes, in two senses. First, the Big Five itself shows modest but real change across the lifespan, especially through young adulthood (Roberts et al., 2007). A user whose underlying traits shift will see their archetype shift accordingly. Second, day-to-day measurement noise can flip users near an archetype boundary. We label the latter explicitly: the result page shows the second-nearest archetype and the boundary distance, so a user near a boundary sees the ambiguity rather than a misleading hard label.