Team analytics and composition optimisation

Averaging trait scores across a team throws away exactly the information you need. Two teams with identical mean Extraversion behave completely differently if one is uniformly moderate and the other is split between four dominant voices and four silent ones. Variance is the signal.

TalentSpark models distribution, not central tendency: where the team is uniform, where it is polarised, where a single trait dominates the group's decision-making, and where a critical trait is absent from everyone.

Each team produces a fingerprint across work-style dimensions — how it makes decisions, tolerates conflict, handles ambiguity and distributes influence. Two teams in the same company routinely have fingerprints further apart than two companies in the same industry, which is usually where the cross-functional friction is coming from.

This is descriptive rather than prescriptive. There is no correct fingerprint. There is a fingerprint that matches the work the team has been given, and a lot of teams have been given work their composition cannot support.

The team optimiser takes a candidate's profile and models what the team looks like with them in it. This is the question hiring managers actually have and cannot currently answer: not 'is this person good' but 'is this person good here, with these people, on this work'.

The same machinery runs for internal transfers, reorganisations and project staffing. Modelling a reorg against trait composition surfaces the teams you are about to break, which is cheaper to know beforehand.

An individual's derailer — the behaviour that emerges under pressure and undoes them — is well understood at the person level. Less appreciated is that derailers interact. A pressured team with two high-narcissism members does not have two problems; it has a coalition problem, and the failure arrives faster than either individual profile predicts.