When Skills Move, What Stays Behind?

What happens to high performance, so closely linked to psychological safety, when a skills marketplace makes teams fluid.

Most skills marketplace platforms are built, at their simplest, to answer one question: who can do this? They hold a skills taxonomy, a proficiency rating, sometimes a credential or assessment score, and match that inventory against a project's requirements.

This is a real improvement over guesswork and manager memory. But it can still be an imprecise representation of capability, because it treats movement as a relatively clean transfer of a portable asset from one context to another.

Capability isn't always portable.

Boris Groysberg, Linda-Eling Lee and Ashish Nanda tracked star Wall Street equity analysts over nine years and found that those who left for a new bank suffered an immediate decline in performance that persisted for at least five years. The decline wasn't universal, though. Analysts who moved as part of an intact team of colleagues, rather than alone, showed no significant decline in either the short or long term.

The important point isn't simply that people work better with people they know. It is that the performance associated with an individual capability can depend partly on the configuration in which that capability is exercised. The analyst's skills hadn't disappeared. Something about the context in which those skills had been effective had changed.

This gives capability intelligence something concrete to work with that a skills-centric view does not: the relationships and context through which capability is exercised.

A skills taxonomy represents what a person can do. It does not, by itself, represent the established working relationships that may contribute to how effectively they do it. Two people with identical proficiency scores are therefore not necessarily equivalent inputs to a resourcing decision if one has established relationships that contribute to performance and the other does not.

A related finding, from the team's side rather than the individual mover's, reinforces the point. Robert Huckman, Bradley Staats and David Upton, studying an Indian software services firm, found that team familiarity — the extent to which members had previously worked with one another — was associated with project performance more strongly than conventional measures such as individual years of experience.

The two studies describe related phenomena from opposite directions. The analyst study shows what can happen when an individual capability is separated from an established working configuration. The software study shows that the accumulated familiarity between team members contributes to how effectively the team performs.

A skills-only view of capability cannot see either very well.

But there is a complication.

Manuel Sosa and Franck Marle found that, for creative and non-routine work, familiarity alone isn't what predicts good outcomes. What matters is whether previous interactions between people have actually generated good ideas together, rather than simply whether they have worked together before.

That matters because the obvious response to the problem of fluid teams would be to optimise for familiarity: keep people together who have worked well together before.

But that would recreate the rigidity that skills marketplaces were built to dissolve.

The objective isn't maximum mobility or maximum stability. It is understanding the capability consequences of changing the configuration.

This is where the distinction between skills intelligence and capability intelligence becomes useful.

Skills intelligence asks:

Who can do this?

Capability intelligence asks a broader question:

What configuration gives us the capacity to do this well?

That configuration may include individual skills and experience, but it can also include team familiarity, established working relationships, organisational context, technology, processes and other conditions through which those skills become productive.

And there is a risk in seeing more.

Once relational data exists, there is a temptation to turn it into a score attached to individuals: a flight-risk index, a belonging-risk index, a mobility-cost index. Gartner analyst Helen Poitevin has highlighted the danger of this kind of predictive labelling, including the fact that once a manager has been given a prediction about an employee, that information can influence subsequent decisions whether or not the prediction deserved the weight it receives.

That distinction matters here.

Knowing the configuration is useful. Labelling the person is something else.

Capability intelligence should help an organisation understand the consequences of changing a configuration without turning those consequences into a judgement about the individual members of it.

It might show that moving one person from a team changes several established working relationships. It might show that moving three people together preserves some of the capability that would otherwise be disrupted. It might show that a particular relationship has contributed repeatedly to successful creative outcomes.

What it cannot determine is whether preserving that configuration is worth the loss of flexibility, or whether moving three people together is worth the additional coordination that entails.

That remains a human judgement.

And that is the point.

Capability intelligence doesn't make the judgement for you. It makes more of the consequences of the judgement visible.

A skill is an attribute of a person.

Capability can be a property of a configuration.

That distinction becomes increasingly important as organisations become more fluid — and as they try to understand not just what their workforce can do, but what the organisation can actually execute.

Sources

Groysberg, Boris, Linda-Eling Lee, and Ashish Nanda. ‘Can They Take It With Them? The Portability of Star Knowledge Workers’ Performance.’ Management Science, vol. 54, no. 7, July 2008, pp. 1213–1230.

Huckman, Robert S., Bradley R. Staats, and David M. Upton. ‘Team Familiarity, Role Experience, and Performance: Evidence from Indian Software Services.’ Management Science, vol. 55, no. 1, January 2009, pp. 85–100.

Sosa, Manuel E., and Franck Marle. ‘Assembling Creative Teams in New Product Development Using Creative Team Familiarity.’ Journal of Mechanical Design, vol. 135, no. 8, 2013.

Zielinski, Dave. ‘The Dangers of Using Predictive Analytics to Gauge Employee Flight Risk.’ SHRM, 12 March 2020.