Score the role before you score the person

Series「People Don't Follow the Org Chart」 · Part 12 of 13 · Part 3 · Why people put in effort
Prologue：The people who aren't on the org chart

One role turns over three people in two years, each leaving for a different stated reason: no growth, too fragmented, can't see what the work is for. The company concludes its hiring bar needs raising. My conclusion: the three were saying the same thing, and the role itself fails the test.

Five dimensions for scoring a role

In 1976 Richard Hackman and Greg Oldham proposed the job characteristics model, which breaks "can this role, by itself, engage a person" into five scorable dimensions.

Skill variety: how many different abilities the role uses. Task identity: whether the person does a whole piece of work or a fragment. Task significance: how much the work affects other people. Autonomy: how much of the how and when is the person's own decision. Feedback: whether, on finishing, the work itself tells you how well it went.

The first three determine whether the work feels meaningful, the fourth determines felt responsibility, the fifth determines knowledge of results. A role low on all five loses anyone's engagement within six to twelve months, largely regardless of who they are.

The model has over forty years of empirical support. The disputes are two: whether the five dimensions carry equal weight (the original used a multiplicative formula; most later research uses simple addition), and individual differences (some people are less sensitive to the five). Neither dispute affects its use as a checklist.

What low-scoring roles look like in a Japan entity

The most common is the HQ liaison role: all day translating HQ's requests to local and local conditions back to HQ. Skill variety is high, but task identity is near zero (always relaying someone else's work), autonomy is low (neither side's decisions are theirs), feedback is low (success belongs to both sides, failure is their poor communication). I wrote about this role in the structural-hole post; it scores low because it was never given decision rights.

The second is the role cut too fine. HQ slices roles for its own thousands-strong division of labor and transplants them into a five-person entity: one person reviews only one type of contract, another only enters data into one system. Task identity and significance are both low; the person can't see how their sliver relates to the company.

The third is the role with no feedback. Japanese companies often give an annual evaluation once at year-end, and for the other eleven months people don't know how they're doing. That is a structural absence of feedback, not a boss who can't communicate.

The design decision: score before hiring, fix low roles first

Add the five dimensions to the role profile and fill them in before writing the job requirements. Score each 1 to 5. A role totaling under 15 doesn't get a hire; it gets redesigned first.

The fixes are usually concrete. Low task identity: pull one or two upstream or downstream steps into the role so the person owns something end to end, even if it's smaller. Low autonomy: write down three things they can decide without asking, the same move as the autonomy item. Low feedback: give the role one metric the person can see themselves, updated weekly, not at year-end. Low significance: put them in front of the people who receive their work, even once a quarter.

Score again after the redesign, and hire only once it clears the line. Reverse the order and you get the three-people-in-two-years story.

Back to the method

When I judge person-role fit, I check veto items first and then the weakest layer. But there's a step before that: does the role itself clear the line? If it doesn't, however finely you assess fit, you're just looking for someone who will tolerate a bad role. These five dimensions are page one of my role profile; the candidate's four-layer portrait is page two.

The role in your company with the highest turnover: what do its five dimensions add up to?

In this series
Previous：Pay removes complaints. It doesn't create commitment.
Next：When bad news stops traveling up