Person-Problem Matching

Match the person to the problem's error budget

Eric Glyman describes a hiring discipline that happens before a role is opened: decide what shape of person a problem actually needs, because a single product contains sub-problems with opposite error-tolerances, and each rewards a different temperament. In his account, "great engineer" is underspecified; what matters is matching a person's way of working to the problem's cost of being wrong.¹

Glyman anchors the idea in Ramp's own work. Because Ramp issues credit cards, part of the org runs on rails where, in his phrasing, "it's not funny if you swipe and it doesn't work." That splits the work in two.

The two temperaments Glyman names

For the never-break rails, the money-movement stack demands very low latency and a very low tolerance for error. Glyman says he hunts specifically for people who obsess, whose working style suits perfection, and that he looks for that quality directly rather than screening for general talent.¹

For the recover-fast surfaces, the right person ships daily, even hourly, turning customer feedback around in ten minutes. Glyman says these people can be a little sloppy where detect-and-recover is cheap. His example: a receipt match that is usually half a second and occasionally ten seconds when queues are long is "not a huge deal, as long as you're detecting it and responding quickly."¹

The mistake, in his telling, is hiring one temperament for the other's problem. A perfectionist on a fast-iteration surface is slow; a fast-and-sloppy shipper on the money rails is dangerous.

Why Glyman frames it this way

The move makes the error-budget of the problem an explicit hiring input. Most job descriptions describe a stack or a seniority level; person-problem matching instead describes the failure mode the team can and cannot tolerate. Glyman treats this as the hiring-side mirror of Ramp's two-speed architecture, where money rails must not break while much of the rest of the product is allowed to break and recover many times a day. Deciding the temperament first is how that split gets staffed.

It also sharpens the more general instinct to hire for spikes: the spike being screened for is problem-specific, since perfectionism and shipping velocity are both spikes but opposite ones. Glyman names the error budget up front, before the role is opened. Related to Glyman's broader hiring stance, it sits alongside slope-over-intercept (which trajectory to buy) and mispriced-talent-pools (where the underpriced version of each temperament lives).

Tensions Glyman acknowledges

People are not purely one temperament, and problems evolve: a fast-iteration surface can harden into a must-not-break one, so the matching is a starting bet rather than a permanent label. Over-segmenting risk-tolerance also risks ossifying into a caste of "the careful people" versus "the fast people," and the healthier reading of Glyman's own examples is that people move across the split as they grow.