Nobody learns the operation by reading the manual

A Brazilian grocery retail chain, 600 people, 14 stores. They hire a regional manager coming from another chain. In her first week she gets a 60-page manual, three system logins, and a calendar full of introductions. By week five, she's still asking why store 7 gets restocked on a different day than the others. Nobody has a good answer: it's been that way since a carrier negotiation four years ago.

It will take her about three months to operate with autonomy — not for lack of competence, but because everything that matters in that operation is outside the manual. It's in the history of decisions, in the exceptions agreed along the way, in what each store learned to do differently.

01The idea: you learn by taking part in the practice #

Jean Lave and Etienne Wenger published Situated Learning in 1991, drawing on studies of traditional forms of apprenticeship — tailors, midwives, butchers, machine operators. The central concept is legitimate peripheral participation: newcomers start with real, lower-risk tasks inside the same community that carries out the main work, and move toward the center as they gain mastery.

Two words carry the weight. Peripheral, because the initial task has lower consequence — it isn't a simulation. Legitimate, because it's real work the operation actually needs done. The apprentice isn't watching: they're contributing. Wenger developed this further in Communities of Practice (1998), describing how groups that share a practice build a common repertoire, and brought the organizational reading to the Harvard Business Review in 2000 with William Snyder.

The implication is direct: the manual is not the vehicle of learning, and never was. It works as a reference for someone who already understands the context. For a newcomer, 60 pages describe what the company would like to be — not what it does on a Tuesday when the truck runs late.

It's the same boundary Nonaka described between tacit and explicit knowledge, covered in what only Paulo knows. The difference in emphasis is useful: Nonaka looks at how the tacit becomes explicit; Lave and Wenger look at how someone enters the practice and starts sharing the tacit without it ever needing to be written down.

02Why the ramp-up takes three months #

Because typical onboarding reverses the order. It front-loads information — presentations, policies, systems — and postpones real participation, on the grounds that the person isn't ready yet. The result is that little of what they hear sticks, because there's nowhere to attach it, and they reach their first meaningful task without having built any repertoire.

The second reason is the unrecorded exception. Store 7 with a different restocking day is the kind of thing that exists by the hundreds in any operation. Each of those exceptions, when unwritten, has to be discovered by collision: the new person proposes something, someone says that's not how it works here, and they learn. Multiplied by a hundred, that's the ramp.

And there's a third, quieter reason: whoever answers the newcomer's questions is always the same person, the one who knows the most — and that person is busy. The ramp then moves at the speed of a third party's calendar.

03The math on a 90-day ramp #

At the scale of the 600-person chain — a mid-size Brazilian company (figures in Brazilian reais). Stated assumptions, to redo with your own numbers:

Want to see how this looks inside a real operation? Explore the platform.

The math: 90 days is about 3 months; with 55% of capacity missing, each hire loses the equivalent of 1.65 months of full-time work. With salary plus payroll costs of R$ 21 thousand a month, that's R$ 34.6 thousand per person; × 18 positions, roughly R$ 620 thousand a year. Add to that the hours spent by whoever does the teaching — in the example, 25 hours per ramp from each area's specialist.

Cutting the ramp from 90 to 45 days recovers half of that, close to R$ 310 thousand a year at this scale. It's not a hypothetical long-term gain: it shows up by the third hire.

04Where this breaks down in practice #

1. Onboarding is made of presentations. Two weeks listening to departments explain what they do, with no real task in between.
2. The first assignment is too big. It jumps from observation to full responsibility, skipping a low-risk peripheral stage.
3. Exceptions live in the head of whoever's been there longest. Each one gets discovered through a mistake, and every mistake costs the newcomer credibility.
4. There's no searchable history. The person can't see what was decided before, so they re-propose what's already been ruled out.
5. The mentor has no protected time. The responsibility to follow up gets assigned, the time doesn't — and the ramp moves at the pace of gaps between the mentor's meetings.
6. Nobody measures the ramp. Without knowing how long it takes today, there's no way to know if a change worked.

05What data-driven management answers #

Much of what a newcomer needs to know isn't content: it's state. How this process is moving, who owns each stage, what was decided in the last six months and why, which exception applies to which unit. When that's recorded where the work happens, the new person participates from week one without depending on someone translating the operation for them.

That's what turns peripheral participation into something practical in a company with no time for formal mentoring. It's the same mechanism described in knowledge that doesn't circulate isn't an asset — with the replacement cost detailed in first who, then what.

At Relevanti, the people, process, and operations modules keep that context available — see the platform, the breakdown by area in solutions, or talk to us at contact.

If the person you hired takes three months to operate on their own, the problem isn't them. It's that your operation only exists in full in the memory of whoever's already there.