What teaching Japanese taught me about onboarding customers
4 min read
Updated October 27, 2026
From August 2015 to August 2016 I taught Japanese at the Centro Universitario de Idiomas, the University of Buenos Aires language center where I had studied a few years earlier. I’d been on the other side of that desk, which helps.
A decade later, most of my teaching happens in rooms that don’t look like classrooms: onboarding enterprise customers at Vercel, or coaching mahjong players and assisting in Kosho Tsuchida’s class. I plan for all of it the way I planned those lessons.
Start from what they can do at the end
A language lesson that starts with “today we’ll cover the te-form” is organized around the teacher’s syllabus. A better one starts with what the student will be able to do by the end: ask for directions, say what they did yesterday, order food without pointing. Language teachers call these can-do statements, and they make every other decision easier, because each exercise either builds toward that ability or it doesn’t.
Then comes sequencing. Hiragana before kanji, present tense before past, nothing that depends on something you haven’t taught yet. It sounds obvious, and it’s the step onboarding plans tend to skip: they open with a two-hour architecture overview before the team has deployed anything, which is the corporate version of a grammar lecture on day one.
Mahjong works the same way. For a beginner I’d go from how a hand is built, to tile efficiency, to when to fold, to reading the table, and not in any other order. Tile efficiency alone filled a whole series when I wrote “Tile Efficiency 101” for Osamuko.com, and it’s only the second step.
For an enterprise onboarding, the same logic gives you something like this:
- Access: accounts, SSO, who can do what.
- One real project deployed, even a small one.
- The team’s review workflow on top of it.
- Production: domains, cutover, rollback.
- Monitoring, usage and cost.
Write each step as a can-do (“the team can roll back a production deployment without us”) and you have the objectives for your enablement sessions.
Check understanding instead of asking “any questions?”
“Any questions?” is the least useful question in teaching. Students who are lost don’t know what to ask, and shy ones won’t ask in front of everyone. In a Japanese meeting room with their manager present, silence is the likely answer, and it doesn’t mean anyone understood.
In class, you find out by making students use the thing: answer in Japanese, conjugate the verb, explain a sentence back in their own words. The answers tell you exactly who is following.
Mahjong has the perfect tool for this, the nanikiru problem (何切る, “what do you discard?”). You show a hand and a situation, and the player picks one tile and says why. The tile tells you what they know; the “why” tells you whether they know it or got lucky.
In an enablement session, replace “any questions?” with tasks:
- “Deploy this branch and send me the preview URL.”
- “Roll back to the previous deployment. How long did that take?”
- “If this page got slow tomorrow, where would you look first?”
If nobody can answer the third one, the next session just found its topic.
Let them try early, and fail where it’s cheap
Students who wait until they’re ready to speak wait forever. Mistakes made out loud get corrected; mistakes made silently stay.
A mahjong beginner should play a real game long before they know every yaku (scoring pattern). They’ll deal into hands and forget to call riichi. Good. That’s what a practice table is for.
With customers, the cheap place to fail is a preview environment in the first week, with you there. The expensive place is production in month three, after the team has built on a misunderstanding nobody caught. So push the first attempt early, on purpose.
In Japan this can be harder than it sounds, because nobody wants to look wrong in front of colleagues. Make the mistake part of the exercise (“this step will throw an error; see if you can find out why”). People relax when the failure was planned.
Give them a first win
Trying early only works if the first attempt ends in something that works. A student who can do a self-introduction (jikoshōkai) at the end of the first class comes back for the second. One who spent it on the history of the writing system might not.
In mahjong, I want a beginner to win a hand in their first session, even a cheap one. Riichi and tanyao (all simples) is plenty. That fits my order of priorities, enjoy it first, improve second, win third, because the first win is really the first item.
For a customer, the first win is something real in production, early: a small site, an internal tool, one page of the main product. It gives your champion something to show internally, which, as I wrote in customer success in Japan, is what keeps the project funded.
Office hours
Office hours look optional and aren’t. A fixed time when anyone can show up with anything is where quiet students ask what they wouldn’t ask in class.
Customers need the same thing: a recurring slot, open to any engineer on the account (the integrator’s too), with no agenda. It catches the problems nobody thought were big enough for the regular meeting, which are usually the ones that become escalations. In mahjong coaching, the equivalent is reviewing a game the player brings. It teaches more than a lecture, because the mistakes are their own.