The short version
- Forty questions in week one, routed to one buddy, at roughly fifteen minutes each including recovery. That is a lost week for the buddy.
- The joiner also loses, because they ration the questions and guess at the rest.
- Both problems have one cause: a single human route to everything the team knows.
- Widen the route. Most of the forty are retrieval, which does not need any particular person to be free.
Assign a buddy and the onboarding plan looks complete. Then the first week happens: a question every twenty minutes, each one small, each one reasonable, all of them arriving at the same person who also has their own sprint commitments.
"Every question costs the buddy 15 minutes."
The arithmetic of forty questions
Forty questions in a week, at two minutes of answering and thirteen minutes of recovered context each, is ten hours. More than a full day of the buddy's week, and it does not appear anywhere in the sprint plan, because each instance is too small to record.
It is also front-loaded into the exact week the buddy was told onboarding would be light touch. This is why buddies dread the role by the second or third time they are asked, and why the good ones start declining.
What the forty questions actually are
| Kind of question | Example | Needs a human? |
|---|---|---|
| Where is it | "Where do the staging credentials live?" | No |
| Is this still how we do it | "Is this pattern current or legacy?" | No, if a decision was recorded |
| Who decides | "Who approves a schema change?" | No |
| Why is it like this | "Why two queues instead of one?" | No, if the reasoning was written |
| Should I do X or Y | "Is this worth refactoring now?" | Yes, this is mentoring |
Four of the five rows are retrieval. Only the bottom row is the thing a buddy is genuinely for, and it is the row that gets least attention because the other four have consumed the week.
Both sides lose
The buddy loses a week and learns to avoid the role. The new hire loses more subtly: aware that each question costs their buddy, they start batching and rationing, and the ones they decide not to ask become assumptions. Those assumptions surface in review, which costs more than the question would have and dents their confidence in the process.
And there is a third loss. Because all the questions went to one person, the new hire ends month one with one relationship instead of five, which slows everything afterwards.
Widening the route
The fix is not a better onboarding document, because the four retrieval rows are not the kind of thing a document answers well: two of them are about current practice and past reasoning, which change faster than any page is maintained.
With StandIn, each person on the team confirms a ninety-second brief at the end of their day, mostly pre-drafted from work that already happened: current state, decisions and why, open questions, next actions. While they are off, their StandIn answers questions from that brief in their words, with a source under every answer, clearly labelled, and it never guesses. When the answer is not in the record, it says so and names who to ask.
The new hire asks the team's StandIns first and gets four of the five rows answered immediately, at any hour, as many times as they need. The buddy's week is spent on the fifth row, which is mentoring, which is what they were actually asked to do. See StandIn for the first 30 days.
What good looks like at thirty days
Four signals, none of which is about how much the joiner has read.
They have merged several changes, including one they scoped themselves. They have asked a question of at least four different people rather than one. They can name who decides what in their area. And they have written down three things that confused them, which is both the best test of engagement and the cheapest improvement to your onboarding for the next person.
Common Questions
How many questions does a new engineer have in the first week?
Commonly several dozen, most of which are retrieval rather than judgement. At roughly fifteen minutes each including the recovery cost, that is more than a day of the buddy's week, which is why the role feels heavier than it is described.
How do we stop the buddy being overwhelmed?
Route the retrieval questions somewhere other than a person, and keep the buddy for mentoring. Four fifths of the volume does not need a specific human to be available, and separating the two makes the role sustainable.
Should new hires be told to ask fewer questions?
No, the opposite. Under-asking is the more expensive failure, because unasked questions become assumptions that surface in review. The aim is to make asking cheap rather than to reduce it.
How do we measure onboarding success?
Count merged changes, the number of different people they have asked something, and whether they can name the deciders in their area. All three are better signals than time spent on documentation.
Log off like you mean it.
Your StandIn answers the questions that come up while you're out, from what you actually wrote down, so the return pile stays small.