Why New Engineers Stay Unproductive for Weeks: A PeopleOps Diagnosis

Why New Engineers Stay Unproductive for Weeks: A PeopleOps Diagnosis

Here is a scene most of us have lived through. A new engineer joins on Monday, spends Monday and Tuesday waiting for a laptop and a login, spends Wednesday reading outdated documents, and by Friday has pushed nothing but a typo fix. Everyone then says the hire is “ramping up”. Often the engineer is fine and the machinery around them is not. Let us diagnose it like an incident, symptom by symptom.

The symptoms

  • Equipment or accounts arrive after the start date.
  • The first week has no written plan, only a series of introductions.
  • Nobody can say who to ask about small things.
  • Questions get answered differently by different people.
  • The manager discovers gaps in access during the first sprint.

These are the kind of problems that the Ops Atlas folds into the discipline it calls PeopleOps, described on its PeopleOps page. The Atlas’s view is that the hiring-to-exit journey is a set of operations that should have owners and promised timings. Let us apply that to onboarding.

Diagnosis one: nobody owns the countdown

In engineering we would never launch a release without a named owner and a checklist. Yet onboarding often runs on goodwill. The Atlas cites a documented company practice worth stealing: a company whose onboarding ticket is created no later than four days ahead of a person’s first day, which gives hardware and accounts time to turn up. If your task opens on the morning the person arrives, you have already lost the first week.

Fix: create the onboarding request when the offer is accepted, assign it to a named coordinator, and measure what fraction of starters have everything on the first morning. The Atlas offers an example target of at least 95 percent for that readiness.

Diagnosis two: access is requested by hand

When every account is raised by a separate ticket, something is always late. The event that matters, “this person is joining on this date”, should flow from the people record to the systems, not be retyped by a manager. That requires a record that holds start dates and stable identifiers, and a provisioning route that reads it.

Fix: list the first-week tools, group them by role, and convert the common bundle into a template that is raised automatically. Keep exceptions manual and rare. Also decide what happens when a person never turns up. The Atlas flags no-shows as a risky case because accounts were created in advance, so make sure they are cleaned up if the start is cancelled.

Diagnosis three: the plan is in somebody’s head

An engineer without a plan fills the gap with anxiety. The Atlas recommends the manager provide a thirty, sixty and ninety-day outline by about the third day, with specific goals. Pair that with the four themes it takes from a well-known onboarding model: the compulsory bits, clarity about the role, introduction to culture, and human connection. The model also sorts organisations into passive, high-potential and proactive onboarders, which is a useful mirror to hold up to your own team.

Fix: write a template plan with space for team-specific goals, and make sure the first week includes one small production-bound change. Nothing builds confidence like shipping something.

Diagnosis four: no buddy, no safe questions

Engineers ask different questions of a peer than of a manager. A named buddy means that “where is the staging config?” gets an answer in minutes, not a polite email chain. It also protects the manager’s calendar.

Fix: assign a buddy before day one and give them a short briefing, including an hour a week of protected time for the first month.

Diagnosis five: feedback is gathered but never used

Some teams do ask newcomers how it went. The Atlas describes one large company that surveys new starters at 30, 90 and 365 days and analyses the results by segment, and it reports that the findings led to a closer link between company-wide orientation and local onboarding. It then adds a caution worth repeating: a copied questionnaire achieves nothing if nobody acts on what it reveals.

Fix: when you ask, publish what changed. Even a single line in the team channel, “you said the staging guide was wrong, so we rewrote it”, shows that the questions matter.

Diagnosis six: moves and absences are forgotten

Onboarding problems do not stop on day five. A newcomer who transfers teams after a year will carry permissions from the old role unless someone reviews them, and the Atlas calls these moves the forgotten event of the lifecycle. Similarly, a person who starts while the previous owner of their work is still leaving can inherit half-finished tasks and unclear ownership.

Fix: treat each change of team as a small onboarding of its own, with a review of access, a refreshed plan and a handover note from the previous manager.

A note on managers

You will often hear that great managers make great onboarding. The Atlas mentions a large internal study at one technology company that drew on more than ten thousand data points to define ten manager behaviours, and then quickly describes it as one company’s finding about itself, not a general law. It is a fine source of ideas and a poor source of certainty. Use it to start a conversation about what your own managers can do, then test the idea in your own team.

What not to do

  • Do not measure new engineers by how many keystrokes or commits they produce in week one. The Atlas places that kind of individual monitoring firmly outside the discipline.
  • Do not mix a development conversation, a pay conversation and a performance concern in the same meeting.
  • Do not assume that a late laptop is the new hire’s problem to solve.

The follow-up check

Two weeks after the start date, ask three questions. Did everything arrive on time? Did the plan exist on day three? Did the buddy meet them? Record the answers and review them monthly. The pattern across several hires will show you which diagnosis applies to your team more than any single story can.

Keep building the career side

Good onboarding is also what lets people grow. For the long view, take a look at how DevOps practice supports growth across a tech career“>how DevOps practice supports growth across a tech career and at what certifications are really worth in modern tech careers“>what certifications are really worth in modern tech careers.

The treatment is rarely dramatic. A named owner, an early start to the paperwork, a written plan and a buddy fix most of what looks like a slow hire. Try them on the next person who joins and see how much faster the first pull request lands.