HR playbook · AI Employee: Harry

Onboarding Stall Detection

Stalled onboarding tasks resolved within 48 hours

Install this playbookBack to marketplace
The problem

Onboarding fails quietly. A new hire's manager forgets to schedule the second-week 1:1, IT forgets to grant access to the internal wiki, the buddy programme quietly lapses. Nobody notices until the 90-day review when the hire says "I've been blocked on X for six weeks." By then the goodwill of week one is gone and the ramp curve is broken.

At a glance
Trigger
Cron (daily during first 30 days)
Approvals
None
What it does
Read-only
Systems
HRIS · Ticketing · Learning management
How it feels in production

An hour-by-hour walkthrough.

Every morning, Remy scans every new hire in their first 90 days. For each, it checks a battery of onboarding signals: has the second-week 1:1 happened, has the buddy sent an intro Slack, does the hire have access to the top 20 apps for their role, have they completed the mandatory training, have they made their first commit / first ticket / first case (role-dependent), have they attended the required syncs. Priya joined engineering 18 days ago. Remy detects: no commit yet (engineering median is 9 days), no access to the internal design docs Notion, buddy hasn't messaged since day 3, 1:1 with manager missed for two weeks. Three signals crossing threshold. Remy DMs Priya's manager Sarah with the specifics: "Priya is stalled — no commits, no design-doc access, buddy inactive, 1:1 slippage. Suggested next actions: grant Notion access (I can do this), reschedule the 1:1 (I can propose times), check in on the buddy pairing." Sarah approves the access grant and 1:1 reschedule with one click. Remy executes both, and separately DMs the buddy to ask if the pairing needs to be swapped. Weekly digest to the People team surfaces every stalled onboarding across the company, with the manager's response time as a leading indicator of manager health.
How it works

Step by step.

  1. 01

    Track every new hire on a signal matrix

    For each new hire in first 90 days, monitor: 1:1 cadence, buddy activity, app access grants, mandatory training completion, first-work milestones (commit / ticket / case), team sync attendance. Signals are role-specific.

    HRIS · Okta · Slack · Zoom · GitHub · Jira
  2. 02

    Detect stalls against role benchmarks

    Not a fixed threshold — compare each hire against the median for their role. A design hire making no commits at day 18 is fine; an engineer at day 18 is behind. Two crossed signals trigger a manager DM.

    Benchmarking · Role model
  3. 03

    Surface to the manager with next actions

    Manager DM naming each signal, the median it fell below, and the specific next actions Remy can take (grant access, reschedule 1:1, DM the buddy). One-click approve.

    Slack · Teams · Approval flow
  4. 04

    Execute the recovery

    On approval: provision missing access via Okta, propose 1:1 slots to the manager + Priya's calendar, DM the buddy separately. Track the recovery — signals should return to normal within 5 days.

    Okta · Google Calendar · Slack · Buddy directory
  5. 05

    Weekly People-team digest

    Every stalled onboarding surfaced with: manager response time, recovery status, most common stall signal by role. Manager response time is a leading indicator of manager engagement.

    Analytics · Slack digest · HRIS reporting
Systems and wiring

What you connect to make this run.

HRIS · Okta

read+write

Read new-hire roster + start dates. Write missing app-access grants when the manager approves. Access grants respect the role's standard entitlement map — no lateral privilege escalation.

GitHub · Jira · Zendesk · Salesforce

read

First-work signals: first commit, first ticket picked up, first case handled. Read-only — the presence of these events tells Remy the hire is unblocked; absence is the stall.

Google Calendar · Zoom

read+write

Read 1:1 cadence + attendance. Write proposed reschedule slots with both parties' availability. Never schedules without manager approval.

Slack · Teams

read+write

Read buddy DMs (frequency + last-message-date, not content). Write manager DM with signals + actions; write buddy DM separately when the pairing looks inactive.

What changes

Before and after, honestly.

Onboarding stalls detected before 90-day review
Before
20-40% (surfaced in review, too late to recover)
After
90%+ (detected within 3 days of first signal)
Median days from stall to recovery
Before
30-60 days
After
5-10 days
90-day attrition of new hires
Before
8-15%
After
3-7%
Time-to-first-commit / -ticket / -case
Before
18-30 days
After
9-14 days (blockers cleared faster)
Related

Playbooks that pair with this one.

HR

New Hire Onboarding

The proactive setup that this playbook watches for gaps in.

HR

30/60/90-Day Check-Ins

Uses the same signal matrix at fixed cadence rather than by exception.

HR

New-Hire Handoff

Sits upstream: a clean handoff from recruiter to manager reduces day-one stalls.

Frequently asked

Answers about this playbook.

Won't managers feel micromanaged by weekly nudges?

Only stalled hires generate a DM — a healthy onboarding gets nothing. Most managers hear from Remy once every 6-8 hires. The framing is "here's what I noticed, here's what I can fix" not "here's your report card."

How does it handle remote vs. in-office hires?

Signal weights adjust. In-office hires have physical badge-swipe + desk-check-in signals; remote hires have more weight on Slack + Zoom + first-work signals. Same detection logic, different inputs.

What if the manager doesn't respond to the DM?

Escalation ladder: manager-of-manager at 3 days, People-team lead at 5 days. Non-response is itself a signal — the manager is either overloaded or disengaged, both worth surfacing.

Can we customise thresholds per team?

Yes — role and team both carry signal weights and thresholds. A hardware engineering role may have longer first-commit medians; a sales role may have earlier deal-touch milestones.

How does this feed into performance reviews?

Onboarding health data is separate from performance — a stalled onboarding is usually an organisational failure, not the hire's. The signals stay in the onboarding record; performance data lives elsewhere.

See it run on your data.

Free plan, no credit card. Connect the systems this playbook needs and run it against a past event first.

Book a demoMore HR playbooks