How to Onboard a New Hire with AI
Meridian Commerce runs cross-border e-commerce for eleven client brands with a team of thirty, and every number in the company still travels through someone’s spreadsheet. That is why the agency is hiring its first data analyst, Sofia Marchetti, who starts in eleven days, and why ops lead Dev Patel is nervous. There is no existing analytics role to shadow, no predecessor’s calendar to copy, and a real risk that the first month disappears into account access and vague introductions.
This workflow builds the onboarding before the new hire walks in. ChatGPT drafts the 90-day scorecard, the pre-start logistics list, a day-by-day first week, and the check-in rhythm, so that a first-of-their-kind role has a shape from the start. The whole plan takes an afternoon to generate and an hour to edit, which is far cheaper than losing six weeks to drift.
Start With the 90-Day Scorecard
Onboarding a first-of-a-kind role begins with one question: what must be true by day 90 for this hire to have worked? Answer that before you plan a single meeting.
- By day 30, Sofia has access to every data source and can reproduce three client reports without help.
- By day 60, one recurring manual report has been automated and one client dashboard exists in Looker Studio.
- By day 90, she has shipped a monthly performance view that the account team actually uses in client calls.
- Success is measured on adoption, not on volume of dashboards built.
Write that scorecard down. Every later prompt refers to it, and it becomes the honest basis for the day-90 review.
The instinct with a brand-new role is to measure effort, like the count of dashboards or reports shipped. Resist it. A first analyst at a small agency succeeds by getting people to trust the numbers, and trust shows up as usage, not as output. If the account team opens the dashboard in client calls by day 90, the hire has worked even if only two dashboards exist.
Clear the Logistics Before Day One
Most first days are lost to missing logins. Generate the checklist early enough to actually finish it, then verify each item yourself.
You are an operations lead onboarding the first data analyst at a 30-person
cross-border e-commerce agency. Build a pre-start logistics checklist to be completed
before day one. Cover:
1. Hardware and accounts: laptop, email, password manager, SSO, calendar.
2. Data and tool access: Shopify admin, Meta Ads Manager, Google Analytics 4,
Google Ads, the BigQuery warehouse, and Looker Studio.
3. Introductions: which two client teams to meet in week one, and why.
4. Day-one essentials: who greets her, where the first meeting is, and one lunch.
5. Anything that requires a vendor or client to approve access in advance.
For each item give an owner and a "days before start" deadline. Flag anything likely
to take more than three days to unblock.What you get: a dated checklist with owners. The access rows are the important ones, because a client-managed ad account can take a week to grant read access, and that wait should happen before Sofia arrives rather than during her first week.
Send a short welcome note three days before the start date. It should cover where to go, who greets her, what the first meeting is, and the one thing to read first. A single page of preparation removes most of the first-morning friction and signals that the company planned for her, which matters more at a thirty-person agency than at a large one where the machinery exists already.
Design Week One Day by Day
A first-of-a-kind role with no peer to follow needs an explicit first week. Left open, it becomes five days of improvised meetings.
Design Sofia's first week, one block per half day, for a first data analyst at an
e-commerce agency. Goals for the week: understand the eleven client brands, learn
where each data source lives, and produce one small deliverable by Friday.
For each half-day give: the task, the person she is with, and the artifact she
leaves with, such as a notes doc, a data map, or a chart.
Keep the first afternoon light on meetings. Build in one open two-hour block daily
for reading and setup.What you get: a week where every half day has a purpose and an output. The Friday deliverable matters most: a single simple chart from real client data gives Sofia a win and gives Dev something concrete to react to.
Keep the first week deliberately under-scheduled. New hires need white space to read client briefs and poke at the warehouse without an audience, and a calendar packed with thirty-minute introductions leaves no room for that. Two or three meetings a day is plenty for a thirty-person company where Sofia will meet everyone within a fortnight anyway.
Map the 30-60-90 Plan
Now expand the scorecard into a working plan with deliverables and a definition of done for each phase.
Turn the 90-day scorecard into a phased plan.
For the 30, 60, and 90 day marks, list:
- two to three concrete deliverables,
- the definition of done for each,
- which existing team member unblocks it,
- how we will know it landed (who uses it and how often).
Then list three skills or domains Sofia should learn, given she is new to
cross-border e-commerce and to agency client work specifically.
Keep it to one page.What you get: a plan where the day-90 item is tied to real usage. “Dashboard adopted in client calls” is a definition of done; “improve reporting” is not. If the plan reads like a wish list, tighten each item until it is checkable.
Plan to revise the document at day 30. A first-of-a-kind role always reveals assumptions the manager could not see from outside, and the 60 and 90 day sections should bend to what Sofia learns in the first month. Treat the plan as a living draft reviewed at each checkpoint, not a contract signed on day one.
The phased structure here is the same one we use to scope any multi-week effort, described in our AI project plan workflow.
Set the Check-in Rhythm and the Questions in Advance
First hires rarely ask for help early enough. Make the check-ins regular and script the answers to the questions new people are too polite to raise.
Design the check-in cadence for Sofia's first 90 days:
1. A daily 15-minute sit-down for week one, then twice weekly, then weekly.
2. A 30-day and a 90-day review with a simple template.
3. A standing question from the manager at every check-in.
Then write a short onboarding FAQ that answers the questions a new hire is unlikely
to ask out loud, such as: what if I do not understand a client's data, who reviews
my work, and how do I flag that a deadline is unrealistic.
Keep the FAQ to five questions.What you get: a rhythm that makes asking easy and a FAQ that removes the early awkwardness. The manager’s standing question, “what is slowing you down right now”, catches blockers that a status update hides.
Give the buddy a job beyond answering questions. Dev, as the assigned buddy, should own the three client teams that supply messy data, so Sofia has one person who knows which feeds are reliable and which need a manual nudge each Monday. That single relationship is often what separates a hire who is productive by week three from one who is still guessing in week eight.
Onboarding Deliverables Checklist
- A written 90-day scorecard with measurable success criteria.
- A pre-start logistics checklist with owners and dated deadlines.
- A day-by-day first-week plan with an artifact per half day.
- A one-page 30-60-90 plan with definitions of done.
- A check-in cadence and a five-question onboarding FAQ.
- A named buddy and a first-week deliverable that uses real client data.
Where First-Hire Onboarding Fails
Treating access as a day-one task. Client-managed tools take days to unlock. Finish it before the start date or lose the first week.
Leaving the first week unstructured. With no peer to follow, an unplanned week becomes a blur of intros. Schedule half days and require an artifact from each.
Setting vague 90-day goals. “Improve reporting” cannot be reviewed. Tie each goal to a thing someone uses, and how often.
Waiting for the new hire to ask. First hires stay quiet to look capable. Put a standing question in every check-in and publish the FAQ.
One more failure hides in plain sight: expecting the first analyst to invent the role alone. Nobody has held this job before, so there is no template to inherit. Hand Sofia the 90-day scorecard in writing, and tell her explicitly that the plan will change as she learns the business. That permission to revise is what turns a risky first hire into a hire who shapes the role instead of drowning in it.
Before this stage, the same discipline applies to hiring. The scope-and-requirements structure in our job description prompt keeps the role definition consistent with the onboarding plan you build here.
Sources & Notes
The scorecard-to-check-in sequence is our own framing of standard onboarding practice, reviewed on 2026-10-09. Meridian Commerce, Dev Patel, and the new analyst starting in eleven days are illustrative examples, and the eleven client brands are representative rather than actual. Tool names such as Shopify, Meta Ads Manager, Google Analytics 4, BigQuery, and Looker Studio are used as common examples and each requires its own access process. Language models change between versions, so re-check any prompt that references your specific tooling or client approval workflow.
