30-60-90 Day Check-ins: What to Ask at Each Milestone, and What to Do With the Answers

Day 30 is about expectations: does the job match what they were told, and did training prepare them. Day 60 is about the manager: are they getting support and clear goals. Day 90 is about the decision: looking at the next twelve months, do they see themselves staying, and if not, what is driving it. The core script is the same each time; the emphasis moves.

Two rules make the plan work. Someone other than the manager asks, because the manager is a frequent answer. And every conversation is scored into the same fields, so by day 90 you have one record per person and a list of names, not three sets of notes.

Below: the questions for each milestone, the reasoning, a copy-paste template, and how to run it when you have four hundred joiners a quarter and not four.

Ask people who left inside a year when they first thought about it, and the answer is rarely month eleven. It is the first quarter, when the job turned out to be different from the pitch, or the training ended before they were ready, or the manager who interviewed them handed them to a supervisor who never checks in.

None of those are surprises to anyone who has run frontline hiring. What is missing is a way to hear them while they are still fixable. A 30-60-90 plan is that mechanism, if the conversations are honest and the answers go somewhere.

The first check-in is short and mostly listening. Open with "How has your experience been so far, these first few weeks?" and let them set the tone. Then steer, without reading a list, toward the three things that go wrong early.

Expectations. Does the work match what they were told at hiring? Expectation mismatch is one of the most common risk drivers on the calls we run, and at day 30 it can still be corrected by a manager conversation rather than a resignation.

Training. "How effective was the training in helping you do your job?" Score it on a five-point scale afterwards. People who say training was slightly or not effective are telling you they are about to underperform and be blamed for it.

Settling in. How settled do they feel in the team, and is there anything they need that they have not asked for? Small asks at day 30 are cheap.

Do not ask the twelve-month question yet. It is too early for a real answer, and asking it makes the call feel like an interrogation.

By day 60 the honeymoon is over and the manager relationship is the variable. "How satisfied were you with the support and guidance from your manager during this period?" Then: are their goals clear, do they know what good looks like this quarter, have they had a real one-on-one.

This is where the neutral-asker rule earns its keep. A supervisor running this check-in gets a 4 out of 5 from everyone. A neutral voice on the phone gets "mostly absent, honestly" from the people who are already updating their CV.

Ask, on this call, whether anything has come up that they have raised with someone here. The answer sorts concerns into the ones nobody has heard and the ones that were heard and ignored, which need different escalation.

Now the gate question, asked plainly: "Looking at the next twelve months, how do you feel about staying on here?" Read the whole answer. "I think so, but let's see" is unsure, not yes.

Branch from there. Settled people get asked what keeps them and what to keep doing. Unsure people get asked the one thing holding them back and what would need to change. People who may leave get asked what is driving it, whether they have raised it, and the single thing that would make them stay. Everyone closes with the 0 to 10 recommend question.

The full branching set, with the reasoning for each question, is in our guide to employee check-in questions.

The milestones are a setting, not a rule. Retail and BFSI teams often add a check-in before the season that historically loses people, and a six-month or annual one that repeats the gate question. Each milestone is its own cohort with its own script, and each person's calls accumulate on one record, so a change from settled at day 90 to unsure at six months is visible as a change.

The manager should have their own one-on-ones; those are not the check-in. The check-in is the conversation the employee can have about the manager. HR can run it, but HR in the same building tends to get the guarded version too.

What works is a neutral voice, on the phone, in the language the employee actually speaks, with one honest line at the start: what you share goes to senior HR leadership, under your name, so that someone can act on it. Attributed, and told so up front. An anonymous program feels safer and cannot be acted on.

Score every call into the same fields: commitment (committed, unsure, at risk), risk drivers and stay drivers from fixed lists, what they asked for, whether they raised it with anyone, training effectiveness, manager support, and the recommend score. Keep their own words beside every category as evidence.

Derive risk from a published rule. High: said they may not stay, or raised something that needs a formal review. Medium: unsure, or settled but would not recommend. Low: settled. Unknown: never reached, and unknown is never counted as low.

Then give HR a queue, not a report: a list of names filterable by team, manager, driver, and time in the job, where every row opens the person, their quotes, and the recording. Protect people in the aggregates by never charting a group smaller than five, and naming withheld groups with their headcount rather than dropping them silently.

Everyone gets the opening and the close. The middle moves with the milestone.

Three check-ins per joiner, four hundred joiners a quarter, is twelve hundred conversations. Nobody has that many HR business partners, which is why most 30-60-90 plans quietly become a day-30 email survey.

The way to keep it a conversation is to make it a call program. An AI counselor phones each person at the milestone you set, opens by asking which language they would like, states where the feedback goes, runs the script with follow-ups where an answer is thin, and scores the call into the fields above. Three attempts inside working hours, callback capture if someone says call me tomorrow at four, and a short call never counted as a completed one. The output is the queue with names on it, per-person reports, and a cohort workbook with the action list already written.