Employee Onboarding: How to Build a Track That Works

Short answer: an onboarding track that works has one business metric (time to productivity), a three-block 30-60-90 structure with an exit criterion per block, content extracted from the people who already train informally, assessed practice in each block, and a cohort component. It is not the welcome deck: it is the path from "started" to "performs independently" — and it is the pilot almost every corporate university should begin with.

Onboarding is also the process companies believe they have solved and have not. Gallup finds that only 12% of employees strongly agree that their organization does a great job onboarding new people — which means the near-universal experience is a first week of firehose followed by silence.

The metric the track has to move

Before writing a single lesson, pick the number. For onboarding it is almost always time to productivity: how many days until the new hire performs, at the company's standard, the work they were hired to do.

This metric has three virtues. It is observable without sophisticated instrumentation — the manager can tell you. It varies by role, which forces the track to be specific instead of generic. And it converts directly into money: every week of ramp removed is salary already paid that starts returning sooner.

The second reason to take onboarding seriously is early attrition. SHRM's widely cited benchmark puts turnover as high as 50% in the first 18 months and estimates replacement cost at six to nine months of the role's salary. Someone who leaves in month four takes the entire hiring investment with them and returns nothing.

The 30-60-90 structure

The established format splits the path into three blocks, each with its own verb:

Days 1 to 30 — understand. Company, product and customer context; tools and access; who is who; the internal vocabulary. The classic error is dumping all of it in week one. Exit criterion: the new hire can explain, in their own words, what the company sells and to whom.

Days 31 to 60 — perform with support. The actual work of the role, done with supervision and review. This is where role-specific processes live: how to log work, how to handle a customer, how to escalate. Exit criterion: performs core tasks with occasional support rather than constant guidance.

Days 61 to 90 — perform independently. Autonomy with spaced check-ins, first owned results, final assessment. Exit criterion: delivers at the company standard without daily supervision — the working definition of full productivity.

Each block closes with a real assessment, not a decorative quiz: a practical exercise, a role-play, a reviewed deliverable. Without assessment nobody knows whether the track taught anything; you only know it was consumed.

What belongs in each phase

Phase	Belongs	Does not belong
1 to 30	Context, product, customer, tools, culture in practice	Deep operational detail of every process
31 to 60	Role processes, real cases, supervised practice	Other departments' content "for awareness"
61 to 90	Autonomy, first goals, final assessment, development plan	Anything that changed and nobody updated

One rule saves a great deal of rework: shared content (history, values, policies, security) lives in a single track reused by every department, while role content lives in separate tracks. Hiring for a new area then requires producing only the specific part.

How to extract content from the people who know

This is the real bottleneck of onboarding, and it was never distribution — it was production. The critical material is not documented; it lives in the head of the senior rep, the long-tenured manager, the founder. That is why so many programs die in the planning phase: asking those people to "write the material" asks for weeks they do not have.

What changed is the size of the ask. With AI structuring content, the flow becomes: the veteran describes the process in a few sentences, or hands over whatever exists in raw form — old decks, recorded calls, scattered docs; the AI organizes modules, lessons, exercises and assessments into a pedagogical sequence; the veteran reviews and corrects. Weeks become hours, and human review stays mandatory — it is what guarantees the track teaches the company's real process rather than a generic version of it. At Tandria this is the native flow: the track is born structured, with narration and assessments, ready for the reviewer who actually knows.

Cohort or continuous entry

The choice depends on hiring rhythm and it has a direct consequence on completion.

Continuous entry is the only viable option with sporadic hiring: each person starts the day they arrive. The risk is isolation — an asynchronous track with nobody alongside is exactly the scenario where self-paced content completes in the 3% to 15% range.

Cohorts work when hiring comes in batches: people advance together, compare questions and build attachment to the company through peers. Completion is materially better.

For most companies the winning design is hybrid: an asynchronous track with continuous entry plus a biweekly live session that gathers everyone who joined in that window. It costs an hour of calendar and recovers much of the cohort effect.

What to measure

Four numbers, in order of importance:

1. Time to productivity, by role — the business metric of the track.
2. Completion rate — if people do not finish, everything else is noise.
3. Pre- and post-assessment — shows what the track actually taught, and where it fails.
4. Retention at 6 and 12 months — the long-term effect, slow to appear and expensive to ignore.

A note on method: none of these say anything in isolation. What informs a decision is the comparison between the ramp of people who went through the track and the ramp of people hired before it existed.

Four mistakes that ruin onboarding

Dumping everything in week one. Volume is not hospitality — it is overload. People forget most of it and have nothing to return to.

A track without an owner. Onboarding without a responsible owner ages badly: the process changes, the track does not. A stale track loses the team's trust once and does not get it back.

Confusing a buddy with a track. Buddies are an excellent complement and a terrible substitute: they transmit what they remember, in the order they remember it, at that day's quality.

Measuring attendance. A meeting attended is not a competence acquired. If the metric does not survive the question "did this shorten the ramp?", it is vanity.

---

A well-built onboarding track is the highest-return pilot in any corporate learning program: the pain is visible, the audience is guaranteed, and the result is measurable in weeks. See Tandria for companies and build your first track during the 7-day free trial.