The Queue Is Not a Curriculum: Build a Training Caseload

To train a new hire on real work, hand them a deliberate caseload instead of the live queue. Choose about ten real cases that are deliberately different from one another, put them in an order you picked, and have someone review each one before it closes. The live queue cannot do that job, because it delivers work in whatever order customers happened to send it. That order is a coincidence, and a coincidence is not a curriculum.

Whatever walked in the door that day

In a retirement servicing team we interviewed, the person who trains new hires described the entire program in one sentence: there is no training case load. New people learn whatever comes into the queue that day, in whatever order it arrives.

The consequence she named is the one every operations lead recognizes. Someone can be two weeks in, fully confident on two case types, and then be handed a case type they have never seen. Confidence and coverage are not the same thing, and a queue only builds the first one.

This is also why the training looks fine from a distance. Completion is high, the new hire is busy, and the gap surfaces only when an unusual case lands on a Friday. If you have tried to work out why your employee training isn't working while every report says it is, this is usually the reason.

A queue hands out work in the order customers happened to send it, which makes it a schedule, not a curriculum.

What is a training caseload?

A training caseload is a fixed set of real cases, chosen in advance to cover every case type the role actually handles, worked by the new hire in an order you decided rather than the order the queue produces.

Three words carry the weight. Fixed, because the list does not grow when the department gets busy. Real, because these are live cases with live consequences and a named reviewer attached. Chosen, because somebody listed the case types before the new hire's first Monday.

It is not shadowing. Shadowing puts a new person next to the work and hopes proximity does the teaching, which is why shadow training fails as a program: the learner absorbs whatever the veteran happens to be doing that hour. A caseload inverts the roles. The new hire runs the case, the veteran reviews it.

Ten cases, as long as they are ten different ones

Asked how many cases someone needs before working alone, the retirement servicing specialist who trains her team's new hires did not give a number of days. She gave a shape: about ten cases, provided they are ten different ones.

That is the decision rule worth stealing. Ten cases of one type teaches speed. Ten different types teaches judgment, which is what you are actually trying to install. Count distinct case types, not days on the job.

It scales in both directions. A two-person department runs it on a whiteboard. When you are training a large department of thirty or more people, the caseload becomes the schedule: cohorts move through the same ten types in the same order, and coverage is visible at a glance.

How to train a new hire on real work: sequence by case type

Sequencing is where most attempts collapse, because the honest version requires somebody to write down what the role really does. Here is the order we use.

1. List the case types. Take them from the person working the queue today, not the org chart or the job description. Both were written by someone who has not run a case in years.
2. Rank by damage. In retirement servicing, a missed required distribution carries a large statutory penalty, which is exactly the case type that should never be learned by accident.
3. Start with single-path types. Others split: a beneficiary update clears online with the custodian the same day, while the same update at an insurance carrier runs by fax and phone over several days. Save the forks for week three.
4. Open a long one early. Cases running up to two months outlast the operator's memory of what they were waiting on. Start one in week one so the new hire learns the waiting, not just the doing.
5. Fill the holes on purpose. If a case type has not appeared by the halfway point, stop waiting. Pull a closed one from last quarter and walk it end to end.

That list is not a syllabus of concepts. Every item is a live case with a customer attached, and the sequencing is the only artificial part. One reliable way to build it is to ask what your last new hire got wrong and treat every answer as a case type that earned its place.

The cost is real, and it lands on your best operator. Every case on the list needs a reviewer who reads it before it closes, and the new hire's throughput drops for the first few weeks because they keep meeting types they have never seen. If the role has only two or three case types, skip the caseload and sit with the new hire for a week.

Practice before the customer is real

Some things should not be learned on a live file. There is a recurring error in that servicing team where mailed forms come back and staff forget to add the social security numbers. Her verdict: "It happens to every single person. It might just be one of those things that it has to happen to you before you're like, oh yeah, I got to remember to do that."

You can at least choose where the first time happens. Let the predictable first mistake land on a case you constructed, not on a client's. A brokerage trainer we recorded got there sideways: asked to screen-share her contact database to demonstrate tagging, she stopped over client privacy, then found close to a thousand tags in it, mostly old ad campaign labels. She built a dummy record with clean categories instead.

Practice also forces you to define the standard. At a staffing agency, a customer service trainer had two representatives on the chopping block, two percentage points apart and both below the 76 percent bar. His plan was to stop grading them on tone, which is arguable, and start enforcing call avoidance and schedule adherence, which are not. He also reworded a rubric item so he was hunting for one specific sentence, findable in the first thirty seconds.

That is what a practice case buys: a vague standard turned findable before the stakes are real, which is the same logic behind mock call training.

Name the vocabulary gap out loud

Ask whether your new hire arrives knowing your account types, then answer honestly. In that servicing team, most do not. A planner says "pull it from the non-qualified account" and the new person has no idea what that means, so they guess, interrupt someone, or sit on it until Friday.

The trainer said the more useful thing. Walking through the account types out loud, she hesitated on one and admitted: "I don't even know if I actually know." That admission is worth more than a confident glossary, because it marks the term the whole department has been quietly approximating.

So write the page before case number one. Plain definitions, no theory, read once and kept open. It is short enough to be the rare kind of training your team actually finishes.

Vocabulary is the cheapest thing in the business to document and the first thing everybody skips.

How do you know they can run one alone?

In interviews with servicing operators, the same two questions surface, and they come from the learner rather than the manager. "Am I doing this correctly?" was named as the single biggest hang-up for new hires in servicing. "Can I be trusted to do this on my own?" is the same worry, asked later.

A caseload answers both with one artifact. A case type is signed off when the new hire has run two of that type with no corrections from the reviewer, and the tracker shows every type with a date beside it or a blank. Readiness stops being a feeling and becomes a list with visible holes.

Watch the internal transfer most closely of all. That specialist trained someone who had spent a year at the company's front desk and assumed the year meant they understood the back office. "I didn't realize that it was always from like a peripheral view," she said, and her conclusion was blunt: it was easier to train a brand new person than someone who was already there.

When we build a caseload at The Systems Effect, the case types come from recorded sessions with the person working the queue today, because the manager's version of the case mix is always shorter than the real one. Run the first pass yourself this week. Sit with whoever works the queue for one hour, write down every case type they can name, and count how many your last new hire has touched.

Frequently Asked Questions

How long does it take to train a new hire on casework?

Long enough to work about ten different case types, which in a busy queue takes a few weeks and in a quiet one takes longer. The useful measure is coverage, not calendar time. Someone who has run ten cases of two types is less ready than someone who has run six cases of six types.

How much shadowing is enough before someone works alone?

Shadowing is enough once the new hire can name the steps and the decision points out loud, which usually takes days rather than weeks. After that, watching is a poor use of two people's time. Move them into a reviewed caseload where they run the case and somebody checks it, signing off one case type at a time.

What is a training caseload?

A training caseload is a fixed set of real cases, picked in advance to cover every case type the role handles, worked in a deliberate order with a named reviewer. It replaces the live queue as the curriculum. The queue still gets worked, it just stops deciding what the new person learns.

Why is a promoted internal hire sometimes harder to train than a new one?

Because everyone, the trainer included, assumes context the person only ever had peripherally. One servicing lead we interviewed trained a colleague who had spent a year at the front desk and found it harder than training an outsider, because nobody thought to start at the beginning. Run internal transfers through the same caseload, in the same order, with the same sign-offs.