How to run safety training for hundreds of students a semester

Most facility software assumes a stable membership that grows slowly. An academic makerspace is the opposite: the user base turns over almost completely every year, most arrivals have never touched the equipment, and the institution carries the liability for what happens when they do.

That changes onboarding from a thing you do occasionally into the central operational workload. Here is how to structure it.

Start by separating three things people conflate

A lot of confusion in this area comes from using one word for three different objects:

• A course is content. Slides, a video, a quiz. It can be taken.
• A credential is a fact about a person. It has an issue date, possibly an expiry, and it can be revoked.
• Access is what a credential permits. The laser turns on, or it does not.

Keeping these distinct is what lets you do things that are otherwise impossible: issue the same credential from an online course or an in-person session, recertify someone without making them retake content they have already seen, or revoke one person’s access without touching anyone else’s.

If your current system collapses these into “completed the training,” you will hit a wall the first time someone needs a credential they did not earn by the default path.

Design the prerequisite graph before you build any content

The instinct is to build a list of courses. Real training pipelines are not lists, they are graphs, and the shape matters:

Shop Orientation & Safety
        ├──> Wood Shop Fundamentals ──> CNC Router Certification
        ├──> Metal Shop Fundamentals ──> MIG Welding ──> TIG Welding
        └──> Laser Cutter Operation

Two properties are worth getting right up front:

AND across levels. CNC certification requires wood shop fundamentals, which requires orientation. Nobody skips a level.

OR within a level. Sometimes several different prior credentials qualify someone for the same next step — a student who did metal shop fundamentals and a machinist hired from industry might both be eligible for the lathe. If your system only supports “all of these,” you will end up issuing credentials by hand to work around it, which defeats the point.

Sketch this graph on a whiteboard with the people who actually teach the material, before anyone authors a slide. It is much cheaper to move an arrow than to rebuild a course.

Decide what genuinely has to be in person

Not everything does, and the ones that do are expensive — they consume staff time that scales linearly with enrollment.

A workable split:

• Online, asynchronous: general shop orientation, PPE, emergency procedures, shop rules, hazard awareness. Anything that is knowledge rather than motor skill. These are the highest-volume courses, and making them asynchronous is what makes 600 students a semester feasible at all.
• In person, hands on: anything where you need to watch someone do it. Machine operation, tool changes, work holding. Keep these short and make the online prerequisite mandatory, so instructor time is spent on the hands-on part rather than on repeating the rules.

The economics matter here. If orientation is in person, you are spending instructor hours on content that a video delivers identically. That is the capacity you need for the machines.

Make the online half enforceable, not decorative

The failure mode of asynchronous safety training is that people click through it. If your defence is a completion checkbox, you have documentation but not learning — and documentation that is known to be hollow is worse than none, because it creates false confidence in an incident review.

Things worth requiring:

• Watch enforcement on video, validated on the server rather than trusted from the browser. A client-side check that a video “finished” is a suggestion.
• A quiz with a real threshold, set by you, with the correct answers never sent to the browser before grading.
• Sequential progression — a block can only be completed once every block before it is done, enforced on every call rather than only in the interface.

None of this makes the training good. It just means the completion record means something.

Run in-person sessions like the constrained resource they are

Hands-on sessions are your bottleneck. Treat them accordingly:

• Cap them and waitlist them. A session with unlimited signups is a session where twelve people show up for six machines.
• Gate signup on prerequisites. If someone has not done the online orientation, they should not be able to book the hands-on slot. This is the single highest-leverage rule you can enforce, because it eliminates the most common way instructor time gets wasted.
• Take attendance at the session, and issue from the roster. Bulk issuance from an attendance list is a minute of work. Issuing credentials one at a time afterwards, from memory or a photo of a sign-in sheet, is where records go wrong.

Plan for expiry from day one

Every credential should have an answer to “how long is this good for?” — even if the answer is “indefinitely.”

Where expiry applies, the important design decision is what expiry does. An expiry that sends an email is a reminder, and reminders get ignored. An expiry that pauses machine access is a control, and it enforces itself.

The recertification path should also be shorter than the original. Someone who has run the CNC for two years does not need the full course again; they need the refresher and a new date on the credential.

Let members see their own status

A surprisingly large fraction of front-desk time is spent answering “am I checked out on the laser?” and “what do I need to do next?”

Both are answerable by the member if the pipeline is visible to them: what they hold, what is expiring, what is locked, and specifically which prerequisite is blocking the thing they want. That last part matters — “locked” is unhelpful, “take MIG Welding Basics next” is actionable.

Instrument it, because someone will ask

At some point a department chair, an EHS officer, or an insurer will ask a question like:

• How many students completed safety training last semester?
• Who was certified on this machine on this date?
• What is our completion rate between orientation and hands-on?
• Can you show that this person was trained before the incident?

These are all trivial if the credential is a dated record attached to a person and the whole chain is one system. They are close to unanswerable if the evidence is split across a learning platform, a spreadsheet, and a filing cabinet.

The shape that works

Putting it together, the pipeline that scales looks like this:

1. Asynchronous online orientation, enforced, available continuously — no scheduling bottleneck.
2. A prerequisite graph that gates hands-on sessions automatically.
3. Short, capped, waitlisted in-person sessions for motor skills only.
4. Bulk credential issuance from the attendance roster.
5. Credentials that gate the machines themselves, so the record and the enforcement are the same object.
6. Expiry that pauses access, with a short recertification path.

The load that scales with enrollment is then almost entirely the asynchronous half, which costs nothing per additional student. The expensive half stays proportional to machines and instructor hours, which is the constraint you actually have.

---

MakerOps includes a native course builder — slide, video, and quiz blocks with server-side watch enforcement — plus SCORM 1.2 import, an AND/OR prerequisite graph, capped and waitlisted in-person sessions with bulk issuance, and credentials that physically gate connected machines. Student accounts are unlimited on every plan. See how training works or read about academic makerspaces.