A training record in an aerospace or defense organization answers four questions: who enrolled, who finished, what they scored, and when the record expires. Those four facts run the training function and, for a certificated repair station, form part of the records the regulation requires. They do not answer the question a Head of Training is asked when a difficult installation is coming up: can this person do this work correctly now.
The short version. A completion record holds four facts: enrolled, finished, scored, expires. None of them show whether a technician will do the work correctly next month. Closing that distance takes a shared record of skill evidence: what someone practiced, what they demonstrated, and what changed in the work since, kept current alongside the systems already in place.
What a completion record holds, and where it stops
The record stops when the course does. A learner finishes the module on connector troubleshooting, passes the assessment attached to it, and the system writes the result. Everything that happens afterwards — the first real intermittent fault, the step that got skipped, the supervisor who corrected it at the bench — happens in a different place and is written down, if at all, somewhere the training team never sees.
None of this is the LMS falling short. An LMS delivers training and keeps the record, and it does both jobs. The gap sits in what gets recorded: the training system holds the course, the work holds the evidence, and nothing joins them.
Why finding the right answer is not the same as doing the work
A newer class of workplace tool has made the first half of this much better. Connect the manuals, the service bulletins and the drawings, and a technician can retrieve the current document in seconds instead of searching for it. That is a genuine improvement, and it removes a common failure mode: working from a superseded revision. It also happens to be the half that knowledge retention work has always sat apart from.
The second half is untouched. Retrieval answers what the correct procedure is. It says nothing about whether this technician will test the right connection first, read the result correctly, recognize when the evidence points somewhere else, and document the repair so the next person can follow it. Pfeffer and Sutton named this distance in The Knowing-Doing Gap, and it is why knowing where the answer lives and being able to act on it are two different capabilities.
What a record of skill evidence holds
The missing record is narrow and specific. For a given person and a given skill, it holds the evidence that they can do it and the date that evidence was produced. That makes it a different kind of object from a completion row: it accumulates, and it goes stale on its own when the underlying procedure changes.
The signals feeding it already exist in most organizations. Courses and assessments show where someone slowed down or moved through cleanly. Scenarios and role plays built from your own material show applied judgment rather than recall. Structured observation of real work shows what happened at the bench. And a revised document is itself a signal, because a change to the source material can make training that was correct last quarter wrong today.
Read none of this as a verdict. A low score does not mean the course failed, or that an instructional designer got it wrong. It marks the part of a procedure that needs another pass, which is the only form the information can actually be used in. Measurement here exists to show what to improve.
What changes when the record exists
The visible change is that a readiness question stops being answered from memory. A supervisor asking whether a technician is ready for a particular avionics installation gets a view backed by dated evidence: what was covered in the course, what was demonstrated in a scenario, what was observed in the field, and which parts are stale because the manual changed underneath them.
The more useful change is upstream. Once you can see where people consistently slow down, the training itself gets specific: this module gets another pass, this procedure needs a scenario rather than a slide, this revision means a cohort needs re-practice. Continuous proficiency means exactly that in practice, and it is the same team producing better training because they can finally see which parts land.
Where this sits next to the systems you already run
Nothing here asks you to move off the systems that work. The LMS stays the record for completion and for the training documentation a repair station has to retain. The authoring tool stays where courses get built. Content comes in as it is, with SCORM 1.2 and LTI 1.3 in and out and xAPI to your LRS, so there is no content migration and completion records stay where they are.
What gets added is the join: a system of record for skill evidence that reads from those sources and holds person, skill, evidence and date in one place. That makes it a companion to the stack rather than a replacement for any part of it. If you are mapping this against what auditors and inspectors ask for, the AS9100 and FAA Part 145 reference guide covers what the standard and the regulation actually require of training records, and AS9100 training software covers the tooling side of the same problem.
Why this matters when you certify a distribution network
For an aerospace or defense manufacturerwith a distribution or service network, the argument is commercial before it is anything else. When you name an approved service partner, you are lending your reputation to work you do not directly supervise, and commercial expectations move faster than regulation does. A completion record from a partner’s training system is a thin thing to stake that on.
Evidence of demonstrated skill, dated and tied to the current revision of the procedure, is a materially stronger position, both for deciding which partners get which tier of work, and for the conversation after something goes wrong. See skill verification for how that plays out across a network, and the aerospace and compliance training statistics for the workforce numbers behind it, each cited to its source.
Where to start
Not with an integration program. Start with one course that already exists and one skill that matters and can be observed. Keep the learner experience as it is. Add one realistic scenario, and connect it to whatever evidence already exists in the real work.
For an aerospace maintenance team that might be one installation course and one troubleshooting scenario, with newer technicians the obvious place to begin. The first useful version answers a short list of questions: where do people get stuck, who has demonstrated the skill, what evidence supports that, and what should happen next. Each pass through the loop sharpens the record, which is what makes the next course better than the last one.
Frequently asked questions
The term comes from The Knowing-Doing Gap by Stanford professors Jeffrey Pfeffer and Robert Sutton (Harvard Business School Press, 2000), which examined why organizations that know what to do frequently fail to do it. Applied to training, it names the distance between a person having access to the correct procedure and that person carrying it out correctly under real conditions.
Four facts: that someone enrolled, that they finished, what they scored on the assessment attached to the course, and when that record expires. Those facts are necessary for training administration and, in a certificated repair station, for the records required under 14 CFR 145.163. None of them describe what the person did the next time the work came up.
No. AmpUp works alongside an existing LMS rather than instead of it: the LMS keeps delivering training and keeps the completion records where they already are. Content moves in as it is, with SCORM 1.2 and LTI 1.3 in and out and xAPI to your LRS, so there is no content migration.
A skills matrix records an assertion at a point in time: someone marked a person as qualified for a task. Skill evidence records what was observed, in what context, and when, which means it carries a date and can go stale as procedures change. The two answer different questions and a matrix is not a substitute for the underlying evidence.
No. Instructional designers and subject-matter experts define the skill, the evidence that counts, and the standard that has to be met. The system collects and organizes signals against that model and surfaces where practice is needed. The judgment about what good work looks like stays with the people who know the work.
From the training that already runs and the work that already happens: courses and assessments in the LMS, scenarios and role plays built from your own material, structured observation of real work, and changes to the source documents themselves. A revised service bulletin is a signal, because it can make previously correct training wrong.