Job aids are the right follow-up after an instructor-led training (ILT) course when the task is procedural, rare or easy to get wrong. eLearning is the right follow-up when people must build a skill, practice judgment or show proof of completion. Most courses need some of each, task by task.
This post gives trainers and instructional designers a scored questionnaire for deciding task by task, a worked example from an invented scenario, a format guide, and a straight answer on tracking, which is where job aids are weakest.
Key takeaways
- Decide per task, not per course. One classroom course usually holds tasks that need an aid, tasks that need practice, and tasks that need both.
- A job aid beats eLearning when memory is the weak link: rare, step-based, detailed or high-stakes tasks that people can look up while working.
- eLearning or more practice is needed when learners lack the underlying skill, must act from memory, or someone must prove who completed what.
- You can rarely track job aid use per person. Count what you can, measure errors and rework instead, and pair the aid with a tracked check when proof is required.
What is a job aid, and how is it different from eLearning?
A job aid is something a person consults while doing a task so they do not have to remember it: a checklist, step list, decision table or reference card. eLearning builds knowledge and skill before the task. A job aid supports performance during the task. They solve different problems, so the choice is rarely either-or.
The University of San Diego’s online degrees site describes job aids as tools or guides that give simple instruction on completing a task, and it lists five common formats: step-by-step manuals, flowcharts, decision tables, checklists and reference guides. A well-known handbook, Rossett and Gautier-Downes, A Handbook of Job Aids (Wiley, 1991), names seven: step job aids, worksheets, arrays, decision tables, flow charts, checklists and combination aids.
The practical difference is where the knowledge lives. After eLearning it is in the learner’s head, where you can test it. With a job aid it is in the document, which is why it is cheaper to update: you change one file, not a course package.
When does a job aid beat eLearning?
A job aid beats eLearning when the task is procedural, infrequent, error-prone or reference-like, and the person can look at something while they work. In those cases memory is the weak link, and no module fixes memory. Put the steps where the work happens instead.
In a 2017 ATD blog post, Patti Shank argues that limited human memory is the main reason to build performance support. She lists five situations where it pays off: tasks done infrequently, information that changes, cases where training would be overkill, content too detailed for training alone to stick, and tasks where poor performance has severe consequences. Those five map closely onto the questions in the framework below.
Checklists have a well-known published example. A 2009 study in the New England Journal of Medicine introduced a surgical safety checklist in eight hospitals. Death rates fell from 1.5% to 0.8% and inpatient complications from 11.0% to 7.0%. Treat that as encouraging, not proof for your context: it compared patients before and after the change, and the checklist was a team routine, not a lone document.
If the procedure came from a written document, turning an SOP into training shows how to split it between a course and an aid.
When do you still need eLearning after a classroom course?
Build eLearning, or more practice, when the learner lacks a skill, must decide under changing conditions, must act from memory, or when someone needs a record of completion. A job aid cannot teach a concept from nothing, and it usually cannot show who used it.
Shank makes the same split: training makes the most sense when people do not yet have the skills and must gain them. A decision table cannot teach someone to read a client’s mood, and a checklist will not help a learner who does not see why a step matters.
Three cases come up often after ILT. For spaced reinforcement, see microlearning follow-ups after ILT. If learners arrive without the basics, pre-work eLearning belongs before the class. If the whole course must move online, converting ILT slides into eLearning is a different project from building an aid.
How do you decide between a job aid and eLearning?
Answer three gate questions, then six scoring questions, for one task at a time. A “yes” on any gate means a job aid alone is not enough. Otherwise the score tells you whether to build an aid, an aid plus practice, or training. The cutoffs are rules of thumb, not validated research findings, so adjust them to your learners.
Step 1: Do any gates apply?
| Gate question | If yes |
|---|---|
| G1. Must the person act from memory, with no time to look anything up (a live conversation, an emergency)? | Build practice or eLearning. An aid can still support preparation. |
| G2. Does a client, regulator or contract require proof that each person completed or passed something? | Add a tracked module or assessment. The aid cannot provide the record. |
| G3. Could a learner still not do the task even with a perfect aid in hand, because they lack the underlying skill or concept? | Teach it first with ILT practice or eLearning. Add the aid afterward. |
Step 2: How many of these six are true?
Give one point for each “yes”. Answer for the task as the learner will meet it on the job, not as it was taught in class.
| # | Question | Why a yes points to a job aid |
|---|---|---|
| 1 | Is the task procedural: fixed steps, set values or a clear if-then rule? | Steps can be written down and followed. |
| 2 | Do people do it once a month or less often? | Rare tasks fade from memory between uses. |
| 3 | Is a mistake costly or hard to undo? | Error-prone, high-stakes work benefits from a prompt at the point of action. |
| 4 | Do the steps, values or rules change a few times a year? | One document is faster to update than a course. |
| 5 | Is the right answer the same in nearly every case, with few exceptions? | Exceptions and judgment need practice, not lookup. |
| 6 | Is there more detail than anyone could reasonably remember (codes, settings, long lists)? | Detail-heavy content is hard for training alone to make stick. |
Read the total like this:
- 5 or 6 points: build a job aid. A course for this task is probably overkill.
- 3 or 4 points: build a job aid plus a short practice activity, such as a scenario where learners must use the aid.
- 0 to 2 points: this is a skill or judgment task. Build training and practice, and add an aid only for the lookup parts.
If any gate said yes, take the more cautious result: the gate adds eLearning or practice to whatever the score suggests.
What does the framework decide in a worked example?
Run the questionnaire once per follow-up task, and one course usually splits into different deliverables, as the four tasks below show. This is an illustrative example, not a case study: the company, tasks and scores are invented to demonstrate the method.
Imagine a training provider has just delivered a one-day ILT course for new team leaders at a logistics company. The client wants on-the-job support for four tasks, and the provider scores each one.
| Task | Gates | Score (of 6) | What to build |
|---|---|---|---|
| A. Correct a timesheet in the HR system | None | 6 (steps, monthly or less, payroll errors, screens change, one right answer, many codes) | Job aid alone: annotated screenshots in step order, reviewed whenever the system updates. |
| B. File a workplace incident report | G2: the client’s insurer wants proof every leader was trained | 4 (steps, rare, costly, same every time; no change, little detail) | Checklist job aid plus a short tracked module with a knowledge check that records completion. |
| C. Hold a difficult performance conversation | G1 and G3: live, and the skill is new | 2 (rare and costly; not procedural, many exceptions) | Practice, not an aid: ILT role-play or a branching scenario. At most a one-page preparation checklist. |
| D. Build the weekly shift schedule in the scheduling tool | None | 3 (procedural, tool screens change, one right answer; weekly, low stakes) | Quick-reference card plus a short hands-on practice. Weekly use means memory will take over. |
One course produced four different answers. Only task A is a pure job aid. In task B the aid does the work and a small tracked component supplies the evidence the client needs. Task C shows what no aid can do. Task D is the middle case: people will learn a weekly task by repetition, so a card for the first month and a practice step in class are enough.
Pro tip: Score With Two People Who Do the Task. Ask two people who actually perform the task, ideally one experienced and one new, to answer the six questions separately. Frequency and exceptions often differ by role, so a task can be “weekly” for one team and “twice a year” for another. Where their answers disagree, you may need two aids, or one aid for newcomers only.
Which job aid format fits the task?
Match the format to the shape of the task. Fixed steps need a step list or annotated screenshots, yes-or-no branches need a flowchart or decision tree, several variables need a decision table, unordered must-dos need a checklist, and lookup facts need a quick-reference card.
| Format | Use it when the task… | Example after an ILT course |
|---|---|---|
| Checklist | Has several must-dos, order is flexible, and omissions are costly | Pre-shift equipment check |
| Step list with annotated screenshots | Is a fixed sequence in software | Correcting a timesheet |
| Decision tree (flowchart) | Branches on yes-or-no questions | Which escalation path applies to a complaint |
| Decision table | Depends on several variables at once | Which form applies by region and contract type |
| Quick-reference card | Needs facts looked up, not steps followed | Product codes, contact numbers, limits |
Choose the medium by where the work happens: a laminated card on a warehouse floor, a searchable page or PDF at a desk, a phone-friendly page for people away from a computer. Give screenshots and diagrams a short text version so the aid still works for screen reader users.
How do you make a job aid people actually use?
Build one aid per task, from watching the task done, and test it with someone who was not in your class. Keep it to one page or one screen, write steps as short verbs, and put it where the work happens. An aid nobody can find is the same as no aid.
- Pick one task and one audience. Use the questionnaire result and name the role.
- Observe the task and talk to the people who do it. Note step order, common slips and expert shortcuts, as the university resource above advises.
- Choose the format and medium from the table above.
- Draft in action verbs. One action per step, exact names for buttons and fields, and a warning placed before the step where the slip happens.
- Pilot with a novice. Ask someone who missed the class to complete the task using only the aid. Every question they ask is a gap in the draft.
- Assign an owner and a review date. Without one, the aid decays quietly as the procedure changes.
How do you track whether a job aid is used?
Honestly, you mostly cannot track it per person. A printed card records nothing, and even a digital page records opens, not correct use. Count what you can, measure outcomes like errors and rework instead, and add a tracked check whenever someone needs proof.
Clients often expect completion reports, so be clear about the options, from weakest to strongest evidence:
- Access counts. A digital aid in a shared folder, intranet or LMS resource area can show views. A unique QR code or short link per aid lets you count scans on printed copies. Views show reach, not use.
- xAPI statements. If the aid is a web page, a developer can send “opened” statements to a learning record store, because xAPI does not require an LMS and can record activity from any enabled device. This takes build effort and still records opens, not accuracy.
- Feedback. A “Did this help?” link or a manager check-in question shows whether people find and trust it.
- Outcome measures. Compare error rates, rework, support tickets or time to complete the task before and after. This is the nearest thing to evidence that the aid works, though other changes can affect the numbers too.
- A tracked companion. When a client needs proof, add a short assessment or sign-off. The aid supports the work, and the assessment supplies the record, as in task B above.
“We cannot tell you who opened the card, but we can show the error rate dropped” is a more honest report than a view count presented as proof.
Pro tip: Stamp Every Aid With a Version and a Date. Printed copies stay on walls long after a procedure changes. Add a version number, a date and a contact for corrections in the footer. When an error appears, the first question becomes which version the person was holding, which is something you can actually answer.
What should you do next?
Take one course you have already delivered and list the five to ten tasks learners must perform afterward. Run the gates and six questions on each, and build the aid for the highest-scoring task first. Pilot it with one new person before you roll it out.
For tasks in the middle or low bands, plan practice alongside it. Blended learning design covers what moves online and what stays in the room.
If you run classroom courses and want to see a training management platform built for training businesses, the SimpliTrain page for training companies describes what it covers. Disclosure: SimpliTrain is made by the same company that publishes this site. You do not need it to create or share a job aid, since a shared folder, a printed card or your existing LMS works.
Frequently asked questions
What is a job aid in training?
A job aid is a checklist, step list, decision table, flowchart or reference card that a person uses while doing a task, so they do not need to memorize it. It supports performance on the job rather than teaching a skill from scratch.
Can a job aid replace eLearning?
Sometimes. For a rare, step-based task that people can look up while working, a job aid is often enough. It cannot replace eLearning or practice when learners lack a skill, must act from memory, or when someone needs a completion record.
What is the difference between a job aid and an SOP?
An SOP documents the full procedure, including purpose, scope and responsibilities. A job aid is a short extract designed to be read while working. Many SOPs are better delivered as job aids than as courses, as explained in our guide to SOP to training.
What are examples of job aids?
Examples include a checklist for an equipment check, annotated screenshots of a software task, a flowchart for choosing an escalation path, a decision table for choosing a form, and a quick-reference card of product codes. Pick the format by the shape of the task.
How do you know if a job aid is working?
Measure outcomes, not just views. Compare error rates, rework, support questions or time on task before and after release, and ask users whether it helped. Pilot it with a newcomer before launch, because their questions show where the aid is unclear.
Do you need an LMS to deliver job aids?
No. A shared folder, intranet page, printed card or QR code can all work. An LMS helps if you want to count opens or place the aid next to a course, but it cannot show whether someone followed the aid correctly.
