Filled example: decision table coverage
The number in brackets is how many rules match. One is right; none is a gap; two or more is a clash. In the first draft C1 matched two rules and C5 to C8 matched none. The Excel check does this count for any three conditions.
| # | In the plan? | Within the limit? | Blocked dates? | First draft: rules that match | Final: rules that match |
|---|---|---|---|---|---|
| C1 | Yes | Yes | Yes | R1, R2 (2) | R2 (1) |
| C2 | Yes | Yes | No | R1 (1) | R1 (1) |
| C3 | Yes | No | Yes | R3 (1) | R3 (1) |
| C4 | Yes | No | No | R3 (1) | R3 (1) |
| C5 | No | Yes | Yes | None (0) | R4 (1) |
| C6 | No | Yes | No | None (0) | R4 (1) |
| C7 | No | No | Yes | None (0) | R4 (1) |
| C8 | No | No | No | None (0) | R4 (1) |
Choose a job aid format
Ask in this order: does the order matter, does the action depend on conditions, do people look something up? If none, use a checklist.
| Format | Use it when | Layout | Fictional example in this kit |
|---|---|---|---|
| Step card | The order matters and people skip or swap steps. | Numbered steps, one action each, with what you should see and what to do if not. | Set up the hybrid meeting room |
| Checklist | The order does not matter, but people forget an item. | Tick boxes, one check each, with where to look. | Check a customer quote before you send it |
| Decision table | The right action depends on two or three conditions. | One row per rule: the conditions, then the action. Every combination has exactly one rule. | Approve a course booking request |
| Quick reference | People look up a value, a code or a name, not a sequence. | A lookup table sorted the way people search, with what to do next. | Sort a returned item by its condition |
Aid or training? Nine questions
The rule: a yes to question 9 means fix the work first. No yes at all means no need found. No training signal means a job aid alone. A training signal with two or more job aid signals means a job aid plus practice. Anything else means training first. The rule is this kit’s working guide, not a standard.
| # | Question | A yes points to |
|---|---|---|
| 1 | Is the task done less than once a month by the same person, so the details fade? | Job aid |
| 2 | Does it have more than about five steps, values or conditions to remember? | Job aid |
| 3 | Do the steps, codes or limits change at least once a year? | Job aid |
| 4 | Can the person stop and read during the task, safely and without losing face? | Job aid |
| 5 | Must it be done from memory, because there is no time or no safe moment to read? | Training |
| 6 | Does it rely on judgement or a skill that improves only with practice? | Training |
| 7 | Is the person new to the basics, so the words on an aid would mean little to them? | Training |
| 8 | Would a first mistake be costly or unsafe, so people must practise before doing it for real? | Training |
| 9 | Do people who know the steps still get it wrong, because of the tool, the rules, the time or the workload? | Process fix |
Filled example: aid or training check
Six fictional tasks. Four lead to a job aid in a different format; the refusal call needs practice first; call logging needs a change to the system, not an aid.
| # | Question | A yes points to | Hybrid room | Quote | Booking | Returns | Refusal call | Call logging |
|---|---|---|---|---|---|---|---|---|
| 1 | Is the task done less than once a month by the same person, so the details fade? | Job aid | Yes | No | Yes | No | No | No |
| 2 | Does it have more than about five steps, values or conditions to remember? | Job aid | Yes | Yes | Yes | Yes | No | Yes |
| 3 | Do the steps, codes or limits change at least once a year? | Job aid | No | Yes | Yes | No | No | No |
| 4 | Can the person stop and read during the task, safely and without losing face? | Job aid | Yes | Yes | Yes | Yes | No | Yes |
| 5 | Must it be done from memory, because there is no time or no safe moment to read? | Training | No | No | No | No | Yes | No |
| 6 | Does it rely on judgement or a skill that improves only with practice? | Training | No | No | No | Yes | Yes | No |
| 7 | Is the person new to the basics, so the words on an aid would mean little to them? | Training | No | No | No | No | No | No |
| 8 | Would a first mistake be costly or unsafe, so people must practise before doing it for real? | Training | No | Yes | No | No | Yes | No |
| 9 | Do people who know the steps still get it wrong, because of the tool, the rules, the time or the workload? | Process fix | No | No | No | No | No | Yes |
| Job aid signals (yes to 1–4) | 3 | 3 | 4 | 2 | 0 | 2 | ||
| Training signals (yes to 5–8) | 0 | 1 | 0 | 1 | 3 | 0 | ||
| Result | Job aid alone | Job aid plus practice | Job aid alone | Job aid plus practice | Training first | Fix the work first | ||
| Format chosen | Step card | Checklist | Decision table | Quick reference | None | None |
Filled example: step card
Set up the hybrid meeting room for a call. Fictional room 3, version 2, after the user test. Owner: office manager.
| # | Do this | You should see | If not |
|---|---|---|---|
| 1 | Press the power button on the table panel. | The wall screen shows “Room 3” within 30 seconds. | Check that the wall switch by the door is on. |
| 2 | Plug your laptop into the cable marked LAPTOP. | Your screen appears on the wall screen. | Press the button marked SOURCE on the panel until it appears. |
| 3 | In the call, open Settings, then Devices. Choose “Room 3 camera” and “Room 3 speakerphone” for camera, microphone and speaker. | Your camera preview shows the whole table. | Unplug the USB cable marked CALL and plug it in again. |
| 4 | Ask one online person to say a sentence. | You hear them from the ceiling speaker. | Raise the volume on the table panel, not on the laptop. |
| 5 | Ask them: “Can you hear the room and see the whole table?” | They say yes to both. | Move the speakerphone to the centre of the table. |
| 6 | Turn on captions in the call and say that cameras are optional. | Captions appear on the wall screen. | Tell people how to turn on captions on their own screen. |
| 7 | After the meeting, press END on the panel and take the cable out. | The wall screen goes dark within a minute. | Report it on the room fault form. |
Blank step card
Add the task, who uses it and when, who to ask, the owner, the version and the date above the table.
| # | Do this | You should see | If not |
|---|---|---|---|
| 1 | |||
| 2 | |||
| 3 | |||
| 4 | |||
| 5 |
Filled example: checklist
Check a customer quote before you send it. Fictional sales team, version 1. The 10% limit and the five working days are this team’s rules.
| Done | Check | Where to look |
|---|---|---|
| ☐ | The customer name and address match the account. | Account page, Details tab |
| ☐ | Every line has a product code, a quantity and a unit price. | The quote lines |
| ☐ | The discount is 10% or less, or a manager’s note approves more. | Discount column; Notes field |
| ☐ | The delivery date is a working day, at least five working days from today. | Delivery field; team calendar |
| ☐ | The quote is valid for 30 days from today’s date. | Valid-until field |
| ☐ | The line totals add up to the subtotal, and the tax line is there. | Totals box at the foot of the quote |
| ☐ | The terms of sale file is attached. | Attachments list |
| ☐ | The file name is QUOTE, the customer name and the date. | Save dialogue |
Blank checklist
One check per line, with where to look.
| Done | Check | Where to look |
|---|---|---|
| ☐ | ||
| ☐ | ||
| ☐ | ||
| ☐ | ||
| ☐ |
Filled example: decision table
Approve a course booking request. Fictional learning team, version 1. If none of the rules fits, ask the learning lead.
| Rule | If 1: in the plan? | If 2: within the limit? | If 3: blocked dates? | Then do this |
|---|---|---|---|---|
| R1 | Yes | Yes | No | Book it. Tell the line manager the dates. |
| R2 | Yes | Yes | Yes | Ask the line manager to agree the dates. Book when they say yes. |
| R3 | Yes | No | Any | Send it to the budget holder with the plan reference. If the dates are blocked, the line manager agrees them too. |
| R4 | No | Any | Any | Do not book. Ask the person to add it to their development plan with their line manager, then send it again. |
Blank decision table
Write each condition as a yes-or-no question. Fill each condition cell with Yes, No or Any.
| Rule | If 1 | If 2 | If 3 | Then do this |
|---|---|---|---|---|
| R1 | ||||
| R2 | ||||
| R3 | ||||
| R4 |
First draft of the decision table
Fictional. The coverage check found the clash in R1 and R2 and the missing rule for a course outside the plan. The final version fixes both.
| Rule | If 1: in the plan? | If 2: within the limit? | If 3: blocked dates? | Then do this |
|---|---|---|---|---|
| R1 | Yes | Yes | Any | Book it. Tell the line manager the dates. |
| R2 | Yes | Yes | Yes | Ask the line manager to agree the dates. Book when they say yes. |
| R3 | Yes | No | Any | Send it to the budget holder with the plan reference. |
Filled example: quick reference
Sort a returned item by its condition. Fictional returns bench, version 1. The last row says what to do when nothing fits.
| Code | What you see | Put it here | Then |
|---|---|---|---|
| A | Unopened. The seal is intact. | Resale shelf, bay 1 | Refund in full. |
| B | Opened. All parts there, no marks. | Check bench, bay 2 | Refund in full after the bench check. |
| C | Opened. A part is missing. | Parts bin, bay 3 | Ask the returns lead before any refund. |
| D | Damaged in delivery: the outer box is crushed or torn. | Carrier claim cage | Take two photos, then fill in the carrier claim form. |
| E | Damaged by use: scratches, stains or wear. | Hold shelf, bay 4 | Pass it to the returns lead with the customer’s note. |
| F | Not what the customer ordered: we sent the wrong item. | Resale shelf, after a bench check | Send the right item today. No return label needed. |
| ? | You cannot tell which code fits. | Hold shelf, bay 4 | Write “?” on the label and ask the returns lead. Do not guess. |
Blank quick reference
Sort the rows the way people search. Keep a last row for “none of these”.
| Look up | What it means or what you see | Where or who | Then |
|---|---|---|---|
Filled example: user test log
Fictional, Wednesday 21 October 2026. Two of three testers finished without help; one error in total; the average time was 2.7 minutes.
| Test | Tester (role) | Aid version | Done without help? | Minutes | Errors | Where they slowed down | What we changed |
|---|---|---|---|---|---|---|---|
| 1 | Office assistant, new to the building | v1 | No | 3.5 | 1 | Step 3: could not find where to choose the room camera. Stopped at time. | Step 3 now names the menus: Settings, then Devices. |
| 2 | Sales representative from another floor | v2 | Yes | 2.5 | 0 | Step 2: looked for the SOURCE button for a few seconds. | Nothing. The pause was short and the step worked. |
| 3 | Trainer who had never used room 3 | v2 | Yes | 2.0 | 0 | Nowhere. | Nothing. Ready: two testers in a row on v2. |
One-page job aid checklist
Read it before the user test. It covers the page, print and screen-reader access.
| Point | Check |
|---|---|
| Title | The title is the task, starting with a verb: “Set up the hybrid meeting room”. |
| Who and when | One line says who uses it and at which moment. |
| One page | It fits one side of A4 or Letter. If it does not, split the task or cut the explanation. |
| One action per line | Each step or check starts with a verb and holds one action. |
| What success looks like | Steps say what the person should see, so they know the step worked. |
| Words from the workplace | Buttons, menus and labels are written exactly as they appear at work. |
| Readable type | At least 12 point in print, high contrast, no text over pictures. |
| Not colour alone | Colour is never the only signal. Every coloured mark also has a word or a symbol. |
| Plain text copy | A plain text or tagged version exists for screen readers and for phones. |
| Stop and ask | It names who to ask when the aid does not fit, and what to do if nobody answers. |
| Owner and date | A named owner, a version number and a last-checked date are on the page. |
| Tested | Two testers in a row finished the task with it, without help and without errors. |
Five-minute user test script
Five minutes per tester. Test with someone who has not done the task recently. The ready rule is this kit’s working rule, not a standard.
| Time | What you do | Words to say |
|---|---|---|
| 0:00–0:30 | Give the tester the aid and the task. Nothing else: no demonstration, no tips. | “We are testing the card, not you. Please do the task with only this card. Say what you are thinking if you can. I will stay quiet unless you are stuck for a minute.” |
| 0:30–4:00 | Watch and write. Note the start time, each pause of more than a few seconds, each error and each question, with the step number. Do not help. If the task needs longer, test the first part only. | If the tester is stuck for a minute: “What would you do now if I were not here?” Then help, and write “No” under done without help. |
| 4:00–5:00 | Ask two questions and write the answers in the tester’s words. | “Where did you slow down?” “Which word would you change?” |
| After the test | Change the aid before the next tester and give it a new version number. The aid is ready when two testers in a row, on the same version, finish without help and without errors. | “Thank you. Your pause at step 3 changed the card.” |
Filled example: job aid register
Fictional, as the register stood on Monday 11 January 2027. One aid is 27 days overdue for review and three are due within 30 days. The Excel register calculates the next review date and the days to it.
| # | Job aid | Format | Owner | Version | Last checked | Review every (months) | Next review | Days to review on 11 Jan 2027 | Where it lives |
|---|---|---|---|---|---|---|---|---|---|
| J1 | Hybrid room set-up | Step card | Office manager | v2 | Wed 21 Oct 2026 | 3 | Thu 21 Jan 2027 | 10 | On the table in room 3; plain text on the intranet |
| J2 | Quote check before sending | Checklist | Sales team lead | v1 | Thu 15 Oct 2026 | 2 | Tue 15 Dec 2026 | -27 | Pinned in the sales channel; printed at each desk |
| J3 | Course booking approval | Decision table | Learning coordinator | v1 | Mon 5 Oct 2026 | 4 | Fri 5 Feb 2027 | 25 | Learning team page, linked from the booking form |
| J4 | Return condition codes | Quick reference | Returns lead | v1 | Tue 27 Oct 2026 | 3 | Wed 27 Jan 2027 | 16 | On the wall above the check bench, at eye level |
What you'll leave with
In 60 minutes a trainer and a subject expert decide whether a task needs a job aid, training or a process fix, draft a one-page aid, test it with two users and leave with a revised version that has an owner and a review date.
Before the session
- For trainers, learning designers and team leads, working with one or two people who do the task well.
- Collect the current instructions, error reports or the questions people ask most about the task.
- Book two testers who have not done the task recently, and a safe practice version of a risky task.
- Accessibility: send plain text for screen readers, let testers stop at any time, never use colour alone.
How to fill it in with your team
Name the task, the user and the moment
Ask the subject expert to describe the task in one sentence that starts with a verb. Then say: “Who does this, where are they, and what goes wrong today?” Write one line on top of the check: “When [user] is [doing what, where], they need to [result] without [problem].” If the expert names three tasks, choose the one that goes wrong most often and park the others.
Aid or training? Nine questions
Read the nine questions aloud one by one. The expert answers yes or no; you write. Two minutes for questions 1 to 4, three minutes for questions 5 to 8, one minute for question 9, then two minutes to count and read the result. Say: “A job aid helps people who can already do the task but forget details. It does not teach a skill.” If the result is “Fix the work first”, stop and say: “The steps are known. What gets in the way is [the tool, the rule, the time]. Who owns that?” If the result is “Training first”, plan practice with the course outline template, and come back to an aid afterwards.
Choose the format
Show the four formats. Ask three questions in this order. “Does the order matter?” If yes, use a step card. “Does the right action depend on conditions?” If yes, use a decision table. “Do people look something up, such as a code or a name?” If yes, use a quick reference. Otherwise, use a checklist. If two formats fit, choose the one that matches the most frequent mistake, and write the choice in the last row of the check.
Draft the aid on one page
Five minutes: the expert does or describes the task, step by step, and you write every action, value and condition on the blank format, without editing. Ten minutes: rewrite each line so it starts with a verb and holds one action, and add what the person should see and what to do if not. Use the words on the buttons and labels at work. Five minutes: read the design checklist together and cut everything that explains rather than instructs. Say: “Would someone doing the task need this line at that moment? If not, it goes.” Add the stop-and-ask line, the owner and the version: v1.
Five-minute user test, two testers
Five minutes with the first tester, following the test script: thirty seconds to hand over the aid and the words, three and a half minutes to watch and write, one minute for two questions. The expert watches in silence. Two minutes between testers: change the words where the tester slowed down or made an error, and write v2 on the aid if you changed it. Five minutes with the second tester. Say to each tester first: “We are testing the card, not you.” A tester may stop at any time.
Revise and hand over
Six minutes: read the test log. Fix every step where a tester needed help or made an error, and decide whether the aid is ready: two testers in a row, on the same version, without help and without errors. If it is not ready, name who tests it with whom and by when. Four minutes: fill in a row of the review register: owner, version, last-checked date, review interval and where the aid lives. Ask the owner: “Is this yours, and is the review date right?”
Participant worksheet
Copy this table into a document, or print the worksheet from the bundle. Share only what you choose.
| Item | Detail |
|---|---|
| Aid or training? check | Nine yes-or-no questions for up to three tasks, counted into one of five results |
| Four one-page formats | Step card, checklist, decision table and quick reference, each with a title, a who-and-when line and an owner |
| Decision table coverage | Every combination of three conditions, with the rules that match it |
| Five-minute user test | The tester, the version, done without help, minutes, errors, where they slowed down and what changed |
| Review register | Each aid with its owner, version, last-checked date, review interval and next review date |
Debrief questions
- Which line on the aid did the testers never read, and can it go?
- Where did a tester slow down, and was the cause a word, a missing step or the task itself?
- Which task in the check came out as “Fix the work first”, and who owns that fix?
- What will tell us, a month from now, that people use the aid: fewer questions, fewer errors or fewer returns to the expert?
Adapt it for your group
- Working alone with one expert: keep all six steps, but do the user test with one tester today and a second one within a week. Do not call the aid ready after one tester.
- Short version, 30 minutes: skip the format discussion if the check makes the format obvious, draft in 12 minutes and test with one tester for five. Record the second test in the log later.
- Online: share the blank format in a document both people can edit. For a screen task, the tester shares their screen and the observer keeps the camera off. Keep captions on and cameras optional.
- For testers who prefer not to think aloud: let them work in silence, and ask the two questions at the end in writing. A tester may stop at any time, and nobody’s name goes in the log, only their role.
- For a team with many aids: run the check for six tasks at once in the Excel version, build the aids with the most yes answers to questions 1 to 4 first, and keep one register for all of them.
After the session
Put the aid where the task happens: on the table, at the bench, or one click from the screen. A week after release, ask two users one question: “When did you last use it, and where did it not help?” Note what changes in the register. On the review date the owner checks every line against the real task and the current system, then updates the version and the date. If people stop using the aid because they know the task, that is fine: retire it and note that in the register.
Questions people ask
What is a job aid template?
A job aid template is a ready layout for a short reference that people use while they do a task: a numbered step card, a checklist, a decision table or a quick reference table. It holds what is easy to forget, not what needs to be learnt. This kit gives you each of the four as a blank page and a filled example.
When should you use a job aid instead of training?
When people can already do the task but forget steps, values or rules, and can stop and read during the task. Use practice when the task must be done from memory, relies on judgement, or is unsafe to get wrong the first time. If people know the steps and still get it wrong, the problem is often the tool or the workload, and neither an aid nor training will fix it. The nine-question check sorts these cases.
Which job aid format should I choose?
A step card when the order matters, a decision table when the action depends on conditions, a quick reference when people look up a code or a name, and a checklist when the order does not matter but items get forgotten. If two fit, choose the one that matches the most frequent mistake.
How is this different from the facilitator guide template?
A facilitator guide is for the trainer: what to say and do in a session. A job aid is for the person at work, after or instead of the session, at the moment they do the task. If the check says “Training first”, plan the practice with the course outline template; the job aid often follows as the reminder people take back to work.
How do you test a job aid?
Give it to someone who has not done the task recently and ask them to do the task with only the aid, while you watch in silence for about three and a half minutes. Note where they slow down, each error and each question, then ask where they slowed down and which word they would change. Fix the aid before the next tester. In this kit, an aid is ready when two testers in a row finish without help and without errors.
What is included in the free and editable files?
The open PDF has the aid-or-training check, the four blank formats, the user test log, filled examples of each, the decision table coverage check and the 60-minute build session. Subscribers also receive the editable Word file, an Excel workbook with the check, a decision table coverage check, a user test log and a review register that all calculate, and a CSV of the nine questions. The aids are workplace reminders. They are not a controlled procedure, a safety instruction or a legal record.
How to use the files
Print the PDF, edit the Word file in Word or Google Docs, or open the CSV or XLSX in a spreadsheet. Open files need no email.
Can I use this with paying clients?
Yes. Use and adapt this original Superglue kit in your own training, including paid client sessions. Please share this page rather than redistributing the unmodified kit. Read the full usage terms.