River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Job Task Analysis Template for Training

Five documents and three sheets that score every task twice, because adding frequency into a priority number buries the tasks nobody gets to practise.

Free download  ·  No account needed

Task Register

Role:

One row per task. The three ratings stay in their own columns and the last two columns are worked out twice, once by the weighted sum and once by the decision table.

CellFCDScoreComposite saysTable saysReason
T-01
T-02
T-03
T-04
T-05
T-06
T-07
T-08
T-09
T-10
T-11
T-12

Where the two answers part company

RowWhat it countsValue
D-1Tasks the composite bands into a training band
D-2Tasks the table sends to a course
D-3Needs a course, composite buried it
D-4Needs no course, composite promoted it
D-5Rows a statutory mandate settles

Frequency runs the other way here. The more often a task is performed, the less has to be built for it, because the shift is already the delivery mechanism.

Every scored job task analysis template collects the same three ratings and then adds them up. Frequency, criticality and difficulty go into one weighted number, the number gets banded, and the bands get sorted descending. The frequency weight is always positive, so a task done every shift rises and a task done twice a year sinks. Federal guidance on job analysis asks instead for work behaviours and their relative importance, which is a different column. This pack scores the register twice and keeps both answers.

Calderwell Water runs one surface water plant and is deciding what to build for its Water Treatment Plant Operator role. 32 tasks came off the procedures and two shifts of observation, rated for frequency, criticality and difficulty, and rolled into 9 competencies. On its own band labels the composite asks for a course on 28 of the 32 and reaches nothing at all for none of them. The decision table asks for 14. The two agree on 21 rows.

The 11 rows they disagree on are the ones worth arguing about. 6 need a course and the composite buried them, including locking out a motor before work on it at rank 16. 5 need no course and it promoted them, led by a filter backwash at rank 3 that outranks 12 of the 14 real courses. 5,151 weightings were searched and none reproduce the table, so this is not a tuning problem. What gets built goes to the programme design.

32 tasks, two scores, and the 11 rows they disagree on

The register scored both ways, every disagreement with its reason, the weight search, and the build list that comes out.

Task Register

32 tasks, scored twice, agreeing on 21

Calderwell Water, Water Treatment Plant Operator, sorted by the composite it is usually sorted by.

CellFCDScoreRankCompositeTable
T-06454901Band 1Course
T-07454902Band 1Course
T-17 *553903Band 1On job
T-29255854Band 1Course
T-01 *551805Band 1On job
T-02 *551806Band 1On job
T-08353807Band 1Course
T-10254808Band 1Course
T-11 *452809Band 1On job
T-12 *3538010Band 1Job aid
T-261558011Band 1Course
T-271558012Band 1Course
T-281558013Band 1Course
T-18 *3447514Band 2Course
T-19 *2457515Band 2Course
T-22 *3527516Band 2Course
T-24 *1547517Band 2Course
T-25 *1547518Band 2Course
T-315427519Band 2On job
T-053437020Band 2Job aid
T-131537021Band 2Job aid
T-141537022Band 2Job aid
T-32 *1537023Band 2Course
T-094336524Band 2On job
T-152436525Band 2Job aid
T-204336526Band 2On job
T-232436527Band 2Job aid
T-035316028Band 2On job
T-045215029Band 3Nothing
T-161335030Band 3Job aid
T-305215031Band 3Nothing
T-213224532Band 3Nothing

The starred rows are the 11 the two methods score differently. Read against its own band labels the composite wants a course on 28 of 32 tasks and nothing at all for none, which is why the ranking rather than the band is what people actually use.

Task Analysis Report

6 buried, 5 promoted

Every row where the weighted sum and the decision table give a different answer, with the reason the table gives.

Needs a course, the composite buried it

CellTaskFCDRankWhy a course
T-18Take a filter offline and return it to service34414Hard enough that the errors need correcting under supervision
T-19Run a filter profile and read the turbidity trace24515Hard and rare, so there is no practice to fall back on
T-22Lock out and tag out a motor before work on it35216Lockout training is required before first assignment
T-24Enter the clearwell as an authorised entrant15417Confined space training is required before first assignment
T-25Don and work in self-contained breathing apparatus15418Respirator training is required before first assignment
T-32Stand attendant on a permit space entry15323Confined space training is required before first assignment

Needs no course, the composite promoted it

CellTaskFCDRankWhat it needs instead
T-17Backwash a filter and verify the ripening sequence5533On-job practice
T-01Read and log filtered water turbidity every four hours5515On-job practice
T-02Record the day's lowest residual disinfectant value5516On-job practice
T-11Collect routine distribution coliform samples to the plan4529On-job practice
T-12Compile the monthly operating report for the state35310Job aid

Reading the ranking instead of the bands

RowWhat it countsValue
P-1Courses the table asks for14
P-2Right, in the composite's own top 149
P-3Rank you must read down to for the last course23 of 32
P-4Tasks needing nothing built above that point9

Backwash a filter and verify the ripening sequence is rank 3 on the whole register and needs no course at all. It outranks 12 of the 14 tasks that do, because it is done every shift and difficulty is the column that only gets a quarter of the weight.

How the Register Is Scored

5,151 weightings, none of them reproduce it

The weight simplex swept at a hundredth of a point, each weighting tested against every threshold its own scores admit.

RowWhat was searchedResult
W-1Weightings tried5,151
W-2Best any of them managed29 of 32 rows
W-3A weighting that reproduces the tablenone
W-4Weightings tied for best671
W-5Frequency weight across all of them0.00 to 0.19
W-6Frequency weight the published formula uses0.25

The pair that settles the direction

CellTaskFCDRankTable says
T-09Make up a polymer batch to the target concentration43324On-job practice
T-16Produce the annual consumer confidence report13330Job aid

Same criticality, same difficulty, frequency the only difference. The table writes a checklist for the rarer one and nothing for the one done every day, so more frequency means less built. The composite ranks the frequent one 6 places higher.

The rows no weighting ever gets right

CellTaskWhat settles it
T-08Change out a one-tonne chlorine cylinder29 CFR 1910.134(k), respirator training, recurring annually
T-22Lock out and tag out a motor before work on it29 CFR 1910.147(c)(7), energy control training for authorised employees
T-32Stand attendant on a permit space entry29 CFR 1910.146(g)(2)(i), before first assignment

All 3 are statutory mandates, which is a date rather than a rating. No arithmetic over the three columns can see them, which is why they sit above the table rather than inside it.

Training Priority

6 courses, 14 of 32 tasks inside them

A competency inherits the most demanding verdict among its tasks, never the average of them. Mandated courses first, because a mandate carries a date.

CellCompetencyTasksMeanMean bands asDriverVerdict
K-1Routine monitoring and shift handover569.0Band 2T-31On job
K-2Filter operation and performance interpretation575.0Band 2T-19Course
K-3Chemical dosing and process control481.2Band 1T-06Course
K-4Chlorine handling and gas safety378.3Band 2T-26Course
K-5Regulatory sampling371.7Band 2T-11On job
K-6Compliance reporting and notification366.7Band 2T-12Job aid
K-7Mechanical maintenance and isolation458.8Band 3T-22Course
K-8Confined space and permit work272.5Band 2T-24Course
K-9Abnormal and emergency operations381.7Band 1T-29Course

4 competencies need a course their own mean score would have missed, and 3 band into a training band needing none. Mechanical maintenance and isolation is carried by one task out of 4.

What to build

CellCourseTasksMandatedDue
K-4Chlorine handling and gas safety33On a date
K-7Mechanical maintenance and isolation11On a date
K-8Confined space and permit work22On a date
K-2Filter operation and performance interpretation2Before next intake
K-3Chemical dosing and process control3Before next intake
K-9Abnormal and emergency operations3Before next intake

And what is deliberately not built

RowInstead of a courseTasks
B-1Structured practice on shift against a sign-off8
B-2A checklist and a spot check on the next one7
B-3Nothing at all, criticality below the gate3

14 of 32 tasks need a course, which is 44% of the register. The other 18 rows are the reason the analysis was worth doing at all.

What's in the pack

01

Task Register

One row per task with the three ratings kept apart, the composite, the rank it gives, the table's verdict, and the reason on every row.

02

Competency Matrix

Tasks grouped into competencies, each with three indicators written so that two people watching the same shift would record the same verdict.

03

Training Priority

The build list and everything left alone, with the mandated courses carrying a date and the rest due before the next intake.

04

Task Analysis Report

The register read as findings: where the two scores part company, which rows a mandate settles, and what the observation could not see.

05

Competency Definitions

Each competency, the tasks inside it, the verdict its hardest task forced, and the indicator an assessor marks against later.

06

Training Recommendation

What to build, in lead-time order, plus the job aids and the on-shift sign-offs that replace the courses nobody needs.

07

How the Register Is Scored

The decision table written out with its gates, the weight search, and the three things the method deliberately refuses to settle.

08

Observation and Interview Guide

What to watch for on shift and what to ask afterwards, built so a difficulty rating comes from a stumble somebody saw.

How to use it

  1. 1

    Open it in River, or download it

    Open the pack in River and have the agent score your own register both ways, or download the eight blank Word and CSV files with no account.

  2. 2

    Send the role, not a job description

    Procedures, work orders, incident reports, the shift log, and whatever an experienced holder of the role says when asked to walk through a day.

  3. 3

    Rate the three columns separately

    Frequency is how often, criticality is what breaks, difficulty is how long somebody takes to get reliable. Averaging them is the thing this avoids.

  4. 4

    Read the disagreement column first

    A row where the composite and the table part company is either a wrong rating or a task the usual arithmetic gets backwards. Both are worth twenty minutes.

Frequently asked questions

Is this template free?

Yes. Download all eight files as Word documents and CSV sheets with no signup and no card. Edit with AI is the optional path, where the agent scores your own register both ways and writes the reason on every row. The rest of the packs sit in the template library.

What format are the downloaded files?

Word documents (.docx) for the report, the competency definitions, the recommendation, the scoring explainer and the observation guide, and CSV (.csv) for the three sheets, zipped together. They open in Word, Pages, Google Docs, Excel, Numbers and Sheets. In River the same content opens as live Docs and Sheets.

What is actually wrong with weighting the three ratings?

The frequency weight points the wrong way. A task done every shift is already being taught by the shift, so the more often it happens the less has to be produced for it. Two tasks on the worked register share a criticality and a difficulty and differ only in frequency, and the table builds a checklist for the rarer one.

Could a better set of weights fix it?

No, and the pack shows the search rather than asserting it. 5,151 weightings were tried at a hundredth of a point, against every threshold each, and the best managed 29 of 32 rows. Across all 671 that tied for best the frequency weight never went above 0.19.

What happens to a task with a statutory training mandate?

It is settled before any rating is read, and the register records which rule settled it. Confined space entry needs training before the employee is first assigned duties, which is a date rather than a priority. Three of the worked register's rows are the ones no weighting ever gets right.

Is a competency the same thing as a learning outcome?

No, and conflating them is why competency lists arrive unusable. A competency is a thing a role does, held by whoever does the job. An outcome belongs to a course and is written to be assessed, so if you need those, write them separately.

The content lives in one person's head. Where does that fit?

Before this. The register needs tasks and ratings, and an experienced role holder is usually the only source of both, so start from the interview. The indicators here then give an assessor something to mark against, which is where an anchored rubric comes in.

Find out how little of the role needs a course

Download the blank pack as Word and CSV files, or open it in River and have your own register scored both ways, with the reason written on every row.

Edit with AI