What a Job Is
A job is the container hours are charged to. How big it should be, when to split it, and what happens when the container is the wrong size.
A job is the container that hours are charged to, and its size decides what can be learned afterwards. Too large and everything inside it is invisible; too small and nobody books accurately. Most costing problems that look like discipline problems are container problems.
The commercial issue in “What a Job Is” becomes easier to audit when estimated work, recorded hours and approved changes remain distinct. Teams considering using mouse jiggler detection in cost reviews for mouse jiggler detection can strengthen the project-time evidence without treating a software report as a substitute for the quote, contract or accepted invoice.
The right size is the smallest unit anybody makes a decision about.
Additional context for “What a Job Is” is available from Asana work-management resources. Current legal, payroll, tax and accounting requirements should be confirmed for the relevant jurisdiction.
Deciding whether to open one
Will anybody ever ask what this cost? If yes, it needs its own container. If no, it can live inside a larger one.
Will anybody ever ask what this cost? If yes, it needs its own container. If no, it can live inside a larger one.
A customer, a site, a phase, a product variant — each gets a job number only if somebody will ask about it separately.
Too large
A single job number for a year of work for one customer. Everything balances, nothing is learnable, and the only available conclusion is whether the customer as a whole was profitable.
Which is a real question, and it is the only one that can be answered.
Too small
Four hundred job codes, most of them touched a few times. People stop choosing and start guessing, usually by picking the first plausible entry or the one they used last.
The data is then precise and wrong, which is worse than coarse and right.
The number that works
In most operations, a list short enough that a person can hold it in mind — twenty to forty active codes for an individual, not for the organisation.
Structure beats length: a hierarchy where the person picks a customer and then a phase is easier than a flat list of the same size.
Phases
The usual reason to split a job: design, build, test, handover. Each behaves differently and each is estimated separately.
Where a job overruns, phase-level data says where. Job-level data says only that it did.
Splitting after the fact
Possible and unreliable. Hours booked to a combined code cannot be split afterwards except by asking people to remember, and they will remember to expectation.
Splitting decisions belong at the point the job is opened.
Standing jobs
Containers for recurring work that is not a project: support, internal maintenance, a retainer. They are legitimate and they should have owners and review dates.
A standing job with no owner becomes the catch-all, and the catch-all grows.
The catch-all
Every system has one. Watching its share is the best single measure of whether the job structure is working.
Above about ten per cent of hours, the structure is wrong and no amount of reminding will fix it.
Jobs with no price
Internal work, warranty, goodwill. They still need containers, because hours spent on them are hours not spent elsewhere.
What they do not have is a recovery mechanism, which means they are carried in the rate and should be visible when the rate is built.
Opening and closing
A job should have a date it opened and a date it closed, and after it closes nothing should be bookable to it.
Open jobs that never close are where rework and goodwill accumulate invisibly, and that has its own note here.
Who decides the structure
Whoever will read the results, with whoever has to book to it. Designed by finance alone it is too fine; designed by operations alone it is too coarse.
An hour's conversation at the start of each significant piece of work is the whole governance required.
Jobs that outlive their usefulness
A job opened for a three-month project and still accepting hours four years later. It has become a standing code without anybody deciding and nobody has looked at its total.
An annual sweep of jobs open longer than their expected duration finds these, and the conversation is usually short: close it and open something named for what the work has become.
The customer who is one job
Where a customer buys a stream of small work, a job per request is unworkable and a job per customer loses everything. A job per month per customer is usually the answer.
It gives a cost per period that can be compared, keeps the list short, and matches how the work is actually reviewed and invoiced.
Jobs that cross a year end
A job spanning two financial years will carry hours at two different rates and will be reported in two periods. Both are correct and both confuse anybody comparing the job's total to its estimate.
Recording the rate in force for each tranche of hours is what makes the comparison reconstructable afterwards, and it is a column rather than a process.
Reviewing the list
Annually: which codes were used, which were not, which absorbed more than expected. Close the unused ones and split the overloaded ones.
A job list that has only grown for five years is the commonest reason people say their time recording is useless.