The Second Job List: The Small Site Jobs That Never Reach the Programme
Every project has a second job list: the repair before the next trade arrives, the measurement nobody took, the material that has to be moved. Agreed by phone, forgotten by afternoon. Here is what it costs and how to record it.
A window opening needs measuring before the joiner arrives on Thursday. It is agreed in a corridor on Tuesday, between two other conversations, and nobody writes it down. The joiner arrives, the opening has not been measured, and he leaves again. Three weeks later, during handover, somebody finally asks why that window is still missing.
Nothing about that job was difficult. It took four minutes of work and cost a wasted trade visit, a delayed handover and an argument about who was responsible. It never appeared on any programme, in any report, or in any project cost — right up until it became all three.
Every project runs a second job list like this. It sits underneath the programme, it is entirely made of small things, and almost nobody records it.
Why the programme never sees these jobs
A construction programme plans phases. It says when the structure finishes, when the screed goes down, when the joiner is on site. It is a good instrument for exactly that, and a useless one for the work that happens between phases.
The second list is made of jobs too small to be a programme line and too consequential to forget:
- a repair that has to happen before the next trade can start,
- a measurement somebody needs to take on site,
- material that has to be moved from the delivery area,
- a defect noticed in passing that nobody logged,
- a check the client asked for during a site visit.
None of these belong in a Gantt chart. They are agreed verbally, they are usually done within a few days, and their entire lifespan is shorter than the interval between two programme updates. So they live in somebody’s head — and heads are where site work goes to be forgotten.
What the second list actually costs
The cost of an unrecorded small job is never the job itself. It is what happens around it.
That last row is the one that surprises people. Small extra jobs get done because saying yes is easier than arguing, and they get done well. They simply never reach an invoice, because no record exists that they were ever agreed. A firm that does forty of these a year is giving away a fortnight of work.
Four workarounds, all failing the same way
Every site already has a system for this. That is the problem — it has four of them, and none of them survive contact with a busy week.
The common failure is not discipline. It is that none of these produce a record anyone else can read, and none of them survive the person who made them being busy.
What a site task actually needs to carry
The instinct at this point is to reach for project management software, and it is the wrong instinct. A site task does not need dependencies, percentage completion or a Gantt bar. It needs five things, and adding a sixth is how these systems die.
Every one of those five earns its place. Drop the title and the record is useless. Drop the category and you cannot see patterns. Drop the owner and it belongs to everybody, which means nobody. Drop the done state and you have a diary rather than a list.
Drop the cost centre and you have a to-do list — useful for a week, invisible to the business.
Why the cost centre is the load-bearing field
A task list that is only a task list gets abandoned. It has no reader except the person who wrote it, and it competes with a notebook that is already in their pocket.
A task tied to a cost centre is something else entirely. It sits beside the hours booked to that cost centre and the materials issued to it, which means the small jobs stop being invisible. You can finally answer questions that had no answer before: how much repair work is this project generating? Which phase produces the most unplanned jobs? Is one site manager absorbing work that was never priced?
This is also what turns the second list into an estimating asset. A completed project leaves a record of the small work it actually took, not the small work somebody remembered. The next tender is priced from that.
Two states, not seven
Most task tools offer a rich set of states: in progress, blocked, waiting on client, ready for review, deferred. On a construction site, every one of them dies within a month.
The reason is simple. A state has to be maintained to mean anything, and maintaining a state costs attention that a site manager does not have at four in the afternoon. Within weeks, half the list is sitting in “in progress” — including jobs finished a fortnight ago and jobs nobody has started. A state that everything drifts into carries no information at all.
Open and done is a worse model of reality and a far better model of what people will actually keep accurate. A binary state is either true or obviously wrong, and obviously wrong gets corrected. This is a deliberate limitation, not a missing feature.
How to start on Monday
This is a small habit, not an implementation project. It works or fails in the first fortnight.
After a month you will have something no notebook produces: a record of how much small work a project really generates, attached to the parts of the project that paid for it.
Tasks is included in every RelayPlan plan, including Free. If you want to test whether your projects are carrying a second job list, that costs nothing to find out — record site tasks for one project for a month and look at what accumulated.
