When you know each project's margin in real time, you can make better decisions before it is too late.
Project cost tracking usually fails because the cost is recorded when it is invoiced rather than when it is decided.
When a site manager orders materials on Wednesday, the cost was created on Wednesday. The invoice arrives three weeks later and lands on the project then. For those three weeks the project looks cheaper than it is.
For one order the gap is not dangerous. When the same happens ten times a month, cost tracking is structurally behind, and a figure that is behind steers decisions the wrong way: it gives permission to spend money that has already been spent.
Three different figures circulate in cost tracking, and confusing them explains much of why tracking does not feel trustworthy.
Actual. What has already been paid or invoiced in. The most accurate figure, and the slowest.
Committed. What has been decided but not yet invoiced: orders, subcontracts, confirmed deliveries. This is the figure that says what you are already bound to.
Forecast. Actual plus committed plus an estimate of what remains. The only figure that answers the question of whether this went over.
In most small construction companies the first is tracked and the third is talked about. The second is missing entirely, and it is precisely the part that would be current without extra work, because the order is placed anyway.
Costs on a construction project do not accrue evenly. They cluster into phases: purchasing at the start, labour in the middle, the costs of finishing at the end.
It follows that a monthly measuring interval lands at random. One month looks good because the invoices have not arrived. The next looks bad because they all arrived at once. Neither says anything about the real state of the project.
Compare this with how a site is otherwise run. Schedule is not reviewed once a month. Safety is not reviewed once a month. Costs are, because cost data comes from the bookkeeping, and bookkeeping runs monthly.
The bookkeeping rhythm is right for bookkeeping. It is not right for running a site.
A small construction company has nobody whose job is cost tracking. Tracking therefore has to be light, and light means choosing.
Four things are enough for most.
Hours against the project. The largest single cost and the only one that is in your own hands daily.
Purchases as commitments. Recorded when the order is placed, not when the invoice arrives.
Subcontracts as contract sums. The commitment is known immediately, and changes to it are rare but large.
Extra work separately. Not inside the base contract, because otherwise you cannot tell which one overran.
Outside these sit small items that accumulate as overhead. It is better to accept them as imprecision than to try to track them, because the effort of tracking grows faster than the accuracy it produces.
When a project has one budget figure and one actual figure, the only possible observation is "over" or "under". That observation arrives too late to be useful.
When the budget is split by work stage, the observation is different: the frame went as expected, but the roof has used 70 per cent of its budget at the halfway point. That can be acted on.
The level of the split should match how costs are recorded. If hours and purchases are booked to the project but not to a stage, dividing the budget into ten parts does not help, because actuals cannot be allocated to match.
In practice four to six stages are enough for most small construction companies. More means recording becomes heavy and allocation stops being accurate.
Cost data alone does not say whether a project is heading over. It says how much has been used. The question is always how much has been used relative to what has been achieved.
Two hundred hours into a four hundred hour contract can mean exactly on schedule or badly behind. The difference is in progress, and it cannot be seen from costs.
It does not follow that a sophisticated progress measurement is needed. It is enough for someone to state an estimate: this stage is roughly halfway. An estimate that has been recorded is infinitely better than a precise figure that does not exist.
The practical form can be simple. Each stage is marked started, halfway or complete. Three states are enough to say whether the cost accrual is in line.
Projects look good but the result does not. The classic sign that some costs are not allocated to projects at all. The usual causes are unrecorded hours and overheads that are never apportioned.
Overruns are always discovered in the final account. If every overrun arrives as a surprise at the same stage, tracking is not measuring, it is reporting.
Nobody can state the position of a live project without an investigation. If the answer takes half a day's work, the information is not available when it is needed.
This is worth asking before building any tracking, because the answer decides what accuracy is needed.
Steering a live project. Needs fresh data but not great accuracy. A ten per cent error does no harm if the direction is right.
Making sure invoicing is complete. Needs accuracy but not speed. Here the question is whether everything billable has been billed.
Feeding the next quote. Needs accuracy and a breakdown by work stage, and no speed at all.
Understanding company profitability. Needs projects to be comparable with each other, which is a different thing from being precise.
Three of the four uses tolerate imprecision. Only one needs speed. That is good news: tracking does not have to be perfect to be useful.
The delay on a purchase invoice is structural. The supplier bills at their own pace, and the invoice travels through bookkeeping. Faster processing does not help.
Two things do.
Book the order against the project when it is placed. Then the cost appears when the decision was made, and the invoice is only a confirmation.
Photograph the receipt immediately. Small cash purchases and collections are the category that disappears entirely, because no invoice arrives for the bookkeeping to react to. A photo of the receipt on the project is often the only trace.
Neither requires a new process, only that an existing act is recorded where it happens.
Figures make the delay concrete. Take a three month renovation budgeted at 60,000 euros: 30,000 labour, 20,000 materials and 10,000 subcontracting.
Month 1. Materials are ordered, 14,000 euros. Labour worth 9,000 euros is done. No purchase invoices have arrived yet.
The month-end report shows 9,000 euros used, 15 per cent of the budget. Reality: 23,000 euros committed, 38 per cent.
The gap is 14,000 euros, and it steers directly to a wrong decision. If someone concludes here that the project has room and orders more, the overrun has already happened before it appears anywhere.
Month 2. The first month's invoices arrive. The report suddenly shows 32,000 euros, which looks like an alarming jump. In reality nothing happened; the bookkeeping simply caught up with the site.
Here a second typical mistake is made: reacting to the jump, even though the jump is a measurement artefact rather than an event.
Month 3. The final invoice reveals the subcontract was 13,000 rather than 10,000, because its scope grew along the way. This was known at the start of month two when it was agreed, but it was recorded nowhere before the invoice.
Had commitments been recorded at the moment of decision, all three points would have been visible in time: month one would not have looked roomy, month two's jump would not have happened, and the subcontract's expansion would have appeared in the month it was agreed.
Cost tracking often fails because nobody agreed who records. Everyone assumes somebody else does.
A workable division in a small construction company looks like this.
The worker records their own hours. Nobody else, because nobody else knows.
The site manager records orders and subcontracts. They make the decision, so they record the commitment. This is the step most often skipped, because an order feels handled once the call has been made.
The worker photographs the receipt. Cash purchases and collections never pass through a purchase invoice, so without a photo they leave no trace.
The office or the bookkeeper reconciles. Compares arriving invoices with recorded commitments and corrects the differences.
The most important of these is the second. If only one thing is adopted, it is recording the order when the order is placed, because that removes most of the delay with a single change.
A common situation: every project looks profitable but the year ends at zero. The explanation is usually one of two things: unrecorded hours, or overheads that are not allocated to projects. The first is covered in the article on time tracking. This section is about the second.
Premises, insurance, depreciation on equipment, bookkeeping, software, vehicle costs, the owner's own administrative time. None of these attach to a project, so none appear in project reports.
The simplest way to bring them in is an overhead percentage. Compare the year's overheads with the direct costs of your projects and add the same percentage to the costs of every project. In construction it is typically 5–10 percent: if overheads are 60,000 euros and direct project costs 800,000 euros, the percentage is 7.5.
This is not exact in accounting terms and does not need to be. It is enough to stop a project looking like a 20 per cent margin when it is in fact five.
The percentage is worth rechecking once a year. If turnover or the cost structure changes, the percentage changes, and an old percentage gives no warning of that.
A rare but essential question, and answering it requires current information.
On a fixed price contract there sometimes comes a point where the remaining work costs more than the remaining invoicing. From then on, every hour worked adds to the loss.
It does not automatically follow that the work should stop. Stopping has its own consequences: breach of contract, liquidated damages, reputation, and an unfinished building. Usually the right move is to negotiate with the client, and negotiation only works if you have the figures.
Three questions you must be able to answer before contacting the client.
How much has been used and how much remains. Not an estimate but recorded hours and commitments.
What caused the overrun. Your own misjudgement, the conditions, or the client's changes. Only the last is grounds for revising the price.
What you are proposing. A negotiation where you present a problem but no solution hands the problem to the client and ends badly.
The timing of that conversation decides its nature. With month-old data it happens after the overrun, and the subject is who pays. With week-old data it happens before, and the subject is how to avoid it.
The FiSAS project cost view is built around the same three figures this article started with:
- Actual builds up from recorded hours and approved purchase invoices and expenses. Hours appear as soon as a worker records them on their phone.
- Reserved corresponds to commitments: materials from the approved quote with no purchase invoice yet, and received purchase invoices not yet approved. When a purchase invoice arrives, it reduces the reservation on the same row, and an approved invoice moves to actual.
- Forecast and margin already include reservations, so at the start of the month a project does not look roomier than it is.
The budget comes from the structure of the approved quote, and the comparison is made at the tracking level the company chooses itself, for example by cost code, and at every level separately for labour, materials, subcontracting and other costs. Purchase invoices come in through the accounting integration and are allocated to the project and the same tracking level, and the lines of one invoice can be split across several targets. Time tracking is included in every plan, expenses and purchase invoices in the Master plan.
The problem with cost tracking is not accuracy but delay. A figure three weeks late is accurate and useless, because the decisions were made in its absence.
The fix is not more comprehensive tracking but earlier tracking: hours recorded on the working day, orders at the moment of ordering, subcontracts at the moment of signing. The imprecision left after those three is small compared with what has been gained.
Try this on one live project: write down what you estimate it has cost so far, then compare that with what the system says. If the gap is large, it does not point to a bad guess. It points to information arriving too late to be used.
