BEYOND THE DRUMBEAT · 1 OF 6 · The Contractor Merry-Go-Round

Anyone who’s been involved in commercial management on a large infrastructure programme knows the feeling. The application for payment comes in, month-end is bearing down, and something in the numbers doesn’t sit right. You can’t quite put your finger on it, but the sniff test says no. The certificate still has to go out. So you make a judgement and advise the PM, tell yourself it’s only an interim, and move on. You’ll sort it out at the final account.

I want to dwell on that phrase, because it quietly does more damage than almost anything else in programme delivery. ‘We’ll sort it out at the final account’ sounds sensible. Pragmatic, even. But it’s the first turn of something I’ve come to think of as the contractor merry-go-round, and once you’re on it, getting off is harder than you’d think.

How the merry-go-round turns

It starts small. Month one, there’s a minor challenge over what you recommend the PM to certify when an application for payment lands. A bit of cost coded in a way that doesn’t match, a quantity that looks off, a figure sitting in a line where it doesn’t belong. Not worth stopping the machine for. You note it, wave it through as interim, and turn to month two. Except month two arrives with its own set of challenges before month one’s are resolved. Then month three. And the contractor, who has seen this film many times, knows exactly how it ends.

Part of the trouble is that you’re rarely comparing like with like in the first place. Ten or fifteen projects, each interpreting the same contract slightly differently, each feeding data through systems that were never designed to talk to one another. An ERP built for running a business, not for project accounting. A name in the wrong box, a transposed number, a cost mapped to the wrong code. Individually, trivial. Collectively, they mean the picture you’re certifying against is a little bit wrong in a dozen small ways, every month.

And it isn’t only the systems. It’s the people, quite reasonably, doing their best with what they’ve got. A common client team, but multiple contractors, each reading the same contract in a slightly different way. Designers doing the same. Everyone applying their own interpretation to the same clause or Scope requirement, in good faith, and producing returns that look alike but aren’t truly comparable. Layer a genuine human disagreement about what the contract even means on top of a dozen small coding errors, and the picture you’re recommending the PM certify against starts to look less like a photograph and more like an impression.

By month six of a twenty-four month contract, you’ve lost the thread of where the problems started. The discrepancies haven’t gone away; they’ve layered on top of one another, and the data you’d need to unpick them was never captured cleanly to begin with. You’re now making commercial decisions on a monthly partial picture you don’t fully trust, under time pressure, month after month. And the party on the other side of the table understands your position better than you’d like.

Then comes the final account. Here’s the cruel part: the moment you finally have to look at it all in detail is the exact moment you find you can’t. The data set isn’t there. The forensic trail you’d need to say ‘that was never substantiated’ doesn’t exist in any usable form. So, the interim positions you promised yourself you’d revisit quietly become permanent. No matter your professional stance, what you called temporary was, all along, the final answer.

Why it gets worse the bigger the programme

On a short scheme you might get away with it. On a modern major programme, you won’t. When a railway or a road runs five to ten years and you may well not be the person who does the final assessment, the projection of that monthly drift across the whole life of the work stops being a rounding error and starts being a number that turns up in a board paper. Small acorns, as one experienced commercial director put it to me, have a way of becoming very large trees.

And your professional accountability has sharpened. In the public sector especially, the scrutiny over cost forecasting and cost exposure is serious now. You’re expected to forecast cost against budget accurately, and to explain upstream where creep has come from and why. ‘We’ll know at the final account’ is not an answer that survives a modern governance process. If you’ve reported green all the way through because the sampled data told you it was green, the day it turns red is the day you’ve run out of room to do anything about it. The project team come back three-quarters of the way through and tell you they need more to finish, and by then you’re committed, with nowhere to go.

Notice what that does to the people around the table. It quietly pushes good professionals into adjusting their own judgement to fit data they don’t really believe, which is a miserable way to work and a direct route to optimism bias. You keep saying it’s fine because the numbers say it’s fine, right up until the point where it’s so obviously not fine that you can’t say it any longer. Then it drops off a cliff, in time or in money or in both. The animation everyone recognises from these programmes is the flat green line that stays flat green until, suddenly, it can’t. Some of that line was bent from the very start, in a forecast that was optimistic long before anyone opened a cost report, which is a problem I’ll come back to later in the series. But a good deal of it is simply deferral, compounding quietly, month after month, and that part is entirely within your gift to stop.

The drum-beat is the point

So, what actually breaks the cycle? Not heroics at the final account. By then it’s too late. The only thing that works is refusing to let the first turn happen: assuring the quality and accuracy of cost data properly, every month, so that nothing is carried forward unexamined.

That’s what the monthly assurance cycle, the drumbeat, is really for. Not a reporting ritual to feed a dashboard, but the mechanism that reconciles at source and evidences each month cleanly before the next one opens. Done by hand, across a dozen projects, with data arriving in different shapes from different systems, it’s close to impossible, which is precisely why so many good teams quietly give up and fall back on the sampling assurance technique of trying to judge what is 80% of the months costs in 20% of the months cost data; the sniff test. Done digitally, with every line of everything to data assured every month rather than a sample extrapolated, it becomes achievable. This is the unglamorous heart of digital cost assurance: it stops the compounding before it starts.

The distinction that matters is between sampling from this month’s data and assuring 100% of cumulative data every assessment. The old way, the only way that was ever practical by hand, is to take a 80/20 sample, check it, and extrapolate across the rest. On a good month that’s fine. On a bad month it’s exactly how you end up reporting green while the unsampled eighty per cent quietly goes wrong. Providing reasonable procedures to prevent fraud means assuring all of it, every line, every month, so there is no unexamined remainder left to surprise you later. That is the single pane of glass people talk about and rarely have. Not a prettier dashboard, but the confidence that what you’re looking at is the whole of the truth rather than a flattering slice of it.

The final account was never a safety net. It’s just where the bill for months of deferral comes due.

None of this needs a cultural revolution. It needs a decision, taken every month, that ‘we’ll sort it out at the final account’ is not an acceptable answer. Because by the time the merry-go-round comes round to the final account, it has already taken its fare, and it isn’t inclined to give it back.