Camp capacity is the constraint nobody models until check-in fails. Planning rooms and seats before names changes what the roster can promise.
The queue at the camp office
There is a particular scene that plays out at remote camps across the country more often than any operations report admits. It is 6:30 in the evening. A bus has just arrived from the airstrip carrying 40 workers at the end of a long travel day. They file into the camp office to check in, and somewhere around worker number 33, the person behind the counter stops smiling. The rooms have run out.
What follows is a small masterclass in improvisation. Someone gets a room that was being held for the drill and blast crew arriving Thursday. Someone gets a room that has not been cleaned. Two people get sent to the exploration camp 45 minutes up the road. One supervisor makes four phone calls and burns a favour with the village manager that he will want back someday. And the next morning, everyone quietly agrees it worked out fine, which guarantees it will happen again.
Here is the thing worth sitting with: every person on that bus was rostered weeks ago. The flights were booked. The site access was approved. The one resource nobody confirmed was the one with the hardest physical limit. You can, at painful cost, charter another plane. You can, with grovelling, get another gate pass approved same-day. You cannot invent a donga at 6:30pm. Of all the constraints in the FIFO machine, bed capacity is the least elastic, and it is routinely the last one anyone checks.
That ordering is the bug. This article is about reversing it.
Names first is the natural mistake
Almost every roster process in the industry runs in the same direction. It starts with the work: the project needs 60 people on for the August swing. Then it moves to the names: which 60, matched for competency, availability and leave. Then the logistics team books flights for the names. Then, last, camp gets a manifest and is expected to produce 60 beds, because camp has always produced the beds before.
This sequence feels natural because it mirrors how the organisation thinks about value. The work is the point, the people do the work, the logistics serve the people. Beds are, in this mental model, a downstream detail, like catering or laundry.
The model survives only as long as the camp has slack. And for years, many camps did. But run the sequence during a shutdown, when three contractors surge simultaneously, or during a construction phase overlapping operations, or in a labour market where every contractor is hiring, and the slack vanishes. Suddenly the last step in the chain, the one with no flexibility, is being asked to absorb all the variability created by every step before it. The camp office queue at 6:30pm is not a camp failure. It is the roster's planning debt, presented for payment at the only counter that cannot extend credit.
The tell is what happens next. Projects learn the camp is unreliable, so they start hoarding. Rooms get block-booked "just in case" and sit empty. Utilisation reports show the camp 92 per cent full while 15 per cent of rooms host nobody. Now the constraint is even tighter, not because of demand, but because of the defensive behaviour the broken process taught everyone. Empty rooms with names on them are the camp equivalent of padded rosters, insurance purchased against a process nobody trusts.
Reversing the sequence
Planning from the beds backwards means treating camp capacity, and its close cousin flight capacity, as the first input to the roster, not the last check before travel. In practice, the sequence becomes: here is the room envelope for each week of the planning horizon, here is the flight seat envelope, therefore here is the maximum on-site headcount the roster is allowed to promise, and only then, which names fill it.
This sounds obvious written down. Airlines have run this way forever: nobody sells the seat and then wonders whether the aircraft exists. But applying it to a swing roster has consequences that reach further than logistics, which is why it is worth spelling them out.
The roster becomes a promise you can keep. When the headcount ceiling is set by verified capacity, the project plan built on that roster is built on something real. The alternative, planning the work first and discovering the bed shortfall three weeks out, does not actually get you more people. It just delays the bad news until every option is expensive: split crews, motels in town two hours away, deferred scope. Beds-first does not shrink your capacity. It tells you the truth about it earlier, while the truth is still cheap.
Peaks get negotiated instead of collided with. With a capacity envelope visible across the horizon, the collisions show up months out: the mechanical shutdown, the SMP contractor's surge and the client's own maintenance crew all want week 34. Seen early, this is a planning conversation, someone shifts a week, someone staggers a start, someone books overflow camp at normal notice. Seen late, it is a bidding war for dongas, and the loser is whoever has the least leverage, which is usually a subcontractor whose delay becomes your delay anyway.
Roster shape becomes a lever. Once beds are the constraint you plan around, roster design starts responding to it. Staggering swing changeovers by even a day or two stops the changeover-night double occupancy spike, where inbound and outbound crews overlap and the camp needs beds for both. Moving one crew from 2/1 to an offset cycle can flatten a peak that would otherwise require 30 phantom rooms held all quarter. These are free capacity gains, available to anyone who can see demand and supply on the same page. Which is, of course, the catch.
The same page
Because here is why nobody models camp capacity until check-in fails: the data lives apart. The roster lives with workforce planning. The room bookings live in a camp management system, often the client's, sometimes a spreadsheet at the village. Flights live with a travel coordinator or the client's booking portal. Each system is locally fine. The collision only exists in the join, and in most operations there is no join, there is a weekly email and a lot of assumed knowledge.
So the practical reform is unglamorous: build one view, per week, across the horizon, with three lines on it. Rostered heads. Booked or available beds. Booked or available seats. Where the first line crosses either of the others, that is next quarter's 6:30pm queue, visible today. The first version of this can be a spreadsheet assembled monthly. It will be wrong in places. It will still catch collisions at a distance where fixing them costs a meeting instead of a favour.
Two refinements pay for themselves quickly. First, count bed-nights, not headcount, because a 60-person crew on staggered travel days does not need 60 rooms every night, and the difference is real capacity. Second, put changeover nights on the chart explicitly, because that is where the hidden double-demand lives and where most check-in failures are actually born.
And retire the phrase "camp will sort it." Camp cannot sort it. Camp can only distribute a shortage that was created upstream, and every time camp performs a miracle, the roster process learns nothing. The kindest thing an organisation can do for its village staff is stop requiring heroics of them.
What the roster can promise
A roster is, in the end, a set of promises. To the client: this many capable people, on these dates. To the workers: a bed, a flight, a predictable rhythm their families can plan around. To the project: the labour curve the schedule was built on.
A roster planned names-first makes those promises on hope and history. A roster planned from the beds backwards makes fewer promises, occasionally, and keeps essentially all of them. Between the two, every experienced client, and every experienced worker, will take the second. Nobody remembers the roster that promised 64 and delivered 58 as ambitious. They remember the night the rooms ran out.
The beds were always the constraint. The only choice is whether they constrain your plan on paper in March, or your people at a check-in counter in August. One of those costs a planning meeting. The other one costs favours, motel invoices and trust, and of the three, the trust is the item that does not come back with the next swing.
