Price the roster before the margin

Most bids for government IT services are priced from the top down. A commercial team estimates the price likely to win, a finance team calculates the margin the company can accept, and delivery is asked afterwards to make the staffing fit. The result is a document in which the price is carefully argued and the roster is an appendix. I have signed bids built that way. Several of the contracts that followed were the hardest years of my career, and none of the difficulty came from the price itself. It came from the people the price had quietly assumed.

I now hold the opposite view. A government service contract is a promise about people: how many will answer the phones at seven on a Monday morning, how many will hold the clearances the work requires, how many will be on leave, in training or newly hired in any given month. The price follows from that promise. A board approving a bid should read the roster first.

A bid is a roster with a price attached

In 2023 a state health department tendered a seven-year service-desk and end-user computing contract for 64,000 users. Its documents valued the work at about A$540 million. The terms required every role to be onshore, many of them cleared, a 90-day transition, service credits of up to 20 per cent of the monthly fee and uncapped liability for data breaches.

Our commercial team believed a price of about A$72 million a year would win. When I built the roster, a 24-hour onshore desk of 310 full-time equivalents and a field team for 140 sites, that price required our people to be chargeable 91 per cent of their paid hours. No allowance for annual leave, sick leave, training, team meetings or the turnover a new contract always brings. The price was achievable only on paper, and the paper would have been ours.

We bid A$83 million a year, proposed a liability cap and a longer transition, and explained why in the submission. We lost. The rules governing Commonwealth procurement make value for money, not the lowest price, the core test, and state rules are broadly similar in intent. But evaluation panels must work with what bidders submit, and a lower price backed by a confident roster is hard to set aside. Restraint, in practice, has to come from the supplier.

Clearances move slower than transitions

The roster is only the first test. The second is time. Many government contracts require staff in certain roles to hold a security clearance from the first day of service. Clearances are sponsored through the Commonwealth's vetting arrangements and processed on their own timetable; baseline checks can take weeks and higher levels considerably longer. A supplier cannot apply for clearances for people it has not yet hired, and it cannot hire people for a contract it has not yet won.

So a 90-day transition with cleared roles carries a hidden assumption: that the winner already employs enough cleared people, available to move, or that it can hire them from the outgoing supplier. Neither is reliable. Cleared people are scarce in every city where government work is concentrated, and the outgoing supplier's staff have their own choices to make, often with offers from several bidders at once. Knowledge transfer depends on the goodwill of a team that has just lost its contract, and goodwill is hard to schedule. A transition plan that assumes all of this will go smoothly is really a second price, hidden inside the first, and the client usually pays it in the opening months of service. In one of our own transitions in 2022, 40 staff waited longer for clearances than our plan allowed, and the service went live three weeks late. We paid for that in credits and in the client's confidence.

Our transition plans now begin clearance applications for named people before contract signature wherever the client agrees, and our bid papers show the clearance timetable as a dated chart beside the transition plan. When the two do not meet, the board sees it before the price is set.

Saying no in writing

The third test is willingness to decline. Since 2023 every Torvelle bid above A$50 million reaches the board with its staffing plan, clearance timetable, transition plan and liability position before the price is discussed. If I cannot sign the delivery plan, the bid does not proceed, and the board knows I hold that right. When we decline, we write to the agency explaining which terms prevented us from bidding, respectfully and briefly, because a reasoned refusal is useful information to a buyer preparing its next tender.

The results have been mixed in the way honest results usually are. We now decline about 31 per cent of the opportunities we qualify, up from 12 per cent. Our win rate on submitted bids has risen from 29 to 41 per cent, and margin on government contracts has held at 12.6 per cent. But government revenue grew only 4 per cent in 2023–24 against a plan of 9 per cent, two experienced bid managers left because they wanted to win large deals rather than decline them, and our chief commercial officer, Stuart Coultard, still believes we walked away from at least one contract we could have run well. He may be right. Discipline in bidding is a judgement, not a formula, and it will sometimes be wrong in the cautious direction.

The case for it rests on the asymmetry between the two kinds of error. A bid declined in error costs a sales cycle and some growth. A bid won in error costs seven years of service credits, exhausted teams and a reputation with an agency that will remember the transition long after it has forgotten the price. Government clients also bear the cost: users wait longer, and the department's own staff spend their time managing a supplier instead of serving the public.

For directors of a services company, the practical questions are simple. What utilisation does this roster assume, after leave and training? How many roles require clearances, and how long will they take? What happens to the transition if they take longer? Which terms would we refuse, and have we said so to the client? A bid paper that cannot answer those questions has priced a margin. It has not yet priced the service.