A twenty-city drive is not twenty independent bookings. It is one calendar with twenty places it can fail, and the failures are correlated: the same week, the same paper, the same operations team stretched across all of it. The plan that survives is the one where the expensive, slow decisions are made first and the fast-changing ones are left until last.
Order of operations
1. Cities come from your candidates, not your venues
The temptation is to start from the cities where booking is easy. That inverts the problem: you end up asking candidates to travel further, attendance drops, and the drive looks under-subscribed for a reason that was designed in at step one.
Do it the other way round:
- Cluster the registered list by city.
- Decide what travel distance you consider acceptable, and say it out loud — it is a policy decision, not a logistics detail.
- Collapse clusters that fall inside that distance of each other.
- Then check which of the resulting cities have venues at the size you need.
Where a city has candidates but no venue at your size, the answer is more shifts or a smaller room, not moving the candidates to the next city along. Our own footprint runs to 730 centres across 225 cities in 35 states, which usually means the constraint is room size rather than the city itself.
2. Size each city on its own
Attendance is not uniform across a drive. A metro where most candidates already live and work behaves differently from a city they travel into. Applying one national attendance assumption to twenty cities guarantees you are wrong in both directions at once — over-booked in some, short in others, with the two errors cancelling out in the total and hiding the problem.
Size city by city, with the buffer applied per centre and shift rather than across the drive. A ten per cent cushion held centrally is no help to the one room where three machines died.
3. Decide the shift pattern before you approach centres
Centres quote against a pattern. Changing from two shifts to three after you have negotiated is a renegotiation, not an amendment.
Two sessions in a day is comfortable almost anywhere. Three is achievable where the exam is short and the centre is well drilled. The limit is rarely the exam duration — it is the changeover: clearing one cohort out of the building while the next is arriving, resetting workstations, and running a fresh identity-check queue without the two cohorts mixing. A centre with one entrance will struggle at three shifts however fast its machines are.
One date or several?
High-stakes exams usually take the single date and pay for it in readiness effort. Screening and recruitment drives are normally better off staggered, because the ability to fix on day two is worth more than the security margin.
5. Leave the roster until last — deliberately
Rosters change until the final week. Names get added, candidates switch cities, shifts get swapped. If mapping candidates to centres is coupled to booking those centres, every late change becomes a venue conversation.
Keep them separate. Book capacity early, against your sized estimate. Map candidates to seats late, as an explicit and reversible step. A late change then edits an allocation instead of reopening a negotiation — which is the difference between an afternoon's work and a week's.
The two constraints that always bite
Room size distribution. Cities have many more small rooms than large ones. A plan that needs a 300-seat hall in every city will fail in most of them; a plan that needs 100 seats three times over will not.
Your own operations capacity. Twenty cities on one date needs twenty accountable people briefed and reachable simultaneously. That is usually the real ceiling on how wide a drive can go on a single day — not venue availability, and not seats.