Now onboardingCenters and field teams across 225 cities.List your center

← Knowledge
Guide8 min read28 September 2026

What makes a good computer-based test centre: a nine-point check

Seat count is the easiest number to get and the least useful on its own. These are the nine things that decide whether a lab can actually run your exam on the day.

Exam centresCBTCentre readinessInfrastructure
In short: installed machine count is the number you will be quoted and the weakest predictor of whether an exam runs. The nine checks below are the ones that actually decide the day: usable capacity, power, connectivity and its fallback, workstation spec, invigilation sightlines, candidate flow, waiting space, technical cover, and accountability.

A centre with 200 workstations and no tested power backup is, for a three-hour computer-based test, a nought-seat centre. So is one with 200 machines on a single internet line during monsoon. Capacity is necessary and nowhere near sufficient, but it is the thing that gets quoted first because it is the easiest thing to count.

Here is what we check before a centre goes into our network, in rough order of how often it is the thing that fails.

1. Usable seats, not installed machines

Installed count includes machines that do not boot, seats too close together for your spacing rules, and positions no invigilator can see. Usable count is what is left: workstations that work simultaneously, are spaced to your requirement, and are visible. The gap between the two numbers is routinely ten to twenty per cent, and it is always discovered on exam morning rather than at booking.

Ask for the count that has been verified with every machine powered on at the same time. A lab that has never had all its workstations running together has not been tested at the load your exam will apply.

2. Power backup rated for the session, not for bridging

What you are told What to ask instead
"We have a UPS."Rated for how many minutes, at what load, tested when?
"We have a generator."How long does the cutover take, and does it carry the network too?
"Power is never a problem here."What happened the last time it was?

A UPS sized to give people time to save their work is not the same as backup sized to finish an exam. A three-hour session with a handover gap wants roughly four hours of continuous cover — and the cover has to include the servers and network equipment, not just the desktops. A lab on backup power with a dead switch is a lab full of candidates staring at frozen screens.

3. A second internet path, and evidence someone has tested the switch

One line is one point of failure for the whole shift. What matters is not that a second router exists but that the fallback is on a different provider or medium, and that the switchover has been rehearsed. An untested failover is a plan, not a backup.

4. Workstation specification

Screen size and resolution decide whether your paper renders as designed — a layout built for a certain viewport can wrap badly on smaller screens and change how long questions take to read. Keyboards matter for anything with typed answers. Where the exam uses audio, so do headsets, and they need to be checked per seat rather than sampled.

5. Invigilation sightlines

Partitions that stop candidates seeing each other's screens also stop an invigilator seeing the room. Good labs resolve this with height and layout rather than by choosing one or the other. Walk the room from where the invigilator will stand, not from the doorway.

6. Entry, verification and candidate flow

Candidate throughput at the door sets how late your session starts. A single entrance handling identity checks for 200 candidates is a queue, and a queue outside a building is a problem for whoever owns the building. Look for somewhere to verify identity that is not the corridor outside the hall, and a route in that does not cross the route out when a shift changes over.

7. Waiting space

Candidates arrive early. If there is nowhere to wait, they wait in the exam hall or in the street. Both create problems: one contaminates the room before the session, the other is where complaints come from. Washrooms and accessible access belong in the same check.

8. On-site technical cover for the whole session

Not a number to call. A person in the building, for the duration, who can replace a workstation and knows where the spares are. The question to ask is how many spare machines are configured and ready — not how many machines exist in storage.

9. One accountable person on the day

Every centre has a hierarchy. On exam day you need one name with the authority to make a decision when something breaks, and a way to reach them that is not the main reception number. Where this is unclear at booking it will be unclear at 9am.

The pattern: eight of these nine are questions about evidence rather than equipment. "Do you have X" is nearly always answered yes. "When was X last tested, and what happened" is the question that finds out.

Why we audit rather than ask

Self-declared facilities are not a reliable input. A centre filling in a form has every reason to answer generously and no way of knowing which of the nine above your exam actually depends on. Our own team checks facilities, connectivity, power backup and access, and audited venues surface first when a drive is being planned — which is the only version of this that survives contact with an exam morning.

Common questions

What should you check before booking a computer-based test centre?
Usable seat count rather than installed machines, power backup rated for the full exam duration, a second internet path, workstation specification, invigilation sightlines, separate entry and exit flow, waiting space outside the hall, on-site technical cover for the whole session, and a named person accountable on the day.
How much power backup does an exam centre need?
Enough to carry the whole session, not a few minutes of bridging. A three-hour CBT with a handover gap needs backup rated for roughly four hours of continuous load on the lab, servers and network equipment together — and it needs to have been load-tested with machines actually running, not just present on paper.
What is usable seat count?
The number of workstations that are simultaneously working, correctly spaced and visible to an invigilator. It is normally lower than the installed machine count once you remove faulty units, seats too close together for the exam's spacing rules and positions an invigilator cannot see.
Does a CBT centre need a backup internet connection?
For anything high-stakes, yes. A single line is a single point of failure for the whole shift. What matters is that the fallback is on a different provider or medium and that someone has tested the switchover, rather than that a second router exists.