Most procurement questions have prepared answers, which is why most procurement produces so little information. The questions below are harder to rehearse, and what matters is as much how a vendor responds as what they say.
1. What does this platform do badly?
Every platform has weak areas. A vendor claiming none is either inexperienced or managing you, and both predict how the relationship will go when something breaks.
The useful answer names a specific limitation and explains the trade-off behind it.
2. How many incidents in the last twelve months, and what was the mean time to recovery?
More revealing than any availability percentage, because recovery time cannot be improved by redefining what counts as available.
Follow with: is there a public status page with incident history, or does information come through an account manager?
3. What happens when one game provider goes offline?
Tests failure isolation directly. The good answer is that the provider’s titles disappear cleanly from the lobby and everything else continues. The concerning answer involves loading screens.
4. Show me a full player history export.
Not a description — the actual output, for a player with several years of activity, timed.
This single request tests audit readiness, data completeness and export capability simultaneously. It also reveals whether verification documents, exclusion records and bonus history are included or merely referenced.
5. Is jurisdiction a player attribute or a deployment setting?
Determines whether market expansion is a configuration change or a new instance. The answer shapes the cost of every market after your first.
6. Can I set a deposit limit and then breach it using a different payment method?
Ask them to demonstrate it in a sandbox. Limits enforced per payment route rather than per account are common and are not limits.
Follow with whether increases carry a cooling-off period and whether self-exclusion propagates across brands automatically, including to marketing suppression.
7. What is your API versioning policy, and when was the last breaking change?
Concrete, verifiable and predictive. Vendors with disciplined versioning answer immediately with a deprecation window. Vendors without it become an emergency engineering project on somebody else’s schedule.
8. Can I simulate a promotion against my historical transaction data before launching it?
The most expensive promotional errors are interaction effects that specification review cannot catch. Platforms supporting simulation prevent a category of loss that platforms without it can only discover afterwards.
9. What are your own critical third-party dependencies?
Your platform depends on infrastructure, studios, payment providers and verification vendors. Those failures reach you through the platform.
A vendor willing to discuss their dependency map has thought about it. Reluctance to answer is itself informative.
10. When did you last perform an actual failover, and what did not work?
A tabletop exercise validates a document. Only a real failover validates the system, and the two find different problems.
The second half of the question matters more than the first. Any genuine failover test surfaces something unexpected, and a vendor who says everything worked perfectly has either not tested or is not telling you.
How to use the answers
None of these questions has a single correct answer. Several have legitimate responses that would rule a platform out for one operator and be entirely acceptable for another.
What they collectively reveal is whether the vendor operates with the discipline the answers imply — versioning policies, incident histories, dependency maps and failover results are all artefacts of an engineering culture. Organisations that have them can produce them quickly. Organisations that do not will offer to follow up.
Ask them of every provider under consideration, including those positioning as premium online casino software, and compare the answers side by side rather than against a feature matrix. A vendor who answers eight of these directly and admits uncertainty on two is a considerably safer proposition than one who answers all ten smoothly.
One thing to do afterwards
Ask the same questions of a customer who left that vendor.
Industry events and mutual contacts make this more achievable than it sounds, and departed customers are the single most useful source of information available during procurement. Their answers to questions two, four and ten in particular tend to differ from the vendor’s.
