1. Connectivity density
For most workloads the connectivity inside the facility is the product. The questions worth answering first are which carriers are physically present, whether the carriers you actually need are on-net rather than via a cross-connect to a remote pop, and whether an internet exchange is in the building or in a separate meet-me room. A facility with one hundred carriers on-net and an internet exchange present is structurally different from one with fifteen carriers and a remote exchange handoff, even if the rack price is identical.
Verify the specific carriers you need are there. The carrier list on the operator's marketing page is sometimes aspirational; the PeeringDB facility record is closer to ground truth.
2. Location, latency, and jurisdiction
Latency to your end users is a function of physical distance and routing path. A facility ten kilometres closer to your user base is not automatically faster if its upstream paths force traffic through a distant peering point. Ask for measured latency from the facility to the destinations that matter to your workload, not the operator's preferred test points.
Jurisdiction matters more than it used to. Where the data physically sits affects which legal disclosure regime applies, which trade controls apply to the equipment, and in some sectors which regulators have inspection rights. Confirm the operator can sign the data-residency clauses your workload actually requires before you commit to the cage.
3. Power, cooling, and density
Power density is the line item that scales hardest as workloads change. A contract that allocates four kilowatts per rack reads fine until the first generation of GPU servers arrives and the actual draw is twenty. Confirm the available power per rack, the redundancy commitment (N+1 or 2N), the overage pricing if you exceed the commit, and whether you can grow power without re-contracting the whole cage.
Cooling is the equivalent gotcha on the thermal side. Air cooling at high density without supplemental liquid cooling becomes a thermal management problem the operator may try to push onto the tenant. If your workload is dense, confirm what the operator supports and what you have to build yourself.
4. Certifications
The certifications procurement teams ask about most are SOC 2 Type II, ISO 27001, and the relevant industry-specific frameworks for the workload (PCI DSS for cardholder data, HIPAA for protected health information, FedRAMP for US federal data). The label is a starting point, not a guarantee. Ask for the SOC 2 Type II report under NDA and read the exception list. Verify the ISO 27001 certificate scope includes the specific facility, not just the parent organization. The certifications hub on viabandwidth filters by each so a shortlist can be drawn directly.
5. Operator credibility
The procurement question that does not appear on most checklists is whether the operator actually owns and runs the building. Marketplaces and resellers can be perfectly good procurement partners for some workloads, but they introduce a layer in the operational chain. When a power event happens at three in the morning, the question is whose engineer is in the building.
viabandwidth flags direct operators (verified own infrastructure, network footprint confirmed), carrier-neutral operators, marketplaces, and resellers separately. Use the operator-type signal to decide which class of partner fits the workload before you start the shortlist.
6. Contract terms
The line items that quietly compound are cross-connect fees, power overage rates, remote-hands rates, and exit terms. A reasonable cross-connect fee is meaningful when you are committing to dozens of them; a punitive overage rate becomes a discipline problem on your operations team. Exit terms become the line that matters when the workload outgrows the building.
Useful questions to ask: what is the monthly cross-connect fee per circuit, what is the unit rate for power over commit, what is the remote-hands hourly rate for in-hours and out-of-hours, what is the early-exit penalty structure, and what notice period applies for the operator to change any of the above.
Shortlist workflow
A workable procurement workflow on viabandwidth is to start at the country hub for the geography that matters, filter to the certification hubs that match the workload (SOC 2 and ISO 27001 for most enterprise asks), check each operator's profile for the operator-type signal and network presence, and unlock the operator dossier for the operators on the shortlist to get the verified contact route. The reveal is one credit per operator, less expensive than the typical cost of an unverified sales-route email bouncing.