Start with the work that must continue.
List the applications, people and locations affected by an outage. Decide what must keep working, what can wait and who will own recovery. Use that brief to compare backup options before choosing a speed or technology.
Ask your technical team to document an acceptance test: the services to check, the disruption you can tolerate and the person who will confirm the result. Include payments, calls, VPN access and any cloud systems essential to the site.
Check shared failure points.
Two connections can still depend on the same infrastructure. Ask suppliers about the building entry, access route and upstream network. Include local power and equipment in the review. Cellular access does not by itself prove end-to-end independence from a wired service.
CISA identifies power and other infrastructure dependencies in communications systems. Its route-diversity guidance is a useful starting point for the questions your suppliers and technical team should resolve.
Compare the available options.
LTE/5G cellular
Ask about signal quality at the installed router, antenna requirements, available capacity, data allowances and any traffic restrictions. Request a site test under representative working conditions and a confirmed activation date.
Fixed wireless
Confirm coverage, antenna placement, mounting access and installation requirements. Ask whether the proposed path shares facilities with the primary connection; do not infer diversity from the wireless label.
Secondary wired service
Request route and building-entry information alongside the quote. Review construction needs, installation timing, term length and recurring charges. Have the technical team assess whether the second circuit addresses the failures you are planning for.
Satellite
Ask about installation, sky visibility, service availability, usage policies and support. Test the applications that matter at your location before accepting the service as a backup.
Size and test the backup.
Build a list of essential applications and concurrent users. Review upload and download demand, latency sensitivity and expected data usage during an outage. Agree which nonessential activity should pause. Choose capacity from that assessment and testing, rather than a universal bandwidth recommendation.
Record the test results and any limitations. Repeat the review when staffing, applications or equipment change. Confirm who receives outage alerts and who can contact the supplier outside normal working hours.
Automatic does not mean interruption-free.
Failover behaviour depends on hardware, monitoring and configuration. Cisco Meraki documents different detection and failback behaviour for its supported devices; that is a product-specific example, not a promise for every network. Active calls, VPNs and application sessions may need to reconnect.
Ask the installer to demonstrate a primary-link failure and recovery. Check essential applications in both directions, confirm alerts and document a manual recovery procedure if automatic switching does not restore service.
Compare complete costs.
Request service, equipment, installation, licensing, data overage and support costs. Compare them with your own estimate of lost transactions, staff disruption and recovery effort. Account for existing alternatives and work that can continue offline. Avoid assuming a backup will pay for itself after one outage.