An invoice for data protection usually pays for copies. The conversation that produced it was about being able to work again, which is a different service with a different design and a different price. Nothing in a monthly figure tells a buyer which of the two they bought, so the useful work happens before signing: fixing how much data the business can afford to lose, how long it can afford to be down, and who is contracted to act when both clocks start. This record holds no provider that publishes a rate for any of it.
A copy is not a return to work, whatever disaster recovery companies call it
Backup as a service is a copy held somewhere else, verified as written. Managed disaster recovery is a commitment that the business operates again, which needs somewhere for the workload to run, an order in which systems come back, someone authorised to declare that the plan is now in force, and a person on a phone at three in the morning who is contracted to do it. Those are separate costs and they appear on separate lines when a provider itemises honestly. Ask which of the two the proposal prices, then ask what the other one would add, because a firm that only sells the copy will still discuss the promise in a sales meeting.
The two numbers are the purchase, and a schedule rarely states them
Every provider asks how much data can be lost and how long the business can be down, because both drive the design. Almost no proposal comes back with either number written as a commitment. They behave differently on price: shortening the acceptable data loss means replicating more often, which is mostly a storage and bandwidth cost that rises smoothly. Shortening the acceptable downtime means paying for capacity that sits waiting, which steps up in jumps as the design moves from rebuilding on borrowed hardware, to standing a copy up on demand, to keeping a warm copy running. Put both numbers in the schedule per system rather than for the business as a whole, and ask what happens when one is missed, because a commitment with nothing attached to it is an aspiration.
Tested means a failover, a date and a named outcome
A dashboard showing green is a monitoring report. A recovery test is the workload actually running from the recovery copy, on a stated date, for a stated period, with a named engineer, a record of how long the switch took, what did not come back first time, and how the business returned to normal operation afterwards. Failing back is the half that gets skipped and the half that causes the second outage. Ask for the date of the last exercise, who ran it, which systems were in scope, and how many were left out; a provider that has never exercised a failover for any client is selling a design rather than a service.
Where recovery quietly excludes the thing you needed
The gaps are predictable. A line of business application is restored and will not start because its vendor holds the licence key and issues a new one only during office hours on a support contract nobody renewed. A cloud identity tenancy cannot be recovered by the provider at all, because the thing lost was the administrative credential, so the recovery depends on a process run by the platform vendor rather than anything in the contract. Anything a third party hosts on the business's behalf sits outside the scope entirely. Ask for the list of systems in scope, then ask directly which systems are out and why, and make sure the answer names who holds the global administrative credential for the Microsoft 365 tenancy, both during the contract and on the day it ends.
What this record can say about the price
Twenty nine IT providers were read on 10 September 2026 and seven publish any figure at all. Not one publishes a rate for recovery, a fee for building a plan, or a charge for invoking one. The figures that do exist are monthly per seat rates for support: Ntiva publishes Starting at $99 per user, Electric publishes $10/user/mo, and the median across the seven is $99.50 per seat. They are priced per person, and recovery is not priced per person, which is why a seat rate never answers this question. The nearest thing in the record to a recovery unit is Powerful IT in Milwaukee, which publishes $75 per workstation per month beside $125 per server, a reminder that anything running as a server is quoted separately. Expect to supply the units yourself when comparing quotes: per protected server, per terabyte held, and a separate charge for the capacity used while running on the recovery copy.
Questions people ask about disaster recovery services providers
What is the difference between backup as a service and managed disaster recovery?
The first is a verified copy held elsewhere. The second is a commitment that the business runs again within a stated time, which needs somewhere to run, an order of recovery, and somebody contracted to execute it.
What do disaster recovery services providers charge?
This record cannot say. Of 29 providers read on 10 September 2026, not one publishes a recovery rate. The seven published figures are monthly per seat support rates, which exclude recovery work.
Which two numbers should be in the contract?
How much data the business can afford to lose and how long it can afford to be down, stated per system, with what happens if either is missed.
What counts as evidence that recovery works?
A failover exercised on a named date, for a stated period, with the switch time recorded, the systems in scope listed, what failed first time, and how the business returned to normal running.
How is data backup and recovery in healthcare different?
By what the two numbers have to be. A clinical system that cannot be reached is a patient safety question rather than an inconvenience, so the recovery time and recovery point are set by the practice rather than by the provider's standard schedule, and the failover has to have been tested on a date somebody can name.