Questions to Ask a Disaster Recovery Company Before Hiring

Questions to Ask a Disaster Recovery Company Before Hiring

Questions to Ask a Disaster Recovery Company Before Hiring

Questions to Ask a Disaster Recovery Company Before Hiring

What happens to your business if your main server room goes dark at 2 a.m. and stays dark for three days? That question is exactly why hiring a disaster recovery company matters, and why the questions you ask before signing can save you from a very expensive surprise later. A provider can look impressive on a sales call and still leave you exposed when a real outage hits.

Quick Answer: What questions should I ask a disaster recovery company before hiring?

Ask for written recovery time and recovery point targets for each system, not a single company-wide promise. Then ask how those targets were set, whether a business impact analysis backs them up, and how often the provider actually tests recovery rather than just scheduling tests.

Dig deeper into whether backups are immutable and offline, how ransomware recovery works in practice, and what standards or certifications the provider holds. Backup and disaster recovery are not the same thing, and a vendor selling only backup may never deliver a working recovery.

Before signing, compare recovery targets per system, test frequency and evidence, ransomware protections, certification scope, contract exit terms, references, and total cost including failover and support.

Why These Questions Matter More Than the Sales Pitch

Why These Questions Matter More Than the Sales Pitch

Disaster recovery is one of those services nobody wants to buy and everybody needs. That imbalance gives providers room to talk in vague, comforting language. “We’ve got you covered” sounds great until you ask what “covered” means in hours and minutes.

The core planning metrics here are Recovery Time Objective (RTO), the maximum acceptable downtime, and Recovery Point Objective (RPO), the maximum acceptable data loss measured in time. Both are defined in ISO 22300, the vocabulary standard for security and resilience. If a provider can’t state them per system, the conversation is marketing, not planning.

A useful framing: backup is a copy, recovery is a capability. Plenty of companies have backups and still fail to recover, because nobody tested the restore path under pressure. Your job in the hiring process is to separate the two.

The Questions That Separate Real Providers From Resellers

1. What are the RTO and RPO for each system, in writing?

Push for a table, not a sentence. Your payroll system, your customer database, and your email archive probably have very different tolerances. A provider that quotes one number for everything is either oversimplifying or hasn’t done the work.

Practical tip: ask them to state the targets in hours and minutes, then ask what happens if they miss. A credible provider has an answer about credits, escalation, and root-cause reporting. A weak one changes the subject.

2. Did you perform a business impact analysis, or will you?

A business impact analysis (BIA) identifies critical functions and their tolerable downtime. It is the standard first step before setting RTOs, and the NIST contingency planning guidance treats it as the foundation of the whole plan. If a provider wants to quote recovery targets without a BIA, they are guessing on your behalf.

Common mistake: letting the provider run the BIA alone. Your department heads know which processes actually stop revenue. Include them.

3. How do you test, and can I see the reports?

Ask for test frequency, scope, and the last three test reports. Testing, training, and exercises are a required phase of contingency planning in NIST SP 800-34 Rev. 1, the Contingency Planning Guide for Federal Information Systems. A schedule is not evidence. A report with timestamps, systems recovered, and issues found is evidence.

Watch for providers who only test backups (“the job completed successfully”) rather than full recovery. Those are different exercises, and only one proves you can get back to work.

4. Are backups immutable, and how is ransomware handled?

Ransomware changed the disaster recovery conversation. Modern attacks target backups first. Guidance from CISA’s #StopRansomware initiative recommends tested, offline or immutable backups alongside a documented recovery plan.

Ask three specific things: whether backups are immutable for a defined retention window, whether at least one copy is offline or air-gapped, and how restoration is verified after an attack. If the answer is “we have antivirus,” keep looking. For a broader view of what a solid recovery plan looks like in practice, the cloud disaster recovery best practices that actually survive a real incident are worth reading before you shortlist vendors.

5. Which standards or certifications do you hold, and what is the scope?

ISO 22301 is the certifiable standard for business continuity management systems. A provider claiming certification should be able to name its certification body and the scope of the certification. Scope matters: a certificate covering one data center says nothing about the region you’ll actually use.

Similarly, if personal data is involved, GDPR Article 32 requires appropriate technical and organisational measures, including resilience and the ability to restore availability and access to data after an incident. In US healthcare, HIPAA’s Security Rule requires contingency planning for electronic protected health information under 45 CFR §164.308(a)(7). Ask how the provider supports those obligations, not whether they’ve “heard of them.”

6. Who actually does the work, and who do I call at 3 a.m.?

Many providers subcontract. That’s not automatically bad, but you should know. Ask for the escalation path, the on-call structure, and whether the people answering the phone can actually start a failover. Test it during the sales process with a non-urgent question at an odd hour. The response tells you plenty.

7. Can I speak to references and see after-action reports?

Ask for named references, ideally in your industry and roughly your size. Then ask the references the uncomfortable question: “When something went wrong, how did they handle it?” Every provider has bad days. The good ones document them.

This is practical advice rather than a standard, but it’s the single most useful step in the whole process.

8. What are the exit and data-return terms?

Ask how your data is returned or destroyed at contract end, in what format, and how long it takes. Also ask what happens to your recovery environment if you stop paying mid-incident. These are contract-diligence points, and they’re where buyers get trapped. Vendors who resist these questions are telling you something.

Comparing Providers Side by Side

Use a simple table to keep sales conversations honest. Fill it in during calls, not afterward from memory.

What to compare Weak answer looks like Strong answer looks like
RTO/RPO One number for everything Per-system table with minutes stated
BIA “Not needed for your size” Documented BIA before target-setting
Testing Annual backup verification Regular full recovery tests with reports
Ransomware Antivirus and firewall Immutable, offline copies plus verified restores
Certification Logo on a slide Named body, defined scope, current certificate
References Logos only Named contacts, after-action reports
Exit terms Silent Written data-return format and timeline
Support Ticket queue Named escalation path, 24/7 on-call

If a provider can’t fill the right-hand column, that’s your answer.

Mistakes Buyers Make Again and Again

The most common mistake is buying recovery for the wrong systems. Teams often protect the servers that are easiest to protect rather than the ones that stop the business. A quick test: if the system vanished for 48 hours, would anyone outside IT notice? If yes, it belongs in scope.

The second mistake is accepting a single recovery target across the whole estate. It feels tidy and it’s almost always wrong. A payment system and an internal wiki don’t deserve the same treatment, and paying for identical targets wastes money.

The third is skipping the failover test. Buyers sign contracts, run a tabletop exercise, and call it done. A tabletop is a conversation. A failover is proof. Insist on at least one real failover during onboarding, before the contract’s first renewal.

The fourth is ignoring the human side. Who declares a disaster? Who authorizes the failover spend? Who tells customers? Providers that hand you a runbook without those answers are handing you half a plan. It’s worth reviewing the questions buyers ask before committing to a managed detection response checklist, because detection and recovery are two halves of the same problem.

Finally, buyers underestimate cost creep. Failover compute, egress fees, and support tiers can double the sticker price. Ask for a total cost estimate for a three-day full failover, rather than only the monthly base.

How to Run the Evaluation Without Wasting Weeks

  • Build a short list of three to five providers based on industry fit and references.

  • Send the same written question set to each, including RTO/RPO tables and test evidence.

  • Score answers against your own system inventory, not the provider’s template.

  • Ask each finalist to walk through one real test report line by line.

  • Run a paid pilot or onboarding test before committing to a multi-year term.

  • Review contract exit terms with someone who reads contracts for a living.

This process takes a few weeks and saves years of regret. It also gives you a document you can hand to leadership when someone asks why recovery costs what it does.

Where Recovery Plans Usually Break

Even well-chosen providers fail in predictable places. Documentation drifts out of date, so the runbook describes servers that no longer exist. Key people leave and take institutional knowledge with them. Dependencies get missed, like the identity provider nobody listed that every recovery step quietly relies on. For a longer look at where these plans fall apart, the common cloud disaster recovery problems that sink real recoveries are a sobering read, and the cloud disaster recovery examples with hard lessons from real incidents show what the failure modes look like in the field.

A good provider will talk about these openly. A provider that claims their plans never drift is either new or not paying attention.

Conclusion

Hiring a disaster recovery company is less about picking the biggest name and more about forcing specifics. Written recovery targets per system, a real business impact analysis, test reports you can read, immutable backups, named certification scope, honest references, and clean exit terms. Those eight questions will filter out most of the market in a single round of calls.

The uncomfortable truth is that you’re buying a capability you hope never to use, and the only way to know it works is to watch it work. Ask for proof, test the proof, and put the answers in the contract. If you want a companion checklist for the security side of the same decision, the AI chatbot security questions to ask before buying cover similar ground for a different purchase. Do the homework once, and the 2 a.m. phone call becomes a procedure instead of a panic.

FAQ

How long should a disaster recovery contract be?

Start with a term that matches your testing cycle, often one to three years. Anything longer should come with clear exit terms and price protections. If a provider pushes a five-year deal before you’ve seen a single test report, treat that as a warning sign rather than a discount.

Is cloud-based disaster recovery always better than a secondary data center?

No. Cloud recovery is often faster to stand up and cheaper at low scale, while a dedicated secondary site can make sense for large, stable workloads with strict control needs. The right choice depends on your RTOs, your compliance obligations, and how much failover you can realistically test each year.

How often should recovery be tested?

More often than most contracts state. At minimum, run a full recovery test annually and a tabletop exercise twice a year, plus a test whenever a major system changes. Testing, training, and exercises are a required planning phase in established contingency guidance, and systems drift quickly between tests.

What is a reasonable RTO for a small business?

It depends entirely on the system. Many small businesses target four to twenty-four hours for core finance and customer systems, and longer for internal tools. The number should come from a business impact analysis, not from a competitor’s brochure or a provider’s default package.

Does a backup service count as disaster recovery?

Usually not. Backup creates copies; disaster recovery restores operations, including infrastructure, access, and process. A vendor selling only backup may never deliver a working recovery. Ask specifically who coordinates the failover, who tests it, and who is accountable when it fails.

What should I do if a provider refuses to share test reports?

Walk away, or at least pause. Test reports are the closest thing to proof in this market, and refusing to share them suggests either no testing or poor results. A provider can redact customer names while still showing timestamps, systems recovered, and issues found.

on this page