Check backup recovery by restoring one forecast file and one CRM slice to a separate location, then matching the totals against numbers you already trust.
Key takeaways
- Treat confidence in a backup as a claim to test, not as proof that a restore will work.
- Restore one forecast file and one CRM slice into a separate location, and time the work.
- Compare the restored copy with totals leadership already knows, such as last quarter's closed-won total.
- Keep the people who can delete backup copies separate from the people who edit the forecast.
- Define "recovered" before you test: complete, current enough to use, and trusted by the people who sign the number.
Every revenue leader has a version of the same fear. The forecast file gets overwritten, a CRM export goes missing, or a territory plan disappears a few days before it is due, and the number that leadership expects on the last day of the quarter turns into a guess. Having a backup is not the same as checking whether it can be brought back quickly, completely, and in a form people can trust.
This post walks through four figures from published backup and recovery research. Each one points to a gap between having a backup and recovering from one. All four figures appear in a Backblaze article on ransomware recovery, which cites a Veeam survey, Veeam's own research, and Backblaze's State of the Backup Report. Those sources describe IT teams and whole organizations, not sales operations, so read them as a pattern to test against your own setup rather than a forecast of what will happen to you. The question that matters for a revenue leader is simple: if the forecast workbook and the CRM records disappeared today, how long would recovery take, and would the restored copy be complete enough to plan on?
Figure 1: 90% say they can recover
A Veeam survey of more than 900 senior IT, security, and risk leaders found that 90% are confident they can recover from a cyber incident. That is a strong majority. It is also a self-assessment. Confidence describes how people feel, not what happens when they actually try to restore.
Revenue teams often have the same kind of comfort. The forecast may sit in a shared drive with version history. The CRM is hosted elsewhere, so someone assumes it is covered. Someone set up a backup tool a while ago, and nobody has opened it since. Each of those things may be real, but none of them shows that a specific forecast model or a specific set of opportunity records will come back in the shape the team needs on the first working day after a problem.
Test the confidence against evidence. Pick one forecast file and one slice of CRM data, such as the open opportunities for a single region. Ask someone who did not set up the system to restore both into a separate location. Time the work. Write down what was missing or hard to find. A result that surprises you is far more useful than a comfortable feeling that has never been tested.
Figure 2: only 28% fully recovered
This is the figure that should change how you plan. Among organizations hit by ransomware that affected their operations or data, only 28% fully recovered it, and 44% got back less than three-quarters. Read that carefully. Fewer than three in ten organizations got everything back. Just under half got back less than three-quarters of it.
The group that got back less than three-quarters deserves the most attention. A partial restore looks like success at first. The file opens. Most accounts are there. Then someone notices that late-stage deals from one region are missing, or that close dates on a handful of opportunities have reverted to older values. A forecast built on partial data is not a forecast with a small error. It is a different number, and it will be presented with the same confidence as the right one.
So when you test a restore, do not stop at whether the job finished. A restore can complete without errors and still be incomplete. Pick a number you already trust, such as the total of closed-won deals for last quarter or the count of open opportunities in one stage, and check that the restored copy matches it. If the numbers do not match, you have learned something important before it matters. For the CRM side of deleted records, see Recover deleted sales records: 4 CRM restore rules.
Figure 3: 96% of ransomware attacks target backups
The third figure is about where attackers go. Veeam's own data protection research puts the number at 96% of ransomware attacks specifically going after backup repositories, and most of those attempts succeed. The backup is itself a target, not only a safety net.
For a revenue team, the practical problem is independence. If the account that edits the forecast can also delete the backups, then one compromised login can damage both the working file and its safety copy. Backups that share the same credentials, the same permissions, or the same administrator are less separate than they look on a diagram.
Ask three questions. Who can delete or change backup copies, and is that list shorter than the list of people who can edit the forecast? Does anyone need a second approval before a backup is removed? Could a single compromised login reach both the live CRM data and its copies? The answers do not need to be perfect before quarter-end, but they should be written down and reviewed.
Figure 4: only 9% say restoring is very easy
The last figure is about effort. Only 9% of IT teams describe restoring from backup as "very easy," according to Backblaze's State of the Backup Report. In other words, most IT teams do not describe restoring as very easy. There is little reason to expect a revenue operations team to find the job easier, especially when the data is spread across a CRM, shared spreadsheets, and a few exports.
Scale changes the job too. The same article makes the point well: "restoring 40, 100, or 300 individual laptops without your entire IT team dropping everything else is a different problem." A single forecast file is a small restore. A shared model that dozens of managers rely on, or a CRM with years of linked activity, is a bigger job, and it pulls people away from their normal work.
Define recovered before you test
Before you run a restore, agree on what recovered means. Without a definition, a team can declare success while the forecast is still missing some of its deals. A useful definition has at least three parts: the content is complete, the content is current enough to use, and the content can be trusted by the people who will sign their name to the number.
Completeness means every record the number depends on is present, including the historical fields that explain how a deal moved. Currency means the restored copy is from a point that is close enough to the problem to be useful, and you know which point that is. Trust means someone outside the restore team has compared the copy with figures they already rely on. Write these three tests into the plan so that nobody has to invent them during an incident.
The currency test is easy to skip. A restored forecast that is older than the live file may be fine for a pipeline review and wrong for a commit call. Note the point in time for every restored item, and say that point out loud when you present the result. Hidden age differences can cause real confusion when people compare numbers from different versions of the same file.
Illustrative example: a mis-filtered export
Suppose a team keeps its quarter-end forecast in a workbook, pulls CRM opportunity records and saves the export beside the workbook. One afternoon the export runs with the wrong filter and overwrites the saved copy. Nobody notices until a manager asks why one region's pipeline dropped.
Version history covers the workbook, but the export sits in a folder a sync tool mirrors, so the synced copy is overwritten too. Earlier versions exist only for the workbook. This is the gap: a copy exists, but not one that brings back what the forecast needs.
A short test would have caught this. Restoring the workbook and the export into a separate folder, then checking regional totals against last week's review pack, would show the gap. The fix is partly technical, such as keeping exports out of a mirrored folder, and partly procedural, such as checking the export filter before saving.
Common mistakes that hide gaps
Several habits make recovery problems harder to see. Each one feels reasonable on its own, and each one hides a gap until the day it matters. Watch for these patterns in your own team.
Treating sync as backup. A shared folder or a synced CRM view can copy a mistake as quickly as it copies good data. A deleted row, a bad import, or an overwritten formula may appear everywhere at once. A copy that updates automatically is useful for access, but it is not the same as a copy that stays put.
Testing only the largest system. Teams often test the CRM because it feels like the important system, while the forecast lives in a spreadsheet that may never have been restored. Test every place where a quarter-end number is built, including the small workbook that combines regional files. Small files are often the ones with the most manual edits and the least documentation.
Trusting the vendor's word alone. A vendor may describe its own protections clearly, but your team still owns the question of whether your data comes back the way you need. Read what a vendor covers and what it leaves to you, and write down the gaps. Do not assume that a service tier or a product feature answers the restore question on its own. For one vendor's backup checks, see Microsoft 365 backup: 3 backup and recovery checks.
Confusing a successful job with a successful restore. A status page that says a job finished is a useful signal, but it describes the job, not the data. Pair every status with a check of the output: row counts, a sample of records, and one total that the business already knows.
Keeping the know-how in one person. The restore works because one analyst knows which folder to use and which password to ask for. When that person changes roles or goes on leave, the team discovers that the steps were never written down. Knowledge kept in one head is a gap that no backup tool can close, and it is among the cheapest gaps to fix.
Who should own the restore test
The person who built the forecast should not be the only person who tests it. Builders know where the sensitive pieces are, which means they tend to restore the parts they already understand. An outside check tends to find the missing pieces faster. Split the work so that one person runs the restore and a second person checks the result against the business numbers.
Name an owner for each system, not just for the whole process. The CRM administrator may own the records, the revenue operations lead may own the forecast model, and a finance partner may own the agreed totals. Each owner should know what they are responsible for when a restore is needed, and who else must be told. If nobody can say who approves a restore, the team has a process on paper only.
Decide who has the authority to pause edits during a restore. If the forecast keeps changing while a restored copy is being checked, the two versions will drift apart and the comparison becomes meaningless. A short freeze on the affected files, agreed in advance, keeps the test honest.
Why the timing matters
Quarter-end compresses everything. Finance wants a number, sales leaders want commit calls, and operations is closing out records at the same time. A restore that would be an annoyance in a quiet week becomes an emergency when the close calendar is full. The cost of a failed test is small. The cost of discovering the gap during close can be large, because every decision made from the wrong number has to be unwound.
This is why the test belongs early in the quarter's last month, or earlier, rather than in the final week. It gives you time to fix what fails and to run the test again after the fix. A team that runs its restore check once, sees a clean result, and then changes the CRM, the forecast layout, and the reporting tool in the same week has a fresh set of unknowns.
Turning the figures into a quarter-end check
Each figure points to a specific check. Here is a practical sequence you can run before the close week starts, without buying anything new.
- List what must survive. Write down every file and data set the quarter-end number depends on: the forecast workbook, any scenario tabs, the CRM opportunity records, activity history, and the exports used in pipeline reviews. If you cannot list it, you cannot test it. For weekly export steps, see Export a weekly pipeline snapshot in 4 systems.
- Restore into a separate place. Pick one item from the list and restore it to a location that does not overwrite live data. Have someone other than the usual owner do the work, and time it. Estimate how long the restore should take before you start; the gap between that estimate and the actual time is the number to watch.
- Reconcile against numbers you trust. Choose two or three totals that leadership already knows, such as last quarter's closed-won total or the count of open deals in a late stage. Check the restored copy against them. Any difference is a finding, not a rounding issue.
- Separate who can edit from who can delete. Review the people and service accounts that can remove backup copies. Make sure a normal editing login cannot also erase the copies, and that removing a backup requires a deliberate second step.
- Write the restore steps down. A procedure that lives in one person's head fails when that person is on leave. Record where each item is stored, who runs the restore, what access they need, and how the result is checked.
- Re-run the test after changes. Any change to the CRM, the forecast model, or the people who maintain them can alter the restore path. Repeat the test when that happens.
Talking to leadership about the risk
Leaders do not need a lecture on ransomware. They need a clear statement of what could go missing, how long a restore would take, and what the team has done to check it. Frame the conversation around the quarter-end number, not around the technology. Which decisions depend on the forecast? Which would be delayed if the CRM were unavailable for a week?
When you report, separate what you know from what you assume. You know the time one restore took. You may not know how long a full restore would take under pressure. Say so. A leader who hears a clear range with stated assumptions can plan. A leader who hears a confident promise that later fails will remember the promise, not the caveat.
Bring the 90% figure into the conversation with care. High confidence in the industry is not evidence about your team. The useful comparison is internal: the confidence in your own plan, compared with the result of the last test. Present the gap between them as the thing to be closed, not as a failure of anyone in particular.
What to do this week
The four figures share one lesson. Confidence, a completed job, and a backup product on the purchase order are not the same as a restored forecast that matches last week's numbers.
This week, choose one forecast file and one CRM slice, and run the restore test from the list above. Put the result in front of the people who depend on the number. If the test fails, that is a reason to fix the process now, while there is still time before quarter-end, rather than during it. Keep the result in the same place each time, so the next test can be compared with this one.