Skip down to the research record

Microsoft 365 backup: 3 backup and recovery checks

Quota Forecast Editorial Team · Published · 12 min read

Before a mailbox backup, check the mailbox cohort and transfer rate, then confirm the separate SharePoint support-request window for deleted files.

Key takeaways

  • Count mailboxes and identify unusually large accounts.
  • Measure data volume and item counts before estimating time.
  • Run a timed backup test before scheduling the full job.
  • Confirm the support-request window's start point with your administrator.

Are the forecast files, proposal folders, and pipeline notes safe? And can the mailboxes that hold customer threads be recovered if someone deletes or damages them at the wrong moment? A backup only answers those questions if it can cover the data in the time you have. Microsoft describes Microsoft 365 Backup as designed to ensure your organization's data is always protected and easily recoverable. That is a goal for the service, not a promise about your particular job, so the details matter.

Picture a sales operations lead on the Monday before close. The pipeline review is on Thursday, three regional managers have shared inboxes that hold most of the open deal threads, and someone just asked whether the last month of proposals can be recovered. The lead has a backup plan on paper but has not checked whether it fits the scope, how long it will take, or what happens if a shared file is deleted during the same week. That gap is what this post is about.

This post walks through three backup and recovery planning checks before you start a mailbox backup. The first two concern Microsoft's backup performance figures; the third concerns separate SharePoint recovery rules. The first is the mailbox cohort associated with the performance figure. The second is how fast data moves, and why item counts matter as much as total size.

None of these limits means you should skip the backup. They mean you should measure your scope before you schedule the work, and leave yourself room to react if the numbers come back different from what you expected.

Check 1: the mailbox cohort behind the performance figure

The overview says Microsoft 365 Backup can back up all or select SharePoint sites, OneDrive accounts, and Exchange mailboxes, and that this gives comprehensive coverage for critical data. Broad coverage is the reason to use the service. It is not the same thing as unlimited capacity for every job you can imagine, and it does not remove the need to size your own scope.

The same page describes a cohort for its mailbox performance figure: up to 1,000 average-sized mailboxes. Two parts of that sentence carry most of the meaning.

The first is "average-sized." The figure describes a typical mailbox, not every mailbox in your company. A revenue team usually has a mix. Some people have modest mailboxes. Others have years of attachments, long-running customer threads, or large shared inboxes that many reps write to. If your scope includes a few heavy mailboxes, the average-based figure tells you less about those accounts, and you should plan them separately instead of letting them disappear into an average.

The second is "up to." Here it describes the mailbox cohort associated with the performance figure, not an absolute capacity ceiling. Reading it as a guarantee would be a mistake. Compare your own count and mailbox sizes with that cohort before using the associated rate in your schedule.

Start by listing every mailbox in scope. Include the obvious ones, such as active sales representatives and managers, and the less obvious ones, such as shared sales inboxes, mailboxes of departed employees you are keeping for records, and any mailbox a deal desk or finance team uses for contract approvals. Many teams forget the shared and departed accounts until the backup is already planned, and those are often the accounts that matter most at close. A list built from the directory of active users will miss them.

Once you have the list, mark the outliers. You do not need exact sizes for every mailbox on the first pass. You need to know which accounts sit well above the rest, so that the planning math does not quietly assume they are typical. Then decide whether to group the mailboxes by expected size. A smaller group of heavy mailboxes is easier to reason about than one long list averaged together, and it lets you test the heavy group on its own.

Finally, compare the total with the documented cohort. A lower count alone does not establish the pace your backup will achieve. For a larger cohort, confirm expected performance with your administrator rather than extending the quoted rate by assumption. Splitting the work into groups can make testing easier, and it also makes it easier to tell which group caused a slow run or an error.

Write the final count down with the date you took it. Scope drifts. People join, leave, and get new shared inboxes during a busy quarter, and a count from three weeks ago may no longer describe what the backup will see.

Check 2: data volume and item counts set the pace

The overview pairs the mailbox cohort with a median rate of up to 1-3 TB per hour, depending on the item counts and sizes in the mailbox. The same page gives a matching rate for SharePoint sites and OneDrive accounts, each with its own count line. Check the figure for the workload you are actually backing up rather than assuming one applies to all of them.

Read that sentence carefully, because it has three parts that change how you plan.

The rate is a median, which means it describes a typical case. The rate is also "up to" a level, which means it describes a ceiling. Neither word promises that your backup will move at that speed. A median with an upper bound is a useful reference point, but it is not a schedule, and it should not be the only number in your plan.

The rate depends on item counts and sizes. This is the part many people miss. Total gigabytes are only one input. The number of items in a mailbox also affects the pace, so a simple "we have 3 TB, so it takes X hours" estimate can mislead you. Two mailboxes with the same total size can take different amounts of time if one holds far more individual items than the other. A shared inbox that receives many small automated messages can have a very different profile from a personal mailbox with the same total size.

The rate is a volume figure, not a count of accounts. You need both numbers. Your account count helps you compare your scope with the documented cohort. Your total data and item profile tell you how long the work could take. Teams often check only one of these and then feel surprised by the other.

Illustrative example: estimate a test window

Suppose a group holds 3 TB. At an illustrative rate of 3 TB per hour, the arithmetic gives an hour; at 1 TB per hour, it gives three hours. Neither calculation establishes a deadline for a real backup. Plan around the slower end until a test shows otherwise, and keep a buffer on top of that for retries and reruns.

This arithmetic is only a starting point. It assumes the rate applies evenly across the whole group, and it does not account for a single heavy mailbox that slows the group down, or for a run that has to be restarted. That is why the next step matters more than the math.

A test is the best way to replace the estimate with evidence. Pick a representative group, not only your smallest mailboxes and not only the heaviest. Include at least one shared sales inbox and one mailbox with a large item count. Run the backup, note the start and end times, and record the total data protected. Then compare the observed pace with the range from the overview. If the observed pace is well under the lower end of the stated range, that gap is worth raising with your administrator before the full run.

Schedule the test early enough that a second attempt is possible. A test run on the Friday before close leaves no room to fix anything. A test run two or three weeks earlier gives you time to adjust groups, change the schedule, or ask for help. Treat the test as the cheapest insurance you will buy this quarter.

Check 3: deleted files and the support-request window

The third limit is different. It is not about how fast a backup runs. It is about what you can recover after something has been deleted, and how long you have to ask for it. Shared sales files are a good example. A rep deletes a folder of proposals by mistake, and the team needs it back before the quarter review. Those are the moments when a team learns whether its recovery plan works. For files attached to deals, also check where to find deal attachments.

Microsoft's guidance on restoring deleted items from the SharePoint site collection recycle bin sets out two limits that matter here. The first is the size of the recycle bin itself. The page says the space allocated to the site collection Recycle Bin is 200% of the site collection quota. The bin is sized relative to the site's quota, so it is a finite place to recover from. Do not treat it as a permanent archive for everything your team has ever deleted.

The practical point is that a recycle bin is a recovery layer, not a substitute for a backup. It can catch a recent mistake quickly, but it has a fixed size, and you should not plan a quarter around the assumption that every deleted item will still be there. If your team relies on shared sales folders for pipeline and proposal history, ask whether those sites have a backup in addition to the bin, and what each one is expected to cover.

The second limit is time. The same page describes an escalation path. If content cannot be restored through the Recycle Bin or the feature specified in Restore your OneDrive, an administrator can contact Microsoft Support to request a restore inside the 14-day window. The key words are "any time inside." Once that window closes, the request path described on the page may no longer help, so the clock is the real constraint, not the bin's size.

Read the page for how the window is counted, and do not assume it starts on the day someone first noticed the problem. Confirm the start point before you need it. Escalate immediately and confirm the documented start point with your administrator; the window could already be running when you discover the deletion.

Write down who the administrator is, how to reach them, and what the escalation looks like before you need it. A deleted-file incident should reach an administrator on the same day it is discovered. A message that sits in a general queue for several days can use up a large part of a 14-day window before anyone with access has looked at it. Name a person, give them the steps, and test the contact path once during a quiet week. If the loss concerns a deal record, use the separate sales-record recovery checks.

The backup and the recycle bin answer different questions. The backup gives you a recovery point you can plan around, with a scope you have measured. The recycle bin is a sized layer that handles recent mistakes. The escalation path is what you use when the layers above it have not worked. Plan for all three, and know which one you are relying on for each kind of loss.

Build a pre-close checklist

Use this list in the week before you start the mailbox backup. Each item maps to one of the limits above, and each one takes a short amount of time compared with recovering from a missed step during close week.

  • List every mailbox in scope, including shared sales inboxes and any departed-employee mailboxes you are keeping for records.
  • Compare the account count with the documented cohort, and flag any mailbox that is clearly not average.
  • Total the data for each group and note which groups have unusually high item counts.
  • Apply the documented rate to each group as an estimate, and plan around the slower end.
  • Run a timed test on a representative group, and compare the observed pace with the stated range.
  • Confirm which SharePoint sites and OneDrive accounts hold shared sales files, and check how each one handles deleted content.
  • Write down the administrator contact and the steps for requesting a restore through Microsoft Support.
  • Note the support-request window and its confirmed start point in your runbook, and escalate any deletion immediately.
  • Set a cutoff date for scope changes, so the list does not drift during close week.
  • Record anything you could not verify, so the next person knows where the gap is.

Each of these items is small. Together they turn a general worry about backups into specific answers. You will know your account count, have a tested estimate of transfer time, and have recorded the support-request window with its start point confirmed separately.

Read the official pages with your own scope in mind

The official pages are the right place to confirm the details, and they should be checked again before you rely on them. Product documentation changes over time, and a figure that was accurate when a team last looked may have been updated since. Note the date you checked each page, and link the specific page in your runbook so the next reviewer can see where each number came from.

It also helps to separate what the documentation covers from what your tenant does. Your administrator knows your licensing, your settings, and your current scope. The documentation describes the general behavior of the service. Ask your administrator to confirm how the figures apply to your setup, and to tell you what they will do if a run falls behind during close week.

Bring the numbers to your administrator

Quarter close is a poor time to learn a backup's limits. Use the inventory and timed test to replace vague worries with a mailbox count, an observed transfer rate, and the window for requesting a Microsoft Support restore, with its start point confirmed separately.

Bring those numbers to your administrator with the list of mailboxes, the test results, and the escalation contact. Ask three direct questions. How does our scope compare with the documented mailbox cohort? Does the observed pace match the range on the overview, or is something slowing us down? If a shared sales file is deleted on the Wednesday before the review, who will open the request, and when?

If the answers are clear, you can plan the backup with confidence. If they are not, you still have time to adjust the groups, move the schedule, or change who owns the escalation path. The goal is not to find a perfect backup. The goal is to know, before close, what the backup can and cannot cover, so that the decisions your team makes in the last week of the quarter rest on facts rather than assumptions.

Today, list the mailboxes in scope and ask your administrator to choose a representative group for a timed test.

Related guides: File size limits: 8 checks before evidence uploads · Check the buying company on a deal: 5 record limits · Deal formula limits: 70 parentheses, 40 fields, 10 fields.

Questions before the backup

Is the mailbox count an absolute capacity limit?

No: it describes the cohort associated with the quoted performance figure. Compare mailbox sizes and item counts as well as the total count.

Does the support-request window guarantee recovery by its end?

No: it is a window for requesting a restore, not a deadline for completing recovery. Confirm its start point and escalate immediately.

Sources