Microsoft 365 Backup publishes a median rate of up to 1-3 TiB per hour, depending on item counts and the size of the content. The page pairs that rate with a scale of up to 1,000 average-sized accounts, sites or mailboxes.
Key takeaways
- Treat the 1-3 TiB per hour figure as a median reference point, not a promised speed.
- Count the items and total size of the forecast site, each OneDrive and each mailbox before you plan.
- Plan at the slow end of the rate and put the forecast site first in any run order.
- Test a restore on a non-critical item before quarter-end, and time the steps.
Why these figures matter at quarter-end
Quarter-end is when a revenue team's files change the most. The forecast workbook gets edited every afternoon, rep mailboxes fill with deal threads and customer replies, and reps keep draft account plans in their own cloud storage. If a backup is running at the wrong moment, or takes longer than anyone planned, the copy you need may be out of date on the day of the forecast call.
For a small sales team, the data that matters is not one large archive. It is a few shared files that many people edit, plus individual mailboxes and drafts that only one person touches.
Microsoft's overview of Microsoft 365 Backup describes the product as providing "business continuity peace of mind by providing performance and reliable restore confidence." That is a statement of purpose. The planning figures sit lower on the same page, and they are the reason this post exists.
Find out whether your team's backups run from the admin center or from a partner application. The timing you plan around should come from the tool that actually runs your backups, not from a guess.
This post follows one example case from start to finish. The example team is a small sales team. Its quarterly forecast lives in one SharePoint site. Each rep has a mailbox, and each rep also has a OneDrive account where draft account plans are kept. The team is far below the 1,000 scale that the page uses, but scale alone does not tell it how long a backup will take. The page gives a rate and says it depends on what is being protected, so the team has to work through its own content.
The three figures on the overview page
The page gives three figures you can plan with: a scale, a rate, and the factors that move the rate.
The scale is the first. Each kind is given as up to 1,000 average-sized items, so a team of this size sits well below that ceiling.
The phrase "average-sized" needs attention. The page is describing average-sized accounts, sites and mailboxes. If one rep has an unusually large mailbox, or the forecast site holds an unusually large number of files, that item may not behave like an average one.
The rate is the second. The page gives "a median rate of up to 1-3 TiB per hour" for each of the three kinds of content. Two words do most of the work. "Median" means the figure describes a typical case rather than a guaranteed one. "Up to" means the top of the range is a ceiling, not a floor. A run can be slower than the top of the range, and no run is promised to reach it.
The factors are the third. For OneDrive accounts and sites, the page ties the rate to "site item counts and the size of the site." For mailboxes, it ties the rate to "the item counts and sizes in the mailbox." These are the things you can measure before quarter-end. Counting files and checking sizes is the work that turns the page's general figures into something you can use for planning.
Setting the rate against your own data
The rate is measured in TiB per hour, which makes the planning arithmetic easy to check. A figure in the 1-3 TiB per hour range is a median for typical content, so it is a reference point, not a promise. The following step is to set your own totals against it.
The table puts the three figures side by side for each kind of content, along with the factor the page names for it (Microsoft Learn overview):
| Content type | Scale figure on the page | Rate figure on the page | What the page says the rate depends on |
|---|---|---|---|
| OneDrive accounts | Up to 1,000 average-sized accounts | Median rate of up to 1-3 TiB per hour | Site item counts and the size of the site |
| Sites | Up to 1,000 average-sized sites | Median rate of up to 1-3 TiB per hour | Site item counts and the size of the site |
| Mailboxes | Up to 1,000 average-sized mailboxes | Median rate of up to 1-3 TiB per hour | The item counts and sizes in the mailbox |
The example team's work is to replace the page's general terms with its own numbers. It needs the item count and total size of the forecast site, the same for each rep's OneDrive, and the same for each mailbox. With those in hand, it can see which rows it sits in comfortably and which individual items are unusual for their kind.
Once you have those numbers, the calculation is a simple division. Take the total size for one kind of content and divide it by the rate you plan at. The result is a rough run time for that kind. Do the sum for each kind separately, because each has its own factors.
Ask your admin whether the three kinds of content run one after another or side by side in your tenant. The answer decides whether you add the three run times together or plan around the longest one.
That second check matters more than the first. A team can be well under the scale and still have one mailbox with a long history of attachments. The page ties the rate to item counts and sizes, so one unusually large mailbox can matter more to the plan than the average of the rest. Flag those items now, while there is time to act on them, rather than in the last week of the quarter.
Planning the quarter-end window
Start with an inventory. List every protected item that matters to the forecast: the site, each rep mailbox, and each OneDrive. Next to each one, write the item count and the size. If the admin tools do not give exact figures, use the best numbers you have and mark them as estimates, so nobody later mistakes a guess for a measurement.
Then plan using the slow end of the rate. The page's rate is a median with an upper figure, and item counts and sizes can slow a run down. Planning at the top of the range means a run that is slower than expected pushes the backup into the forecast week. Planning at the slow end gives you slack. The cost is a longer window on paper, which is a small price for a backup that finishes when you need it.
If the runs do go one after another, put the forecast site first. It is the item the call depends on most, so it should be the one that is already protected if the window runs short. Mailboxes and personal drafts can follow.
Set the target completion date well before the final days of the quarter. The point is to leave room to rerun a backup that did not finish, or that finished with an item you did not expect. If the forecast call is the thing that needs the backup, the call date is the real deadline, and the completion date sits ahead of it.
Ask whether a run can overlap with heavy editing. It is unclear how a running backup interacts with active edits, so do not assume they are independent. Put that question to your admin. If the answer is unclear, schedule the run for a quieter stretch of the quarter.
Last, give the window an owner and write it on the forecast calendar. Quarter-end plans fail when nobody is clearly responsible for confirming that a run finished. The owner checks completion, records how long the run took, and notes any item that ran noticeably slower than the rest. Those notes become the numbers you use next quarter, which is worth more than any general figure. For the restore side, see backup recovery before quarter-end.
Restores are a separate question
Backup speed and restore speed are different questions, and the figures above cover only backup. The page describes both as strengths. It says that "Microsoft 365 Backup provides ultra-fast backup and restore capabilities by creating backups within the protected services' data boundaries." It also says the product "not only provides uniquely fast recovery from common business continuity and disaster recovery (BCDR) scenarios like ransomware or accidental or malicious employee content overwrite or deletion."
Those are claims about what the product is designed to do, not a time to restore a forecast workbook, so do not reuse the backup rate as a restore estimate. A median backup rate of 1-3 TiB per hour describes the pace of a backup run. It does not tell you how long it will take to get one file back on the afternoon before a call.
The realistic restore case for the example team is a rep who overwrites the forecast workbook shortly before the call. What matters is how quickly the team can get the prior version back and who has to wait for it. Planning for that means knowing who can start a restore, how they would find the right version, and how long the restore actually takes in your environment.
Write the restore steps on one page and keep that page with the window plan. Otherwise the owner works out the steps for the first time in the middle of a dispute over a workbook, which is the worst moment to learn them.
The only reliable way to learn restore time for your own content is to test it. Before quarter-end, pick a non-critical item of each type, restore it to a safe location, and time the steps. Write the result down with the date, so the number can be traced later. A test on a small item does not prove that a large restore will take the same time, but it does show whether the process works and where the delays come from. For the restore-point figures, see Microsoft 365 Backup restore points: 3 figures to read.
Questions to put to your admin
Before you commit to a window, ask your Microsoft 365 administrator these questions:
- Do our backups run from the Microsoft 365 admin center, or through a partner application? Which one controls the schedule?
- What are the item counts and total sizes for the forecast site, and for the largest rep mailboxes and OneDrive accounts?
- Which of our items are well above average size for their type?
- Can a backup run overlap with heavy editing on the same content?
- Who can start a restore during the quarter-end window, and who approves it?
- When did we last test a restore, and how long did it take?
Each answer changes the plan. If the admin cannot provide item counts, the window needs more slack. If one mailbox is very large, it may need its own row in the plan, with its own completion date, rather than being hidden inside the total for the team.
Telling the team
Reps do not need the technical detail, but they do need to know what to do during the window. Tell them when the backup window starts and ends, and ask them to avoid bulk-moving or renaming forecast files while it runs. Keep the forecast in the one site the plan covers, because a file parked somewhere else may fall outside the plan altogether.
Ask reps to tell the owner at once if a file looks wrong or if they cannot find something they expect to see. A quick report on the day of the problem is far more useful than a discovery during the forecast call. Keep the message short, name the owner, and say where the restore request goes. Reps who know the path will use it, and reps who do not will guess.
Add one more line to the note: the window itself. Say when it opens, when it closes, and who to ask about it. Without that line, a quarter-end week tends to turn into a string of separate questions to the admin, and each one takes time away from the forecast.
Putting it together
The page gives a small sales team three figures to plan around: a scale of up to 1,000 average-sized accounts, sites, or mailboxes; a median rate of up to 1-3 TiB per hour; and the factors that move the rate, which are item counts and sizes. Those are the planning inputs. The example team uses them the way any small team should. It inventories its content, plans at the slow end of the rate, sets a completion date with room to spare, and confirms with an admin which tool runs the backups.
Restores need their own plan. The page's claims about fast recovery describe what the product is built to do, not how long a specific restore will take. Test a restore before quarter-end, time it yourself, and write the result down.
Your next step
Today, find out which tool runs your backups and write down the item count and total size of the forecast site. Those two numbers start the inventory described above.
Frequently asked questions
Is 1-3 TiB per hour the speed my backup will run at?
No. It is a median rate of up to 1-3 TiB per hour, so a run can be slower, and item counts and sizes affect where yours lands.
What does the up to 1,000 figure mean for a small team?
It is a ceiling for average-sized accounts, sites or mailboxes, so a small team sits well below it. One unusually large mailbox can still behave differently from an average one.
Can I use the backup rate to estimate a restore?
No. The backup rate describes backup runs only, so test a restore on a non-critical item to learn how long it takes in your environment.