Microsoft Forms is usually already there. Nobody signs up for it and nobody compares it against anything, because it arrives with the Microsoft 365 account that the organisation already pays for. That makes it the fastest way to get a survey out, and it also means the decision about what happens to the answers gets made by accident rather than on purpose.
This is a practical walk through both halves. First, building a survey that behaves the way it should when it reaches a phone. Second, what the tool does and does not do once the responses are in, because the second half is where the time actually goes.
What your account type decides before you start
Microsoft Forms lives at forms.office.com, and it accepts three kinds of sign in: Microsoft 365 school credentials, Microsoft 365 work credentials, or a personal Microsoft account. The editor looks identical in all three cases. The ceilings are not identical, and the gap between them is large enough to change a plan.
| Account signing in | Responses allowed on one form |
|---|---|
| Work or school (Microsoft 365) | Up to 5,000,000 |
| Personal Microsoft account, paid | Up to 1,000 |
| Personal Microsoft account, free | Up to 200 |
| GCC High and DoD tenants | Up to 50,000 |
Two hundred is not a lot. A conference sign up, a customer satisfaction sweep sent to a mailing list, an all staff pulse at a company of any size: all three clear 200 without trying. If the form was built on a personal account because that was the account already signed in on the browser, the survey stops collecting at a point nobody planned for, and the answers that did not make it are simply not there.
The fix is boring and worth doing before the first question is written: check which account the browser is signed in as, and rebuild in the work or school tenant if it is the wrong one. Rebuilding a twelve question form costs ten minutes. Discovering the cap two days into a campaign costs the campaign.
Other ceilings apply to everyone. A single user can hold up to 400 forms and up to 400 polls, and that count includes items sitting in the recycle bin. Each form holds up to 200 questions.
Building the survey
Select New Form. Give it a title and, if useful, a description. The title allows up to 90 characters and the description up to 1,000, which is more room than most surveys need and enough to explain who is asking and why. The form saves as you work, so there is no save button to hunt for.
Questions go in with Add new. Four types sit at the top level: Choice, Text, Rating, and Date. Behind More question types are Ranking, Likert, File upload, and Net Promoter Score. Section is in the same menu, and sections are how a long survey stops feeling long, because each one starts on its own screen.
Question text allows up to 4,000 characters, and so does each answer a respondent types. Multiple choice options allow 1,000 characters each. These are generous, and the practical limit is attention rather than characters. A rating question with a clear label gets answered. A 4,000 character preamble does not get read.
Question text can be formatted with bold, italic, underline, colour, size, numbering, and bullets. Use that sparingly. Formatting in a question stem reads as emphasis, and emphasis on three words in a row reads as no emphasis at all.
Then use Preview. It shows the form as it appears on a computer and on a phone, and it lets you answer the questions yourself and press Submit to check the whole path. The phone view is the one that matters, because that is where most survey links get opened, and it is where a matrix question that looked fine on a laptop turns into something nobody finishes.
The settings that decide how clean the data is
Most of the difference between a survey that produces usable answers and one that produces an argument is settled in the settings panel, before the link is sent.
Who can respond has three positions: anyone, inside or outside the organisation; only people in the organisation; or specific people in the organisation. Pick the tightest one that still lets the intended audience in. Opening a staff survey to anyone means the results can be diluted by anybody who forwards the link, and no amount of analysis afterwards will tell you it happened.
Record name decides whether the response carries an identity. This is the single setting that most changes what a survey is for. Anonymous responses get more honest answers to uncomfortable questions and cannot be followed up on. Named responses can be replied to and will be answered more cautiously. Choose deliberately, and say which one it is in the form description, because respondents who cannot tell will assume the worse of the two.
One response per person stops the same respondent submitting repeatedly. It is worth switching on for anything where the count matters, and worth leaving off for something like a daily check in where repeat submissions are the point.
Accept responses can be switched off to close a survey, and Start date and End date do it on a schedule, which is more reliable than remembering. A deadline can also be set with Set time duration. Use the dates rather than the toggle. The toggle depends on somebody being at a keyboard on the right afternoon.
Shuffle questions randomises the order, which matters for research and not much else. Show progress bar and Disable question number change how long the form feels without changing how long it is, and the progress bar is worth having on anything with more than about five questions.
Finally, Get email notification of each response sends an alert as each submission arrives, and Allow receipt of responses after submission lets respondents keep a copy of what they wrote. Both reduce the number of emails asking whether a submission went through.
Branching, and the rule that catches people
Branching makes the form change according to the answers given, so questions appear only when they are relevant. To add it, select More settings for question, then Add branching.
There is one hard rule, and it is the source of most broken Microsoft Forms surveys: you can only branch to a later question, never to an earlier one. Pointing question 4 back at question 2 does not loop the respondent around. It breaks the path, skipping questions 5 through 7 and dropping the respondent straight at the end of the form. They submit, the response lands, and nothing announces that two thirds of the survey was never shown.
So branching has to be designed forward, on paper, before it is built. Order the questions so that every route runs in one direction. Screening questions first, then the branches, then any question everyone answers, at the end. End of the form is a valid branching destination and it is the right one for a disqualifying answer, which is better than walking someone through fifteen irrelevant questions to be polite.
After building branches, test every route in Preview, not just the main one. Each combination of answers is a different survey, and the only way to know they all reach the end is to walk them.
Sharing the survey, and sharing the work
Sending the survey out is the easy part: a link, a QR code, an embed, or an email invitation.
Sharing the work is different. Microsoft Forms handles it through a share to collaborate link, which lets several people view and edit the structure, the layout, and the responses of a form someone else started. That covers the case where a colleague needs to fix a typo or pull the numbers.
What it does not cover is dividing the responses between those people. The collaborate link is all or nothing at the form level. Everyone with it sees everything and can change anything, and there is no field on a response saying which of them is dealing with it. For a survey that ends in a chart, that is fine, because nobody needs to be assigned to a chart.
It stops being fine the moment a survey has a follow up. Applications, support requests, event registrations that need confirming, interest forms that need a reply: these are surveys in shape only. Each row is a person waiting. Teams in that position usually end up dividing the exported spreadsheet by hand, which is the pattern worth naming before it starts, and it is the reason some form tools carry an owner and a status on every single response rather than on the form as a whole.
Reading the answers
The Responses tab gives a summary straight away: counts, averages, and a chart for each question. For a survey that exists to produce a number, that view is often the whole job, and the work is finished in the browser.
For anything more, export. Selecting the export option opens the responses in an Excel workbook, one row per submission. From there it is PivotTables and filters, the tools most people already know.
Be clear about what that export is. It is a snapshot taken at the moment of the click. Responses that arrive afterwards are not in it, and edits made in the workbook do not travel back to the form. If two people export on the same afternoon, there are now two workbooks that disagree, and a week later nobody can say which one was used to decide anything. For a one time survey this does not matter. For a form that collects continuously, the spreadsheet becomes the thing that has to be maintained, and maintaining it is unpaid work that grows with volume.
The same limit applies to replies. Microsoft Forms can send an automatic acknowledgement, but the actual answer, the one that says yes or no or asks a question back, gets written in Outlook. Nothing connects that email to the row it came from. Three weeks later the only record that a reply was sent is in the sender's own mailbox, which is why the two failures every team recognises keep happening: the person who receives two different answers, and the person who receives none.
Deciding whether a survey is really a survey
The useful question is not which tool is better. It is what happens on the day after the form closes.
If the answer is a chart in a deck, Microsoft Forms is already the right tool and there is nothing to change. It is included, it holds five million responses on a work account, and the summary view produces the chart without an export.
If the answer is that somebody has to contact each respondent, the form was never the hard part. Collecting is solved, and every tool in this category does it well enough that the differences are cosmetic. What separates them is whether a response can be given an owner, whether it can carry a status from arrived to answered to closed, and whether the reply can be sent from the same screen as the answers, so that the record of the reply lives with the response instead of in one person's sent items. Recruitment, course sign ups, enquiry handling and event registration all sit in this second group, and the shapes they take are fairly consistent across organisations.
Sort your forms into those two piles first. Most teams find the piles are uneven, and that a small number of forms are generating nearly all of the manual work.
What to change first
Check which account your most important form was built under, because a 200 response cap on a personal account is the one failure here that loses data instead of time. Then take the forms that end in a reply rather than a chart, and give those responses an owner and a status, whether that is a disciplined column in the export or a tool that keeps the submission and the reply on one screen. Everything else can stay where it is.
Q1. How many responses can one Microsoft Forms survey collect?
It depends entirely on the account that created it. A Microsoft 365 work or school account allows up to 5,000,000 responses on a single form. A paid personal Microsoft account allows up to 1,000, and a free personal account allows up to 200. GCC High and DoD tenants are capped at 50,000.
Q2. Can a Microsoft Forms survey send a different set of questions to different people?
Yes, through branching. Select More settings for question, then Add branching, and route each answer to the question that should come next. The important restriction is that branching only moves forward. Sending a respondent back to an earlier question causes the form to skip everything in between and jump to the end.
Q3. Can two people manage the responses to the same form?
Both can see and edit everything, using a share to collaborate link that covers the structure, the layout, and the responses. What is not available is a way to split the responses between them, because there is no owner field on an individual response. Teams that need that usually add a column to the exported spreadsheet and maintain it by hand.
Q4. How do you get Microsoft Forms responses into Excel?
Open the Responses tab and choose the export option. The responses open in an Excel workbook with one row per submission. Treat it as a snapshot rather than a live link: later responses are not included, and changes made in the workbook do not go back to the form.
Q5. Is Microsoft Forms free?
It is included with Microsoft 365 work and school subscriptions, and it is also usable with a personal Microsoft account at no cost. The free personal route is the one to watch, because that is the tier capped at 200 responses per form.
Q6. How many questions can a single form hold?
Up to 200 questions per form, with up to 4,000 characters in each question and in each typed answer. One user can hold up to 400 forms in total, and that count includes forms sitting in the recycle bin, so emptying it is occasionally necessary.