The survey closed, the Responses tab is open, and the charts look finished. That is the trap. The Summary view in Google Forms is an aggregate of each question taken on its own, produced for a quick look rather than for a decision, and it has documented points at which parts of it stop appearing altogether. Knowing where those points are, and which questions the view can never answer, decides whether the reporting job takes an afternoon or turns into a monthly rebuild.
What the Responses tab actually provides
Google Forms gives three views of the same data, reached from the Responses tab at the top of a form.
Summary is the aggregate. Each question gets its own chart or list, built from every response received. Google's documentation notes that the Summary tab only appears on a form once there is at least one response, which is why a freshly published form looks broken.
Question shows one question at a time with all of its answers, which is the view to use when reading a single free text field end to end.
Individual shows one submission at a time, with Previous and Next controls and a drop down to jump to a specific response. This is the view that matters for intake forms, where each submission is a person to deal with rather than a data point.
Charts in the Summary view can be copied out. Each chart has a Copy control at its top right, and pasting into Google Docs, Slides or Drawings keeps the chart connected to the source, so it can be updated from inside the document as long as the person has edit access to it. For a recurring report, that link is the most useful thing in the whole feature set, because it removes the habit of pasting screenshots that quietly go stale.
What none of the three views do is compare one question against another. There is no segment filter, no cross tab, and no way to exclude the three test submissions made while building the form. Every percentage shown is computed on everyone who answered that question, with no control over the base.
The four ceilings Google documents
This is the part most write ups leave out. Google publishes the points at which response volume changes what the Forms interface will show. The form keeps collecting either way, but the reporting surface degrades in stages.
| Responses on the form | What may stop working |
|---|---|
| Over 10,000 | The Question and Individual views may no longer appear |
| Over 10,000 | CSV downloads may no longer be sorted by submission time |
| Over 50,000 | The response summary may be unavailable |
| Over 100,000 | Responses may stop syncing to the linked spreadsheet |
For a survey that runs for two weeks and closes, none of this is likely to matter. For a contact form, an application form or a support request form that has been live for three years, all of it does, and the order in which it degrades is awkward. The Individual view is the first thing to go, and that is the view a person handling submissions uses every day.
The practical reading is that a Google form is reliable as a survey instrument and gets fragile as a permanent intake channel. Two responses to that are available. Either the form is retired and replaced on a schedule, which keeps each one well under the ceilings but scatters the history across several files, or the responses are moved to something built to hold them continuously. Deciding which is a reporting decision, even though it looks like an infrastructure one.
The cuts the summary cannot make
The gap between the Summary view and a usable report is almost always one of these four jobs.
Cross tabs. Nearly every real question is a comparison: how did the newer staff answer compared to the rest, did the people who rated the service poorly also mention the wait. The Summary view cannot express this, because each chart is built from one question in isolation.
Cleaning. Test submissions, duplicates from someone who submitted twice, and a respondent who typed a paragraph into the postcode field all need removing before percentages mean anything. There is no filter in the Summary view, so the cleaned version has to live elsewhere.
Free text at volume. The Question view will show every comment, but reading 300 of them in a browser column and counting themes by memory is how reports end up quoting whoever wrote the longest answer.
Per question base sizes. When a question is optional or sits on a later page, fewer people answer it. The chart still fills the width of the screen, and a reader who does not check the count under it will assume the whole sample answered.
All four jobs point at the spreadsheet. From the Responses tab, Link to Sheets creates a connected spreadsheet, and Download responses as a CSV file produces a static copy. Once the data is in Sheets, the cross tabs are pivot tables: select the range, then Insert and Pivot table, with the segment in Rows and the question in Columns. Each column in the source data needs a header for this to work. Pivot tables recompute when a response is corrected, which is the reason to use them instead of a column of COUNTIF formulas.
One access detail is worth knowing before that spreadsheet is shared. Google notes that a collaborator on the form may also have access to the form's linked spreadsheet, and that removing someone means removing them from the form and from the spreadsheet separately. For a survey with free text about managers or pay, that is not a minor footnote.
Assembling the report itself
With the cleaned data in Sheets, the document comes together in a predictable order, and the order matters more than the visual design.
Lead with the question the report answers and the headline finding, in one short block with no chart. Follow with three to five findings, each a full sentence stating a conclusion, with one chart or one small table underneath. Then the method block: how many were invited, how many answered, the dates the form was open, and how people were reached. Put it after the findings so it does not lose the reader, but never leave it out, because it is the first thing a sceptical reader looks for.
For the charts, the linked copies from the Summary view are good enough for simple distributions, and they refresh. Anything involving a segment has to be built in Sheets and inserted from there. Mixing the two sources in one document is fine as long as every chart carries its base size in the caption.
Close with next steps that name a person. A report that ends at the findings creates another meeting to decide what the findings meant.
Reporting on a form that never closes
A survey has a base size that stops moving. A live form does not, and reports built as though it did will contradict each other.
Fix a window before anything else. The linked spreadsheet records a submission time for every row, so a calendar month or an ISO week is easy to cut and easy to defend. Reporting on everything received to date sounds thorough and produces a number that only ever goes up, which tells nobody anything.
Freeze the previous period. If last month's figure came from a live pivot table, it will change as late submissions and corrections arrive, and next month somebody will notice that a number in a circulated document no longer matches the sheet. Copy the closed period's figures into a static tab, or paste them as values into the report, and treat that as the record.
Report counts alongside rates. On a continuous form, volume is usually the finding: 240 requests this month against 180 last month changes staffing, while a satisfaction average that moved from 4.1 to 4.2 changes nothing. Averages on an open form also drift for uninteresting reasons, such as a link being shared with a new audience halfway through the month.
Watch the denominator when comparing periods. A month with a public holiday, a campaign that drove a burst of traffic, or a fortnight when the form was linked from a newsletter are not comparable to an ordinary month, and the report should say so in one line rather than leave the reader to discover it.
What the report cannot tell you, and usually needs to
There is a second report that nobody asks for by name and everybody wants: not what people said, but what was done about it.
Forms has pieces of a workflow. Email notifications for new responses can be turned on per form from the Responses tab, the Form notifications add-on extends that to customised follow up messages, and if email addresses are being collected, respondents can be sent a copy of their answers either always or on request. Email collection itself has two modes, Verified, which confirms the signed in Google account address, and Responder input, which is typed and unverified. For a survey, that is enough.
For anything that arrives continuously, it is not, because none of it records state. A notification lands in several inboxes, and nothing in the response itself says who picked it up, whether a reply was sent, or whether the matter is closed. That is the mechanism behind the two failures every intake queue produces: the same person answered twice by two colleagues, and the one nobody answered at all. Tools that treat the owner and the status as part of the response exist to close that gap, and the monthly number they make reportable is the share of submissions closed inside a target, which is the figure a manager actually wanted when they asked for a survey report. The same pattern applies across applications, enquiries, event signups and support requests, where the volume is steady and the useful measure is throughput rather than an average rating.
Setting up the next form so the report is easier
Most reporting pain is created at form design time and paid for at reporting time. Four choices do most of the work.
Use choice questions wherever a choice will do. A dropdown with six options aggregates itself. A text field asking for a department produces six spellings of the same department and an afternoon of cleaning.
Ask one thing per question. Fields that combine two questions cannot be charted without splitting them by hand afterwards.
Add the segment as a question rather than inferring it later. If the report will compare teams, ask for the team. Working it out from email addresses afterwards is slow and often wrong.
Close the form deliberately. Under Accepting responses, a close date or a maximum number of responses can be set, along with the message people see afterwards. For a survey, closing on a date makes the base size final and the report reproducible. For an intake form, the same setting is the emergency brake when a link gets shared further than intended.
What to change first
Decide whether this form is a survey or a permanent intake channel, because the answer changes everything else. A survey can live inside the Summary view and a pivot table. An intake channel that has been running for years will hit the documented ceilings and, long before that, will need to record who handled each submission.
If it is the second kind, look at what a response looks like when it carries its own owner and status from the moment it arrives. Halict shows that on one screen, with the submission and the reply in the same place.
Q1. Can Google Forms produce a cross tab or filtered report?
Not in the Responses tab. Each chart there is built from one question across everyone who answered it, with no segment filter. Use Link to Sheets, then Insert and Pivot table, putting the segment in Rows and the question in Columns. Remember that every source column needs a header.
Q2. Why did the response summary disappear from the form?
Two documented reasons. The Summary tab only appears once a form has at least one response, and on very large forms parts of the interface stop appearing: the summary may be unavailable above 50,000 responses, and the Question and Individual views above 10,000. The form keeps collecting, and the responses can still be downloaded as a CSV file.
Q3. Do the charts stay up to date if they are pasted into a document?
Yes, if they are copied from the Summary view using the Copy control on the chart and pasted into Google Docs, Slides or Drawings. They stay connected to the source and can be updated from inside the document, provided the person has edit access to it. A screenshot pasted into a slide does not update and is the usual cause of a stale figure in a quarterly deck.
Q4. Is the linked spreadsheet private to the form owner?
Not automatically. Anyone collaborating on the form may also have access to the linked spreadsheet, and removing that access means removing the person from the form and from the spreadsheet as two separate steps. Check this before sharing a form that collects free text about people or pay.
Q5. Should responses be kept in Google Forms permanently?
For a survey that closes, keeping them there is fine. For a form that has been collecting for years, the documented ceilings arrive eventually, and the first view to stop working is the one used to read individual submissions. Either retire and replace the form on a schedule, or move continuous intake to a system designed to hold it.