Key Takeaways
- Building the form is the easy part. Response rate and data quality are decided by the question set you write before you open the editor, and by three settings buried in the Settings tab.
- Question count is the biggest lever on completion. Across 267,564 responses, Survicate found 83.34% completion at one to three questions and 41.94% at fifteen or more.
- Open-ended questions cost you answers. Pew Research measures 1% to 2% item nonresponse on closed questions against a median near 18% on open ones. One open end, at the bottom, is usually the right budget.
- Collect email addresses and Limit to 1 response both change who can reply. Either one pulls a Google sign-in into the flow, which excludes anyone answering from a non-Google identity.
- The Responses tab degrades at scale. Question and Individual views stop rendering past 10,000 responses, and from September 2026 each form is capped at 1 GB and 500,000 responses.
A Google Forms survey fails for one of two reasons, and neither is the form builder. Either the questions were written inside the editor, so they drifted into whatever was easy to click, or a setting nobody checked forced a Google sign-in and cut the sample in half. Both happen before a single response arrives, and both are invisible once the link is out.
So this guide spends most of its time on the question set, the settings that gate who can answer, and what happens to answers once the spreadsheet fills.
What a customer survey has to do that a form doesn't
An intake form has a captive audience. Someone filling out a booking request wants the booking, so they tolerate eleven fields. A customer survey has no such hold. The person has already bought, already been served, and owes you nothing.
That asymmetry shows up in the data. In a roundup of response rates by survey length, Survicate's sample of 267,564 responses puts one-to-three-question surveys at 83.34% completion, four-to-eight at 65.15%, nine-to-fourteen at 56.28%, and fifteen or more at 41.94%. By time, under three minutes holds 85% to 95% and ten to fifteen minutes falls to 45% to 60%.
Those numbers only describe who finished. The quieter problem is who answered which questions. Pew's review of 92 open-ended items across more than 40 survey waves found closed questions draw 1% to 2% item nonresponse against a median near 18% on open ends, with the worst item skipped by 52% of respondents. Cognitive burden predicted it most strongly, and nonresponse ran higher on mobile.
So a customer survey is a budget, not a canvas. Every question spends completion, every open end spends it faster, and you should know what you are buying before you build.
Write the question set before you open the editor
Pick the one decision the survey has to inform
Write the decision down first, as a sentence: we need to know whether to keep the Thursday onboarding call or move it into the help centre. Then keep only the questions whose answers would change it.
This cuts most draft surveys roughly in half. Company size, role, and how did you hear about us survive because they feel like hygiene, not because anyone will slice results by them. No plan to filter on a field means it costs completion for nothing.
Lead with a closed question, keep open ends to one
Put an easy closed question first. A multiple-choice or scale item at the top gives the respondent momentum, and a partial answer still tells you something.
Then budget exactly one open end, at the bottom, phrased narrowly. What would have made the handover smoother pulls better answers than any thoughts on your experience, because it names the moment and asks for one thing. Pew also found that asking for multiple sentences draws more nonresponse than asking for a word or phrase. For wording, meetergo collects open-ended survey question examples and multiple choice survey questions.
Choose the scale before you choose the question type
Forms gives you two rating mechanics and they are not interchangeable. Linear scale starts at zero or one, ends on any whole number from 2 to 10, and labels both ends. Rating renders as stars, hearts, or thumbs up, and only accepts a range from 3 to 10.
That matters because the standard metrics have fixed shapes. Net Promoter Score needs 0 to 10, which only Linear scale produces; a five-point satisfaction question works in either. Lock the scale now, because changing it later makes two waves incomparable.
Multiple choice grid is where surveys lose people: questions on the rows, options on the columns, which reads fine on a laptop and badly on a phone. Past four rows, split it into separate scale questions.
Build it, then stop adding
Open Forms, start blank, and add the questions you already wrote. Worth knowing from Google's question-type reference: a form holds 300 items and 75 sections, Required is per question, and response validation on short answer and paragraph fields is a character-count rule.
Branching only works from multiple choice, checkboxes, and dropdown, so a triage question that routes unhappy respondents has to be one of those three.
Skip this if your survey is five questions long: sections only earn their keep once the scroll bar starts discouraging people. Under about eight questions, one page converts better than three.
Three settings that decide whether anyone can answer
All three live in the Settings tab, and each changes the population that can reach your survey. This is the part that gets skipped.
Collect email addresses
Under Responses, Collect email addresses has two modes. Verified captures the respondent's Google Account address and shows a confirmation notice on every page. Responder input asks them to type an address before submitting.
Verified is the one to be careful with. It requires a Google Account, so a customer on a work laptop signed into a non-Google identity provider hits a wall. Responder input keeps the form open to everyone but gives you a field people can mistype. If you need the address to close the loop, use Responder input and accept the noise. If you do not need it, switch the setting off and let the survey be an anonymous form, which usually buys franker answers anyway.
Limit to 1 response
Limit to 1 response is a sign-in requirement wearing a different hat. The only way Forms can tell one person from another is a Google Account, so switching this on forces authentication even when you are collecting nothing identifying.
For a satisfaction survey sent to a known list, the trade is rarely worth it: duplicate submissions are a small problem, a sign-in wall in front of a two-minute survey is a large one. Allow response editing is more useful than it looks on longer forms, since it lets someone come back and finish.
Publish and the Responders list
Forms gates distribution behind an explicit publish step. Publish makes the form live, and Manage controls who counts as a responder. Anyone on that list can answer; a restricted form turns away everyone else.
Two consequences people discover late. An unpublished form cannot accept responses at all, so a link copied before publishing goes nowhere. And Add expiration on a responder's access is separate from closing the survey, so an expiry set here cuts someone off mid-wave without changing the form's status.
Share the link people will open
Use Copy responder link, not the URL in your browser bar; the editor URL only works for people who can edit. Shorten URL trims it for SMS or a printed card.
For a survey inside a customer email, Pre-fill form is worth five minutes. You fill in what you already know, take the generated link, and the respondent lands on a form that is partly done. Fewer fields left, better finish rate.
Embed HTML drops the form into a page, which suits an always-on feedback widget more than a timed wave. One thing to avoid in any customer-facing survey: a file upload question. It forces a Google sign-in, and it is unavailable if the form lives in a shared drive or your admin has Data Loss Prevention switched on.
Read the answers without losing them
The Responses tab has three views: Summary charts each question, Question walks one question across all respondents, and Individual shows a single submission with Previous and Next controls. Summary only appears once a response exists.
Send the data somewhere durable early. Link to Sheets creates or attaches a spreadsheet, and More then Download responses (.csv) gives a flat export. Turn on Get email notifications for new responses so a complaint does not sit unread for a week.
Two gotchas. Anyone added as a collaborator on the form may also reach the linked spreadsheet, and revoking access means removing them from both. And the views thin out at volume: Question and Individual stop rendering past 10,000 responses, CSV exports stop arriving in timestamp order above the same mark, the Summary chart disappears past 50,000, and Sheets syncing breaks past 100,000.
The ceiling is also moving. From September 2026, Google's documented response limits cap each form at 1 GB of storage and 500,000 responses, counting form and response data but not files respondents upload. Far above a normal survey wave, but it matters for an embedded form you leave running for years.
Then close the wave properly. Under Published, turning off Accepting responses shows visitors a notice you can edit, and Set close date or response limit does the same on a schedule, either on a date or after a response count.
Where Google Forms stops being the right tool
For a one-off customer survey that ends in a chart and a conversation, Forms is the right answer and it costs nothing. Be honest about that before shopping.
It stops fitting when the answer has to trigger something. A detractor score that needs a call booked this week, an intake answer that needs a signed engagement letter, a complaint that needs an owner: Forms collects all three and acts on none of them. A human has to notice the row and start the next step by hand.
There is a data-protection edge too. GDPR's data minimisation principle asks for only what is adequate and relevant to your stated purpose, which argues for switching off email collection rather than for leaving Forms. Residency is the part that does become a tooling question: if your customer DPAs specify EU storage, that is a conversation with your DPO, not a settings change. meetergo's look at GDPR-compliant software covers how the assessment runs.
Tools that close the loop
meetergo is a scheduling and revenue-pipeline platform with a form builder attached, so the useful comparison is not survey-versus-survey. Its own polling tool finds a meeting time rather than measuring satisfaction. The forms layer is the overlap, and what it adds is the step after the answer.
25+ field types cover the survey basics plus rating, yes/no, signature, file and image upload, and a composite contact block. Conditional logic runs without code, supports AND and OR, and evaluates EQUALS, CONTAINS, BLANK and GREATER THAN, so it branches off more question types than the three Forms allows.
Routing is the difference that matters for feedback work. An answer can send the respondent straight into a booking with the right person, which turns a low score into a scheduled call rather than a row in a sheet. Forms can go directly to a named recipient by email or SMS with open and completion tracking, and responses land against a contact in the built-in CRM instead of a separate file. Data sits on EU servers, with no US corporate parent in the chain.
Honest limits: multi-step forms, funnels and smart routing start on the Suite plan at €29.90 per month (regular €34.90), above a Free tier at €0 with a 7-day trial on paid plans. And meetergo is not a research instrument. No panel, no weighting, no cross-wave trend reporting, so a longitudinal study still wants a dedicated online survey tool.
Try meetergo's form builder free. No credit card required.
Common mistakes
- Writing the survey in the editor. The question set should exist in a document first. Built in place, surveys accumulate questions nobody will analyse, and the completion data above prices that habit.
- Asking for a rating and a reason in one question. Split them. A scale question plus one narrow open end gets you numbers and texture; one field asking for both gets you a number and a blank.
- Shipping without a test submission. Submit the form yourself, then open the Responses tab and check what landed. Ninety seconds catches a forgotten email field, a branching loop, and a required question nobody can answer.
- Treating the spreadsheet as the finish line. Someone has to own reading it and someone has to own replying. A survey with no named owner produces a chart and no change, which is the main reason customers ignore the second one. meetergo's guide to running customer survey and feedback programmes covers the ownership side.
Booking + video conferencing in one tool.
Booking + video conferencing in one tool.
FAQs
Is a Google Forms survey free?
Yes, on personal and Workspace accounts, with no cap on respondents. The costs are indirect: the 2026 storage and response limits, and the manual work of acting on what arrives.
Can you make a Google Forms survey anonymous?
Yes, if you switch off Collect email addresses and Limit to 1 response and ask no identifying questions. Respondents may still see their own account email displayed while filling it in, which worries people even when nothing is recorded.
How many questions should a customer survey have?
Fewer than you want. The completion data points at one to eight questions for anything sent to customers, with a single open end last. If you need more than ten, the honest fix is two shorter surveys at different moments.
Do respondents need a Google account to answer?
Not by default. They do if you turn on Limit to 1 response, set email collection to Verified, or include a file upload question. Any one of those turns an open survey into a gated one.
How do you stop a survey after a set number of responses?
Open the published form, go to Set close date or response limit, and choose after a number of responses. The cap has to exceed the count you already have, and simultaneous submissions can slip above it.
Can you move Google Forms answers into another tool?
Through the linked spreadsheet or a CSV export, and both are one-directional snapshots. If the answer itself has to start a workflow, a form with automated follow-ups attached beats a spreadsheet plus a reminder to check it.
