There's no shortage of advice on how to collect customer feedback. Most of it is some version of "send a survey", with a couple of paragraphs about question wording and a chart of average response rates from someone's customer base. That's not what this post is.

This is a practical guide for small businesses. Cafes, hotels, salons, gyms, dental practices, spas, anywhere you serve people in person and want to know what they actually think. It covers the seven main methods that work, when each one earns its place, and the parts that matter more than the method itself (which are timing and question design, not which channel you pick).

If you've already got something running and it isn't producing useful answers, the problem is almost never the channel. Usually it's bad timing, a question that was never going to get a useful response, or a backend where the answers go to die. We'll get to all of that.

On this page

What good feedback collection looks like

Before the methods, the standard you're aiming for. Good feedback collection has a few traits that distinguish it from the kind that produces a folder full of unread responses.

It happens close to the experience. The further you drift from the moment something happened, the worse the answer gets. By the next day, the customer remembers a vague impression. By the next week, they remember whether they liked it. The detail you actually need is gone within hours.

It asks specific things. "How was your visit?" is the worst question in the world. It produces "great" or "fine" or nothing, and you've used up the customer's attention without learning anything. Specific questions get specific answers (we'll come back to this).

It has somewhere for the answers to land. A form that emails responses to an inbox nobody checks is worse than no form. The collection step is the cheap one. Reading and acting are the expensive ones, and most operators get the ratio backwards.

It treats silence as data. The people who don't fill in the form told you something with that decision. Either they didn't care enough, or they didn't see the prompt, or the experience wasn't memorable enough to bother. Treating low response rates as a failure to chase rather than a signal to read is one of the most common mistakes (there's a piece on that here).

It runs continuously, not in batches. Quarterly customer surveys are too slow for anything you'd actually want to fix. By the time you read about the broken espresso machine, three weeks of customers have already noticed.

With that as the bar, here's what's actually available.

The seven main collection methods

There are more than seven if you count edge cases, but these are the ones that show up reliably across the businesses I've seen. Each has a profile of what it's good for and where it falls down.

QR codes (in-person) Email surveys On-site web forms SMS / text Receipt prompts Phone follow-up Public reviews Your feedback inbox AI SUMMARY All sources, one view. Themes, sentiment, gaps.
Seven channels in. One place to read it.

In-person QR code

A QR code at the table, on the counter, on the receipt, or on a thank-you card you hand the customer. They scan it, the form opens on their phone, they fill it in. This is the dominant method for hospitality and retail right now, and for good reason. Customers are already on their phones, the friction is low, and you can put it physically next to the moment of experience.

The thing that makes or breaks it isn't the QR code itself. It's where you put it and what wording goes next to it. A QR sticker on a table that says "feedback" produces almost no responses. A small framed card that says "Two questions, one minute. We read every one." next to a friendly face on the way out produces a lot more. The mechanism is the same. The framing changes the response rate.

Limitation: only works in physical contexts where you can put a code in front of someone.

Email surveys

The post-experience email, sent some hours or days after the visit, with a link to a form. The default for online retail and increasingly the default for hotels and bookings.

Email's main advantage is reach. You can send it to everyone, you can time it precisely, you can include it in a transactional email someone's already going to open. Its disadvantage is timing. By the time the customer is reading the email, the experience is already in the past tense. They give you a summary judgement instead of a specific observation, and the answers come back shorter and less useful as a result.

Email surveys work best when there's a reason the customer would want to think back: an anniversary visit, a major purchase, a hotel stay where the experience was an event in itself. They're weakest for routine, repeat experiences (a coffee, a quick shop, a regular service) where the customer has already moved on by the time the email arrives.

On-site web forms

A feedback link in the website footer, a "leave feedback" button on the contact page, a popup on the order confirmation page. The form is there for anyone who decides to fill it in.

These are the lowest-effort collection method (set up once, runs forever) and the lowest-yield. The customers who go looking for the feedback button are usually the ones with strong opinions, in either direction. Quietly satisfied customers don't go hunting for forms. The people who do are over-represented at the ends, which means the data is biased toward complaints and superfans.

That doesn't make it useless. It makes it a different kind of signal. Treat on-site web forms as a way to capture the customers who feel strongly enough to seek you out, not as your main collection channel.

SMS

A text message after the experience with a short question or a link to a form. Common in healthcare, fitness, and any business where you already have phone numbers and a clear post-experience moment.

SMS has the highest open rate of any digital channel, and that's not nothing. The downside is that it's also the most intrusive. A text feels personal in a way email doesn't, which means a feedback request by SMS reads as more of an ask. If it's badly timed or frequent, it crosses into pushy fast. Send one message at the right moment with a short ignore-friendly link and SMS can outperform email by a wide margin. Send three reminders and you'll teach the customer to mute the number.

Receipt-based prompts

The line at the bottom of the receipt. "Tell us how we did at [URL]" or a small QR code printed alongside the total. Old-school, still works, especially for retail and hospitality with a clear receipt moment.

The advantage is that the customer is already looking at the receipt. They're checking the total, putting it in a wallet or a bag. The feedback link rides along on attention they were already giving. The disadvantage is that paper receipts are a fading channel. Email receipts and digital wallets are eating into the surface area, and a digital receipt with a feedback link tends to get the same response as an email survey (which is to say, not much).

Phone follow-ups

Someone from the business calls the customer after their visit. Used in high-touch services (luxury hotels, premium clinics, B2B), almost never used in volume businesses because of the labour cost.

A good phone follow-up produces the deepest feedback you can get. The customer is talking to a human, the conversation can branch, the questions can adjust based on the answer. The risk is that the customer feels obligated rather than invited, in which case they tell you what they think you want to hear, and the data is worth less than no data because you can't tell which is which. There's a post on that specific failure mode if you want it longer.

Phone is a method to use deliberately, on a small subset of customers, when the relationship justifies the labour.

Public review platforms

Google Maps, Yelp, Trustpilot, TripAdvisor, Booking.com. The customer leaves a review on a third-party platform with their name attached, intended to influence other people's decisions.

Public reviews aren't really a method you "collect" in the same sense as the others. You don't control the form. You don't control the timing. What you control is whether you ask for them, and whether you make it easy. The advantage is reach: you capture customers who would never have filled in your form. The disadvantage is psychology. People write public reviews when they feel a duty to warn or to recommend, which means the middle gets clipped. Five-star reviews tend to be about loyalty. One-star reviews tend to be about grievance. The customers who had a normal pleasant experience mostly don't write anything at all.

Treat public reviews as a parallel source rather than a substitute for direct feedback. You want both. Direct feedback is where you get the timely, specific, middle-of-the-distribution detail. Public reviews are where you reach the customers you'd never have heard from otherwise.

Comparison table

Method Best for Response rate Limitations
In-person QR code Cafes, restaurants, retail, salons, anywhere with a physical visit Medium to high (depends heavily on placement and framing) Only works in physical contexts
Email surveys Hotels, online retail, services with delivery moments Low to medium Arrives after the moment; answers tend toward summary judgements
On-site web forms Capturing customers who actively seek you out Low Biased toward strong opinions at both ends
SMS Healthcare, fitness, services with phone numbers and a clear post-visit moment Medium to high Intrusive if mistimed or repeated
Receipt-based prompts Retail, fast-service hospitality with paper or digital receipts Low to medium Paper receipts fading; digital receipt links underperform
Phone follow-ups Premium services, B2B, high-value relationships High (when used selectively) Labour-intensive; risk of polite inflated answers
Public review platforms Reach, capturing customers who wouldn't fill in a form Low (across customer base, but visible) Skewed to extremes; you don't control the questions

Response rates here are deliberately qualitative. The actual numbers vary so much by industry, timing, and form design that any specific percentage I gave you would be misleading. Use this as a relative guide, not a benchmark.

How to choose the right method

There's no universal answer, but there are useful starting points based on the shape of your business.

If your customers come to you in a physical space and there's a clear endpoint to the visit, start with in-person QR codes. Cafes, restaurants, salons, dental practices, spas, gyms, retail stores. Place the code at the moment the experience ends and the customer is most likely to have a few seconds free.

If your business has a delivery or completion moment but no physical handoff, start with email surveys. Online retail, services with a project completion, anything with an obvious "we just finished something for you" trigger.

If you have phone numbers and a clear post-visit timing window, SMS is worth testing alongside whatever else you're doing. It outperforms email on open rate, underperforms on length of response.

If you serve high-value customers in low volume, phone follow-ups make sense for a subset. Don't try to scale them. Pick the customers whose feedback would change something, and call them.

Public reviews are a parallel system in all cases. You should be paying attention to them whether or not you're running a feedback form. If you're using a tool to collect direct feedback, picking one that also pulls in public reviews from the platforms your customers use saves you the cross-channel reading. (This is the bit Qria handles, by syncing reviews from Google, Yelp, Trustpilot, TripAdvisor, and Booking.com alongside the direct responses, so the same AI summary covers both. The relevant feature is on the reviews page.)

The mistake is trying to run all seven methods at once. You don't have the attention to read seven channels. Pick one or two primary methods, do them well, treat public reviews as a parallel source, and ignore the rest until something forces a change.

Why timing beats method

Most arguments about how to collect customer feedback focus on the channel. QR codes versus email surveys, paper versus digital, popup versus post-purchase. The channel matters, but it matters less than timing.

The channel matters, but it matters less than timing.

The asymmetry is this. A bad question asked at the right moment will still produce decent feedback. A good question asked at the wrong moment produces shallow feedback or none at all. The channel is one of several levers on timing, but the timing is the actual variable that moves response quality.

The right moment depends on the experience. For a meal, it's after the bill but before they leave the table. For a hotel stay, the morning of checkout. For a service appointment, within minutes of the appointment ending. For a class, before the customer has moved on to whatever's next in their day.

The wrong moment is anything that introduces a gap big enough for the customer to compress the experience into a vague impression. By the next day, the dinner is "fine". By the next week, the hotel is "lovely". The detail that would have told you what to fix or protect is already gone.

This is why in-person QR codes outperform email surveys for most physical businesses. The mechanism isn't really about the channel. It's about the seconds. A QR code on the table catches the customer while the experience is fresh enough that they can remember whether the bread was stale or the server was good. The same customer, three days later, remembers "it was nice".

There's a longer post on timing specifically that gets into the moment-by-moment trade-offs. The short version: pick the method that lets you ask closest to the experience, and worry about everything else after that.

Question design that gets useful answers

The other thing you control once timing is sorted out is what you actually ask. Most feedback questions are bad in predictable ways, and fixing them is one of the highest-leverage things you can do.

Ask about specific, observable things

"How was the food?" is general. "Was your food at the right temperature?" is specific. The first one gets you "great" or "fine". The second one gets you a yes or a no, which is countable, plus an open-text follow-up where the customer might tell you the soup was lukewarm.

The pattern: pick specific, observable elements of the experience and ask about them. Not "how was your stay" but "how was the bed". Not "how was your visit" but "how was the wait at lunch". Specific questions force specific answers because the question itself contains the prompt the customer needs to recall the detail.

Pair structured with open-text

A rating gives you something countable. Open text gives you the context that makes the count mean something. A 4 out of 5 with no comment is data without information. A 4 out of 5 with "the coffee was great but the music was too loud" is the thing you actually wanted.

Most forms either skip the open-text follow-up or make it the whole form. Both are mistakes. The combination (a small number of structured questions, each followed by an optional comment) produces the best ratio of completion rate to depth of answer.

Avoid the leading question

"How great was your experience?" is leading. So is "what could we have done better?" (because it presupposes you did something less than perfect). Customers tend to give the answer they think you want, and the way you phrase the question telegraphs that wanted answer. There's a piece on the survey-pleaser problem that goes through how to write questions that resist it. Short version: ask neutral questions about specific things, and pair them with open text where the customer can volunteer something you didn't ask about.

Don't ask things you can't act on

If a question's answer wouldn't change anything you do, drop it. The classic offender is demographic questions on a feedback form for a coffee shop. You're not segmenting your customer base by age bracket. You're asking because the survey template had the field. Cut it.

The same logic applies to long lists of feature requests with no realistic chance of building any of them. If you don't have the bandwidth to act on the answers, you're collecting a wishlist that breeds frustration, both for the customer (who feels ignored) and for you (who feels guilty).

Length

Three good questions get more useful answers than ten generic ones. The standard advice is "keep it short", and that's directionally true. The thing it misses is that brevity isn't really the goal. Fitness for purpose is. A two-question form that asks the wrong things is worse than a five-question form that asks the right ones.

The form Qria builds for this is described on the forms feature page if you want to see the structure. Star ratings, open text, multiple choice, with the option to route positive responders to leave a public review on Google (review routing is the relevant bit).

Driving response rates without nagging

The headline number people obsess over is response rate. There's a real lever to pull on, and it's not the one most people pull.

The thing that doesn't work: reminders. Sending a second email three days after the first one because the customer didn't respond gets a small bump in absolute volume and a large drop in goodwill. The customer who was going to respond responded. The customer who didn't has now been asked twice for something they already chose not to do, which makes them less likely to engage with you next time.

The thing that does work: better timing and better placement. A QR code on the table asked once, with a small printed line explaining why, will outperform a multi-touch email sequence in both volume and depth of response. The same applies to SMS sent at the right moment versus email sent two days later.

The other thing that helps is making the form genuinely fast. The customer who clicks through to a form with twenty fields bounces. The customer who clicks through to a form with three fields and a comment box finishes. Length and structure are response-rate levers in their own right.

What you should not do is bribe people. Discount codes for completing the survey produce volume at the cost of signal quality. The respondent is no longer telling you about the experience, they're filling in a form to get the discount, and the answers reflect that motivation. If you want to thank respondents, do it after the fact, not as a precondition.

Finally, the loop. Customers who give feedback and never hear anything back stop giving feedback. Even a short auto-thank-you message changes the dynamic. A reply to someone who left a substantive comment changes it further. You don't need to write personal essays to every respondent. What you do need is a system where the time between feedback and acknowledgement isn't infinite. The first time someone leaves you a long, thoughtful response and you say nothing back, you've taught them not to bother next time.

  1. Better timing and better placement

    A QR code on the table asked once, with a small printed line explaining why, will outperform a multi-touch email sequence in both volume and depth of response.

  2. Make the form genuinely fast

    The customer who clicks through to a form with twenty fields bounces. The customer who clicks through to a form with three fields and a comment box finishes.

  3. Don't bribe people

    Discount codes for completing the survey produce volume at the cost of signal quality. If you want to thank respondents, do it after the fact, not as a precondition.

  4. Close the loop

    Customers who give feedback and never hear anything back stop giving feedback. Even a short auto-thank-you message changes the dynamic.

Industry-specific guidance

The general principles cover most of what you need. The specifics are where it gets useful, because what works for a hotel doesn't always transfer to a cafe and vice versa. Here are the dedicated pieces for each of the businesses we see most often:

Each one digs into the specific timing windows, channel choices, and question phrasing that work in that context. If your business doesn't fit one of these neatly, the closest match is usually a fine starting point.

Frequently asked questions

What is the best way to collect customer feedback?

There isn't a universal best method. The right answer depends on the shape of the relationship. In-person QR codes work well for physical retail and hospitality. SMS or email after the appointment fits scheduled services. Email surveys after delivery suit online transactions. Selective phone follow-ups make sense for high-value relationships. The best method, whichever channel it is, is the one that lets you ask closest to the moment of experience.

How do I collect customer feedback for free?

A QR code linked to a Google Form is free and works. So is a feedback@yourbusiness.com email address with a printed card that invites comments. The limitations show up later: no analysis, no public review aggregation, no good way to handle multiple locations, manual reading. For a single-location business getting fewer than ten responses a week, free is genuinely fine. Past that, the time you spend reading and organising starts to outweigh the cost of a dedicated tool.

How often should I ask customers for feedback?

Once per visit is the cap for most businesses. Two requests for the same experience is one too many, and it crosses the line into nagging. Across visits, frequency depends on the cadence of the relationship. A monthly customer can be asked monthly. A weekly customer should be asked occasionally, not every visit. The general principle is that any single customer should feel asked, not pestered.

What's a good response rate for customer feedback?

It varies more than the surveys-industry numbers suggest. A 5% response rate on a well-placed in-person QR code is fine if the responses are specific and timely. A 30% response rate on a long email survey can be worse, because the volume is going to people who reflexively complete forms rather than people with something to tell you. The number to watch is depth and signal, not headline rate. (There's a post on low response rates as a signal that goes deeper.)

Should I use multiple methods at once?

Two is usually the right number. One primary method tied to the moment of experience (QR code, SMS, email at the right time), plus public reviews running in parallel. Three or more channels means you're spreading your reading attention thin, and the data starts to fragment across systems. If you're considering a third method, it's usually a sign that one of the first two isn't working and the answer is to fix it, not to add another.

How do I make customers actually want to leave feedback?

Keep the form short, ask close to the experience, and acknowledge the response when it lands. The biggest single lever is timing. Question design is a close second. Acknowledgement is the bit most operators forget about, even though it's the cheapest part of the whole loop. A customer who gives you something and gets a brief thank-you back is more likely to do it again next time. The customer who gets silence usually doesn't.