Most customer satisfaction surveys fail before anyone reads a word of them. They’re too long, they arrive at the wrong moment, and they ask questions whose answers can’t change anything. You get a 4.2 average, no idea what to do about it, and a quiet suspicion that the people most annoyed with you didn’t reply.
The fix is mostly subtraction. A survey that asks five good questions two days after a delivery will teach you more than a twenty-question form sent at the end of the month, and it will teach you the one thing you actually want to know: is this customer going to order again?
This guide is the question bank. Which customer satisfaction survey questions earn their place, how many to send, when to send them, which ones predict a reorder, and a five-question form you can copy as it stands. It’s the private half of your reputation, the half you read before anyone else does, and the first step in the sequence laid out in public relations for small business owners who deliver.
The Bottom Line
- Keep it to five questions or fewer. A 10-question survey averages an 89% completion rate; a 40-question survey drops to 79%, and length is the most consistent predictor of abandonment in SurveyMonkey’s platform data (SurveyMonkey, retrieved 2026-11-08).
- Send it 24 to 48 hours after the delivery arrives, not after the invoice clears. You’re asking about an event, and people forget the details of an event fast.
- Ask one rating question, one reorder-intent question, and one open-ended question. Everything else is optional.
- 48% of respondents will only spend one to five minutes on a survey (SurveyMonkey, 2026 benchmark guide, based on 2025 platform data). Budget your questions against that, not against what you’d like to know.
- The single most useful question is not about satisfaction. It’s “how likely are you to order again” because satisfaction is an opinion and reordering is a decision.
Lower your delivery costs by 23%
How we reduce costs:
- No delivery vehicle expenses
- Optimized local routes
- Pay-per-delivery model
- Average 23% delivery cost reduction
How many customer satisfaction survey questions should you ask?
Five or fewer for a post-delivery survey. Ten is the absolute ceiling, and you should need a reason.
The data here is unambiguous and it all points the same direction. SurveyMonkey’s completion-time research puts a 10-question survey at roughly 89% completion and a 40-question survey at 79%, with abandonment rising sharply past the seven-to-eight-minute mark (SurveyMonkey, retrieved 2026-11-08). Their 2026 benchmark guide adds the constraint that matters most: 48% of respondents will only spend one to five minutes, and the guide’s own recommendation is to aim for under ten questions, or five for a quick pulse check (SurveyMonkey, 2026).
There’s a second reason to keep it short that the completion statistics miss. Long surveys don’t just lose people. They lose a specific kind of person. The customer with a mild complaint abandons at question eight. The delighted one and the furious one finish. So a long survey filters out exactly the moderate, fixable feedback you most need, and hands you a bimodal picture of your own business.
Short surveys sent often beat long surveys sent rarely. Three questions after every delivery for a month is better data than twenty questions once a quarter.
When to send a customer satisfaction survey after a delivery
Send it 24 to 48 hours after arrival. Two competing timings exist and the right one depends on what you’re asking about.
- Asking about the delivery itself (timing, packaging, condition, the handoff): send within a day or two. These are perishable memories. Ask on day ten and you’ll get “fine, I think.”
- Asking about the product (did the bread keep, were the flowers still good on Thursday): wait five to seven days, because the customer needs to have used the thing.
For most businesses that deliver, the first matters more, so default to 24–48 hours. If you need both, send two separate two-question surveys rather than one combined six-question one.
Two timing mistakes are worth naming. Don’t attach the survey to the invoice; you’ll measure feelings about being billed. And don’t batch them into a monthly send. A survey that arrives detached from a specific delivery gets answered from general impression, which is the least useful data there is.
The three rating questions that do the real work
Nearly every satisfaction survey framework reduces to three metrics. You don’t need all three in one survey, but you should know what each one measures and what it can’t tell you.
| Question type | The question | What it measures | Best used for |
|---|---|---|---|
| CSAT | “How satisfied were you with this order?” (1–5) | Feeling about one specific transaction | Spotting a bad delivery immediately |
| NPS | “How likely are you to recommend us to someone else?” (0–10) | Willingness to stake their own reputation on you | Tracking a trend over quarters |
| CES | “How easy was it to place and receive this order?” (1–5) | Friction in your process | Finding operational problems |
CSAT is the one to use if you only use one. It’s immediate, it’s specific to the order, and a 1 or 2 is an alarm you can act on the same day. How the percentage is calculated and what counts as a good result is worth understanding before you start tracking it, and what a customer satisfaction score is and how to measure it covers that side.
NPS gets criticized for being a single number pretending to be a strategy, which is fair. But the question itself is excellent, because recommending you costs the customer something. A 9 means more than a five-star rating does. Just don’t expect the aggregate score to tell you anything on a monthly basis at small volumes; twelve responses can’t produce a trend.
CES is the underrated one for delivery businesses. “How easy was it” surfaces problems customers never complain about: a confusing order form, a delivery window too wide to plan around, an invoice they had to chase. These things don’t make people angry enough to call. They just make them not reorder.
Post-delivery survey questions about the arrival itself
This is where a generic satisfaction survey wastes its questions and a delivery business should spend them. The order went into a vehicle, travelled, and was handed to someone. Five things can go wrong, and each has a question:
- Did your order arrive within the window we gave you? Yes / No / It was early. Note the third option. Early is a failure for a restaurant receiving produce before anyone’s there to sign.
- Was everything you ordered in the box? Yes / No, something was missing. Short-shipping is the most common complaint in wholesale and the one customers are least likely to volunteer.
- What condition did it arrive in? Perfect / Minor damage / Unusable. Scale this to what you ship. For a florist, “how did the stems look on arrival” is the question.
- How was the handover? This covers the part of the delivery you never see: whether the driver found the right door, waited, called ahead, left it somewhere sensible.
- How satisfied were you with the packaging? Only if you ship something fragile or temperature-sensitive, in which case it’s one of your most important questions.
You won’t ask all five every time. Rotate them. Run the window question for a month, then the completeness question, and you’ll build a picture of your operation without ever sending a long survey.
One note on the handover question: it’s the one that produces feedback about people, and it should be read carefully rather than averaged. Customers rarely rate a driver poorly, but they will mention in passing that nobody rang the bell, and that detail is more useful than a score.
It’s also the feedback that should change who you bring on next. The qualities customers keep praising in the handover (calling ahead, patience at a tricky loading dock, knowing which door to use) are the ones worth screening for, and the driver interview questions worth asking turn that survey feedback into a hiring filter.
Open-ended survey questions worth the space
One open-ended question per survey. Not two, and never three. Qualtrics notes that completion rates start falling once a survey carries more than three open-text questions, and the typing is the reason.
The open question you ask should not be “do you have any other comments?” That produces blanks and “no.” Ask one of these instead:
- “What’s one thing we could have done better on this order?” The constraint to one thing is what makes it work. It gives permission to complain mildly, which is the feedback you’re otherwise missing entirely.
- “What almost stopped you ordering from us?” Uncomfortable, and the single most informative question in this article. The answers are about your website, your minimums, your delivery radius, your prices.
- “If you were recommending us to someone, what would you tell them?” This one does double duty. It produces usable feedback and it produces the exact wording customers use about you, which, if they agree, becomes a testimonial you didn’t have to write.
That last question is the bridge between a survey and your public reputation. A customer who has just written a sentence explaining why they’d recommend you is the single best person to ask for a public review, and they are never more willing than in that moment. The mechanics of the ask (the timing, the wording, the link) are covered in how to get more Google reviews without asking twice.
Which answers actually predict a repeat order
Satisfaction and loyalty are not the same measurement, and this is where most survey programs fail: they track a number that goes up while customers stop reordering.
A satisfied customer is reporting a feeling about something that already happened. A loyal one is making a prediction about their own future behaviour. Only the second correlates with revenue, and you have to ask for it directly:
“How likely are you to order from us again?” (1–5, or Definitely / Probably / Not sure / Probably not)
This is the question to add if you add only one. And the interesting signal is the gap between it and your satisfaction score. Three patterns, three different problems:
- High satisfaction, high reorder intent. Working. Ask these customers for reviews and referrals, because they’re your reputation engine.
- High satisfaction, low reorder intent. The most common and most misread pattern. The order went fine and they still aren’t coming back, which means the problem isn’t quality. It is a need you only met once, a price that doesn’t work for a standing order, or a competitor who’s closer. No amount of improving the delivery fixes this.
- Moderate satisfaction, high reorder intent. The most valuable customers in your file. They’ll keep ordering while being faintly unimpressed, which means small fixes compound and they’ll tell you what the fixes are if you ask the open question.
Two more questions sharpen this considerably. “Is this your first order with us?” lets you split new from returning, which you need because first-order satisfaction and tenth-order satisfaction mean completely different things. And “What would make you order more often?” gets you answers about frequency rather than quality: usually a delivery day you don’t run, a minimum that’s too high, or a product you don’t carry.
The pattern worth watching over time is what happens at order three. Customers who reorder twice and then rate reorder intent highly tend to keep going. That’s the point at which someone has stopped trying you and started relying on you, and the number of customers reaching it is a better health measure for a delivery business than any satisfaction average.
A five-question delivery feedback form you can copy
Send this 24–48 hours after arrival. It takes about 45 seconds to complete.
- How satisfied were you with this order? (1–5)
- Did it arrive within the window we gave you? (Yes / No, it was late / It was early)
- Was everything you ordered in the box, and in good condition? (Yes / Something was missing / Something was damaged)
- How likely are you to order from us again? (1–5)
- What’s one thing we could have done better? (open text, optional)
That’s it. Questions 1 and 4 give you your two tracking numbers. Questions 2 and 3 tell you whether a bad score was an operational failure or something else. Question 5 tells you what to fix.
Make question 5 optional and put it last, so an abandoned survey still leaves you four usable answers.
You don’t need to build the form from nothing either. A pre-built customer satisfaction survey template gets you a working form in a few minutes, and the only edits worth making are swapping in the arrival questions above, since generic templates are written for software and retail rather than for something that had to be driven to a door.
What to do with the answers
A survey nobody acts on is worse than no survey, because you’ve asked a customer to spend time and then demonstrated it didn’t matter.
Three habits make the data worth collecting:
- Reply to every score of 1 or 2 within a day. Personally, from your own email. Recovery from a bad delivery handled well produces more loyalty than a delivery that simply went fine, because you’ve shown them what happens when something breaks, which is the thing they couldn’t have known.
- Read the open answers monthly, in one sitting. Don’t average them, don’t tag them. Read fifteen in a row and the repeated phrase will be obvious. That phrase is your next operational fix.
- Watch for the complaints that are really routing problems. Late arrivals, wide windows, drivers who can’t find the loading entrance, orders arriving before anyone’s on site. These cluster by route and time of day, not by customer. Once you see the pattern you’re solving a dispatch problem, and consolidating stops or keeping the same drivers on familiar routes addresses more of your survey feedback than anything you can change about the product.
Where that fix lives depends on your volume. Below a certain number of weekly stops, re-sequencing the route yourself is enough. Past it, the arithmetic favours handing the driving to a local delivery network instead of adding a second van. In a dense market like courier services in Miami, the same survey complaint about a 40-minute window usually disappears the moment the stops are optimized across more drivers rather than squeezed into one.
And keep the numbers. A year of monthly satisfaction and reorder-intent scores, alongside your on-time percentage, is the only credible evidence of reliability a small business ever gets to produce. It’s what makes your reputation claims defensible, and it’s the raw material a reporter needs if you ever want local press coverage for something more interesting than a new product.
Frequently asked questions
How many questions should a customer satisfaction survey have?
Five or fewer for a post-delivery survey, and under ten in any case. SurveyMonkey’s data shows completion falling as length rises and abandonment climbing sharply past seven to eight minutes, with 48% of respondents willing to spend only one to five minutes (SurveyMonkey, 2026).
What is the best single customer satisfaction question?
“How likely are you to order from us again?” Satisfaction is an opinion; reorder intent is a forecast of behaviour, and it’s the one that tracks revenue.
What is a good response rate for a post-delivery survey?
For a short survey sent to people who have an existing relationship with you, anything above 15–20% is healthy, and small businesses with personal relationships often do considerably better. Response rates have fallen 10–15 points across the industry over the past decade, so compare against your own trend rather than a published benchmark.
Should I offer an incentive for completing a survey?
A small one is fine and lifts response rates. A discount on the next order works, and it has the useful side effect of encouraging a next order. Don’t tie the incentive to a positive rating, which buys you useless data.
Can I use the same survey for wholesale and retail customers?
The rating questions transfer; the arrival questions don’t. Wholesale buyers care about order completeness, timing precision and invoice accuracy. Retail customers care about condition and the handover. Run two short versions rather than one that serves neither.
Ask less, act faster
The survey that works is almost embarrassingly small: a rating, a reorder-intent question, two factual questions about the arrival, and one open prompt. Sent two days after delivery. Read within the week.
What makes it valuable is the loop. Ask, read, change one thing, ask again. A business running that loop for a year knows precisely why its customers reorder and can prove its reliability with numbers, which is worth considerably more than a dashboard showing a satisfaction score nobody has acted on since March.
