Reviews give you the extremes. Surveys are how you reach everyone else — the large, quiet middle who thought the meal was fine, noticed one thing that was not, and would never dream of writing it publicly. The catch is that almost every restaurant survey ever made is too long, asked at the wrong moment, and phrased so that the answer is already in the question. This topic is about building the short, honest one that people finish.
✂️ Shorter than you think, then shorter again
Every question you add costs you answers. A survey with one question gets answered in the parking lot; the same survey with twelve gets abandoned halfway through by everyone except the two people with a strong opinion — which is exactly the bias you built the survey to escape. Length does not just reduce your sample, it skews it.
So start from the opposite end. Instead of asking what you would like to know, ask what you would change on Monday, and let that pick the questions. Almost always the honest answer is one or two things, which fits in three questions:
- One scoring question that you can track over time. "How likely are you to recommend us?" or simply "How was it, out of five?" — the exact wording matters less than never changing it, because the value is in the trend.
- One open question with a box to type in. "What is the one thing we could do better?" This is where the real information lives; the score tells you the temperature, the sentence tells you why.
- One optional question that changes with what you are working on right now — the new menu, the wait on Fridays, the new patio.
That is the whole thing. If someone insists on adding demographics, table number and how they heard about you, remember that each of those is paid for in abandoned surveys. Anything you can already get from your POS or booking system should never be asked of a customer.
⏱️ Timing beats wording
When you ask changes the answer more than how you ask. Catch someone while the meal is still in their mouth and you get politeness; catch them three days later and you get a vague memory. There is a short window — leaving, or within a few hours after — where the experience is fresh and the social pressure of your staff standing there is gone.
| Moment | What you get |
| At the table, mid-meal, staff waiting | "Everything's great, thanks" — politeness, not information |
| On the check, to answer later | Honest, and the highest-volume channel most restaurants have |
| A few hours after a delivery or takeaway order | Very honest; they are home, alone, and just ate |
| The next day by email | Fewer answers, but more considered and better written |
| A week later | Blurred memory, low response, mostly the extremes again |
Distribution follows the same logic: whatever is already in the customer's hand wins. A QR on the check presenter or receipt costs nothing and reaches every single guest. A link in the confirmation of an online order reaches every online order. An email the day after works if you have addresses — and if you took the email course, you already know not to burn your list on constant surveys; once after a first visit, and occasionally, is plenty.
🪞 Questions that do not answer themselves
A badly phrased question does not just fail to inform you — it actively misleads you, and it does it in the flattering direction, which is why so many restaurants have survey results that look wonderful and a dining room that keeps getting quieter. Most of the damage comes from a handful of habits.
| Instead of | Ask |
| "How much did you enjoy our amazing new menu?" | "What did you think of the new menu?" |
| "Was our service fast and friendly?" | "How was the service?" — one thing at a time |
| "Any problems?" (nearly always gets "no") | "What is the one thing we could do better?" |
| "Rate the food, atmosphere, service, value, parking…" | One score, one open box |
| "Would you like to join our mailing list?" as question 1 | Ask that at the end, after they have answered |
Two more traps worth naming. Never ask something whose answer you cannot act on — asking about parking you cannot change produces complaints you have to read and cannot fix. And be careful about anonymity: people are markedly more honest when they are not identified, so unless you specifically need to follow up, let them answer without giving a name. If you do want the option, make the contact field optional and say why you are asking.
📉 What the numbers do and do not mean
Once answers start arriving, keep two limits in mind so you do not overreact. First, only a small share of guests ever answer — often a few in a hundred — and the people who bother are not perfectly representative. Second, a score on its own is nearly meaningless in isolation; a 4.2 is neither good nor bad until you know it was 4.5 last quarter.
So read them this way. Watch the trend rather than the level, and pay attention when it moves after something changed — a new supplier, a new shift pattern, a price rise. Treat the score as the smoke alarm and the written answers as the fire: the number tells you something happened, the sentences tell you what. And do not compare your average with the restaurant down the street, because their question, timing and audience are different from yours; the only fair comparison is with your own last quarter.
The most valuable habit is the least technical one: read the open answers yourself, all of them, every week. It takes ten minutes at this volume, and it is where you will spot the sentence that explains why Fridays have felt strange since March. Topic 6 is about what to do when there are too many of those sentences to hold in your head.