At the end of a job you have one clean opening to send your customer something. A survey wants it. A review request wants it. Both arrive as a text or an email, both ask for a minute of goodwill, and sending both turns a small favour into a decision.
So the useful question is not what to put in a customer satisfaction survey. It is whether this job needs a survey at all, and if it does, where the survey goes so that it is not standing in the review request's spot.
What ranks for this term was written for a CX team
Search the phrase and the results are question banks: fifty questions, a hundred questions, templates sorted by journey stage, scales to pick between, dashboards to pipe the answers into. That material is written for a company with a customer experience function, a research budget and a product roadmap that feedback is supposed to steer.
A four-van plumbing outfit, a dental practice or a salon has a different problem. There is no roadmap for the answers to steer. There is a service that either went well or did not, and an owner who already has a rough idea which.
A survey is a diagnostic, not a routine
Here is the distinction worth keeping. A survey is the right instrument when you suspect something and cannot see it.
You have a feeling that jobs booked through the Saturday phone line go worse than the ones booked online. Two technicians are getting repeat callouts and you do not know whether it is the work or the explaining. Your Tuesday lunch cover has turnover and you cannot tell if customers notice. These are answerable questions, and asking customers directly is a reasonable way to answer them.
What a survey will not do is produce anything a future customer can read. The answers land in your inbox, in private, written by someone who has already bought from you. Useful to you, invisible to the person deciding whether to call you instead of the next business in the list.
That is the division of labour. The survey is for you. The review is for the stranger.
So run it as a campaign with a start and an end
Because it is a diagnostic, a survey has a shape: you have a question, you ask it, you get enough answers to act, you stop. Send it for a fortnight to the jobs that bear on your question, not to everybody forever.
Running it permanently is where it goes wrong. A standing survey quietly becomes the thing you send at the end of every job, which means it has taken the slot the review request needed, and it is returning private answers about customers who were already yours.
The questionnaire: one question you would act on
Since the survey exists to settle one question, the questionnaire should be close to that short.
- Open with a rating question, because it is easy to answer and it sorts the replies for you. A simple scale is fine. What matters is that every respondent gets the identical question.
- Add one open box with a specific prompt. Not "any other comments", which collects nothing. Ask about the part you are actually investigating: how the arrival time worked out, whether the quote matched the invoice, whether anything was left unclear.
- Stop. Every extra question is a reason to abandon the form, and you will not act on the answer anyway.
- Ask it where the job happened. A paper slip at a clinic reception desk, a link in the follow-up email for a service call.
If you catch yourself adding a question because the data might be interesting later, cut it. Interesting is not actionable.
Do not wire the answers into who gets a review link
This is the part the template banks never mention, and it is the one with a real penalty attached.
Once you are collecting satisfaction scores it feels obvious to use them: send the Google review link to the people who rated you well, and handle the unhappy ones privately. That is review gating, and Google's prohibited merchant behaviour lists exactly it, "Discourage or prohibit negative reviews, or selectively solicit positive reviews from customers." The same policy treats incentives, "payment, discounts, free goods and/or services," as fake and misleading content.
So the survey result cannot be an input to the review request. Nothing branches on it. Every customer gets the same ask, in the same words, whatever they told you privately, which is what makes the resulting rating worth anything to the person reading it. Our post on what Google's review policy forbids goes through the rest of the list.
Reading an unhappy survey reply and then fixing the thing it describes is not gating. That is the entire point of having asked. What you may not do is let the reply decide who hears from you next.
A split that works
Default to the review request. It is the ask that attaches to every finished job, in the same words each time, pointed at your Google link, and it compounds: each one keeps working on strangers long after the job is forgotten. Google's own guidance on getting reviews describes the method in a line, "To leave reviews, you can ask customers to visit a Google link or scan a QR code." Wording for that message is in our review request templates, and the best time to ask for a review covers attaching it to a trigger instead of a mood.
Then, when you have a question you cannot answer from the office, run a survey as a short separate campaign, aimed at the jobs that bear on it, through a different channel or a different week so the two asks never land together. Read the replies, change the thing, stop sending it.
One instrument tells you what to fix. The other tells everyone else you were worth calling. Neither does the other's job, and the ask you have room for is the second one.