Most teams know they should survey customers. Fewer can say what a good program looks like at scale. Three well-known examples show the choices involved: where to ask, who to ask, and what to do with the answers.
For each, we pull out one practice you can borrow. It sits within our wider guide to customer surveys that produce decisions.
A note on sources: what follows reflects publicly known descriptions of each program. The companies are examples only, and nothing here implies any affiliation with or endorsement by them.
Key takeaways
- Strong customer surveying follows a loop: invite people at a natural moment, ask about things the business can change, and analyze the results into decisions.
- Taco Bell shows that response rate is mostly a design problem: every extra step between the customer and the first question costs you completes.
- Ford shows the value of scope: ask separately about the product, the people who sell it and the next product, and route each answer to an owner.
- J.D. Power shows that a score needs a reference point: use identical wording each time and compare against peers or your own past results.
- In the illustrative example, 60 completes per location means a 70% satisfied score could really be anywhere from roughly 58% to 82%.
What do these three examples have in common?
They sit in very different industries, yet each treats a survey as the start of a loop rather than a one-off questionnaire. The loop has three parts: invite people at a natural moment, ask about the things the business can actually change, and analyze the results to reach conclusions. The examples differ mainly in where they ask and who they compare against.
| Company | Where it asks | Practice to borrow |
|---|---|---|
| Taco Bell | At the point of sale | Make the invitation impossible to miss |
| Ford | Across cars, dealers and future plans | Ask about the whole relationship, then act fast |
| J.D. Power | Across companies in an industry | Benchmark against peers, not just yourself |
How does Taco Bell survey customers at the point of sale?
Taco Bell is known for inviting customers to take a short survey about their dining experience, covering factors such as food quality and service. The program has been described as having a few distinctive features: survey details printed prominently on receipts, employees encouraged to ask customers to take part (sometimes with incentive programs), an online option, and a sweepstakes entry for completing it.
The transferable practice
The invitation arrives while the experience is fresh, and it is easy to find. Response rate is mostly a design problem: every extra step between the customer and the first question costs you completes.
How to apply it
- Put the invitation where the customer already looks, such as the receipt, the confirmation email or the delivery box.
- Keep the survey short enough to finish in a couple of minutes, and say so in the invitation.
- Offer a small, clearly described reward if it suits your customers, and watch for answers that look rushed.
- Ask customer-facing staff to mention the survey.
For a deeper look at what moves participation, see our guide to survey response rate.
A worked example: why volume matters
Suppose a chain has 20 locations and 60,000 receipts a month, each carrying a survey invitation. The numbers below are illustrative, not from any real program.
| Scenario | Completed surveys | Completes per location | Margin of error on a 70% satisfied result |
|---|---|---|---|
| Receipt invitation only (2.0% response) | 1,200 | 60 | about 11.6 points |
| Receipt plus staff request (3.0% response) | 1,800 | 90 | about 9.5 points |
The arithmetic: 1,200 divided by 20 is 60 completes per location, and 1,800 divided by 20 is 90. The margin of error for a proportion is 1.96 times the square root of p times (1 minus p), divided by n. For 60 completes that is 1.96 times the square root of (0.7 x 0.3 / 60), or about 11.6 points. For 90 it is about 9.5 points.
So a location scoring 70% satisfied could really be anywhere from roughly 58% to 82% with 60 completes. Boosting responses narrows that range, which is what lets you compare locations with some confidence. The formula assumes a random sample, which self-selected respondents are not, so treat these figures as a minimum level of uncertainty.
How does Ford survey across cars, dealers and future plans?
Ford is known for running a customer surveying program that brings together several things at once: customers’ experience with their cars, their experience dealing with the company and its dealers, and what they would like to see in future vehicles. That is a demanding job given the number of models and dealers involved. The program has been described as letting Ford respond promptly to problems with its cars while planning future innovation around what customers say they want.
The transferable practice
Ford’s example is about scope and speed. The product, the people who sell and service it, and the next product are three different questions. One program can serve all three if the survey separates them and routes each answer to the team that can act on it.
How to apply it
- List the touchpoints in your customer relationship, such as purchase, delivery, support and renewal, and give each its own short survey rather than one long one.
- Add a forward-looking question, such as what the customer would most like improved.
- Tag every response with the product, location or channel it concerns, so problems can be isolated.
- Decide in advance who receives each type of finding, and how quickly they must respond.
Open-ended questions are especially useful for the forward-looking part, though they take more effort to analyze. We cover the trade-offs in open-ended survey questions.
How does J.D. Power use surveys to compare companies?
J.D. Power is known for customer surveying that lets companies be compared with others in the same industry. For utilities, the factors covered include overall customer satisfaction, customer service and uninterrupted service.
Note that J.D. Power surveys customers of companies in an industry, rather than running a program for a single company. Earlier versions of this article described it as a power company, which was wrong.
The transferable practice
A score on its own is hard to interpret. Is 78% satisfied good? Only a reference point can tell you, whether that is comparable companies or your own earlier results measured the same way.
How to apply it
- Use the same question wording and scale every time, so results from different periods can be compared.
- Where an industry or published benchmark exists, ask comparable questions so you can line your results up against it.
- Report each result against a reference: last quarter, last year, or a peer group.
- Read the gap, not the score. A rise of one point may be noise, as the margin-of-error example above shows.
Our overview of the customer satisfaction survey explains common satisfaction measures and how to read them.
What separates a program that works from one that just collects data?
The programs that stand out take the next step with their data and analyze it to reach real conclusions. Collection is the easy half. A survey earns its keep when someone can say, “this result means we should do that.”
In practice that means three habits. Analyze by segment, since an average hides the location or customer type that is struggling. Test differences before acting on them. And report to the people who own the fix, in a form they can use.
If you do this in a reporting tool, crosstabs and banner tables with weighting and significance testing cover most of it, and Halo Reports provides those along with charts and interactive report pages. To keep watching a tracker after the first read, mTab Pulse monitors trackers and studies and surfaces meaningful movement, and mTab for customer experience teams offers a continuous view of satisfaction drivers, journey friction and emerging issues.
What are the common mistakes when copying these programs?
- Copying the tactic and skipping the follow-through. A prominent invitation gets responses, but responses without owners change nothing.
- Treating incentives as free. Rewards can lift participation and also attract people who answer only to enter. Compare incentivized and non-incentivized answers where you can.
- Reading small differences as real. With a few dozen responses per location, gaps of several points can be noise.
- Changing the wording each wave. You lose the ability to compare over time or against a benchmark.
- Ignoring honesty problems. Respondents sometimes shade their answers, a topic we cover in why people lie on customer surveys.
What should you do next?
Pick one of the three practices and apply it to a single survey this month.
If your response rate is low, move the invitation. If your survey is one long form, split it by touchpoint. If your scores have no context, write down last period’s result next to this one. Then name one person who will act on whatever the survey finds.