NPS vs CSAT vs CES: Which Metric Fits Your Shopify Survey?

NPS asks whether a customer would recommend you. CSAT asks how satisfied they were with a particular experience. CES asks how easy a task felt. Pick the metric that helps your team make a specific decision, then choose a moment when the customer can answer it.

For a Shopify checkout study, CSAT or CES can describe the purchase just completed. An NPS question can describe willingness to recommend at that moment, but a post-checkout answer does not prove satisfaction with a product the buyer has not yet received.

NPS vs CSAT vs CES at a glance

The scores are different measures. An NPS of 50, a CSAT of 75% and a CES average of 5.6 out of 7 cannot be ranked as if they shared a scale. None proves that a particular customer will refer someone or buy again.

How to calculate each score

NPS: recommendation on a 0–10 scale

Ask: “How likely are you to recommend our store to a friend or colleague?” Use the NPS question type in OrderSurvey, with the standard 0–10 scale.

Classify valid answers as promoters (9–10), passives (7–8) and detractors (0–6). Calculate:

NPS = percentage of promoters − percentage of detractors

The denominator includes all valid NPS answers, including passives. The result ranges from −100 to +100 and is usually reported as a score, not a percentage. This follows Bain's NPS calculation.

Illustrative example: 100 valid answers include 60 promoters, 30 passives and 10 detractors. NPS is 60 − 10 = 50. These are invented numbers to demonstrate the formula, not OrderSurvey merchant results.

Use the NPS calculation guide for the denominator and common reporting mistakes. The group labels describe answers; they are not observed referral or retention outcomes.

CSAT: satisfaction with one experience

Ask: “How satisfied were you with placing your order?” Define the scale clearly. For a 1–5 satisfaction scale where 4 and 5 mean satisfied, a common calculation is:

CSAT = answers of 4 or 5 ÷ all valid answers to that question × 100

That convention is described in Qualtrics' CSAT guide. If you choose different labels or a different scale, document the calculation instead of silently applying the same cutoff.

Illustrative example: 30 of 40 valid answers are 4 or 5. CSAT is 75%. A mean star rating would be a different statistic; label it as an average rating rather than substituting it for this percentage.

OrderSurvey's 1–5 star rating can collect the answers. Make the question specific to checkout, order information or another experience the respondent has actually had.

CES: ease of one task

Ask: “How easy was it to place your order?” For this example, use seven labeled choices from 1, very difficult, to 7, very easy. OrderSurvey does not have a dedicated CES question type; use a single-choice or dropdown question.

CES formats and score directions vary. Qualtrics' overview describes its focus on effort during a particular interaction. For the scale defined here, you can report the mean answer, with higher values meaning easier:

Mean ease score = sum of valid numeric answers ÷ number of valid answers

Illustrative example: ten answers total 56 points, so the mean is 5.6 out of 7. Include the question, endpoints and calculation with the result. Do not compare it directly with a reversed scale or a percentage-based CES method.

Match the question to what the buyer knows

Thank you page: the buyer has just completed checkout. Ask about placing the order, finding information or the reason for buying. They may not yet have a delivery or product-use experience to evaluate.

Order status page: the visitor is returning to an order, potentially before shipment. This page is not a delivery timer. OrderSurvey does not automatically wait for confirmed delivery, a week of product use or a support-ticket resolution, and it does not send scheduled email or SMS surveys. If those milestones define your study, you need a workflow that verifies them.

Shopify POS: staff can open the survey action after a completed sale. Ask about the retail visit. The POS experience has different targeting and branching limits; follow the POS setup guide.

Keep the question, audience and invitation timing consistent when comparing periods. A change in the mix of first-time and repeat buyers can change a score even if the underlying experience has not changed. See the Thank you versus Order status guide for placement choices and first/repeat-order prerequisites.

Set up one useful question in OrderSurvey

1. Write down the decision. For example: “We need to investigate where checkout feels difficult.” This would support a CES-style question rather than a general recommendation score.
2. Choose the appropriate question type and labels. Use NPS for the standard 0–10 recommendation scale, a star rating for a clearly defined 1–5 satisfaction scale, or single-choice options for the CES scale above.
3. Add an optional follow-up. “What could we improve?” can help explain a score. Configure conditional follow-ups if the next question should depend on an answer; verify every intended branch.
4. Enable the survey and publish its blocks. Add and save the OrderSurvey block separately on each selected Shopify page. Use the setup walkthrough.
5. Verify a saved response, then review real answers. Use approved test orders and keep their references so test answers can be excluded from customer reporting.

There is no supported basis here for promising that one metric always receives more answers than another. Start with a question your team can act on and inspect the actual results.

Review scores, exports and alerts accurately

Free includes analytics and individual response review. Open Analytics in the app menu for the dashboard and Responses to read answers. The NPS dashboard uses NPS answers from the last 90 days; ordinary ratings and numeric choices are excluded. Keep its date range in mind when comparing it with a separate spreadsheet calculation.

For a CSAT percentage or CES mean, select the relevant question and reporting period and calculate the defined measure from valid answers. Exclude test answers and report the answer count alongside the score. A multi-question survey can produce multiple answer rows; those rows are not all separate customers.

Unlimited CSV export from Responses contains answer dates, order IDs, questions, answers, other text, placement, tags and customer-tag status. It does not include order value or product details. Filter by survey before exporting; matching an answer to order revenue or products requires a separate, authorized order-data join. The agency reporting guide explains a practical reporting workflow.

Configured low-NPS webhook alerts require Unlimited. They apply to an NPS answer at or below the selected threshold. A CSAT star rating or a CES numeric choice does not trigger the alert. Settings can retain alert details on Free, but alerts do not send on Free. See the NPS follow-up guide for the supported workflow.

Free includes unlimited surveys, branching, targeting, analytics and 100 submitted responses across the store per rolling 30 days. Unlimited is $24 USD per store per month for unlimited responses, CSV export, configured low-NPS webhook alerts and priority support. Compare the plans.

Install OrderSurvey to start with one useful question, or book personal setup help to verify the question, both page placements and the first saved response.

Frequently asked questions