How to Use Shopify Surveys to Investigate Product Returns

Use surveys to investigate why customers report a problem, then use actual order and return records to check its scale and whether a change helped. A low rating is not a verified prediction of a return. Collecting feedback does not itself reduce returns, and this guide promises no percentage improvement.

Start with one product or a clearly defined group of variants. Decide whether you are investigating purchase uncertainty, experience after use, or a completed return. Each calls for a different audience and question.

Separate the three research moments

Do not ask a new buyer to rate fit before they receive a physical product. Do not describe a purchase-confidence question as CSAT about product use. If you need feedback from people who returned an item, select them from verified return records through an appropriate permitted process.

OrderSurvey does not send return or delivery invitations. Its online blocks appear on eligible Thank you and Order status page visits. An Order status visit does not prove receipt, use or a return. A qualifying question can control relevant follow-ups for visitors, but it cannot guarantee that all affected customers are reached.

Build a focused fit investigation

For an online survey, target the relevant product or variant IDs and choose the intended page. Product targeting selects orders containing the item; it is not a dynamic collection-membership rule. Keep the selected IDs current when the catalog changes.

An original example for an apparel item:

1. Ask whether the buyer has received and tried the named item.
2. After a qualifying Yes, ask whether the fit was Smaller than expected, As expected, Larger than expected, or Not sure.
3. Offer an optional follow-up: “Which measurement or part of the fit was different from what you expected?”
4. Finish the branch for people who have not tried it, without asking them to guess.

Use conditional branch testing to check both paths. For a multi-item order, ask which item the answer concerns or use a survey that clearly names one product. Otherwise a comment about one item could be assigned to all items incorrectly.

The satisfaction question bank includes alternatives for packaging, description accuracy and product use.

Reconcile the answer with actual return data

Keep reported concerns and recorded events in separate columns. A survey respondent may describe a fit issue and keep the item. Another buyer may return an item without ever seeing or answering the survey.

Shopify's sales-report documentation distinguishes sales reversals from physical returns: reversal measures can include refunds, cancellations and order edits. For physical-return analysis, check the Returns-table returned-quantity metric and line-item reason. Do not relabel every reversal as a returned product. Verify the metric description and filters in the report you use.

OrderSurvey Unlimited's CSV includes saved question answers and Order ID. It does not include SKU, variant ID, returned quantity, refund value or product cost columns. Join against separately obtained order and return records where the IDs and item context permit it; retain unmatched or ambiguous rows explicitly. This is an analysis workflow, not a built-in returns integration.

The export has a row per question answer. Count the relevant question's answers rather than treating every CSV row as a different customer or completed survey. Keep setup tests out of the customer analysis.

Decide what needs inspection

Use the following as investigation prompts, not automatic diagnoses:

A concentration of comments helps choose what to inspect. It does not by itself prove the size chart, photograph, supplier or carrier caused the problem. Include comments that contradict the initial theory.

Assign an owner to review the answers and investigate urgent issues through your normal support process. Ordinary fit ratings do not trigger OrderSurvey's low-NPS alerts; those alerts require an actual NPS answer, Unlimited and a configured webhook threshold.

Evaluate a change using comparable purchase cohorts

Suppose you correct an inaccurate measurement after checking the product. Record the product/variant, the old and new information, and the publication date. Then compare purchases before and after the change with the same opportunity to return.

An illustrative example, not a customer result: 10 physically returned units from 100 eligible units purchased is 10%; 6 from another 100 is 6%. That is a four-percentage-point difference. It does not prove the page change caused the difference. Promotions, product mix, fulfillment, seasonality and customer mix could also have changed.

Specify whether your unit of analysis is items or orders. Give both cohorts enough follow-up under the return policy. Returns processed this week may concern purchases from earlier weeks, so dividing this week's processed returns by this week's orders can mix unrelated cohorts.

Track reasons separately from rates. A fall in one stated reason with unchanged overall returns may still merit investigation. A lower overall rate does not show that a particular issue was resolved for every customer.

Estimate costs from your own records

There is no universal cost per return in this guide. Use your own carrier, handling, inspection, inventory recovery and refund records. Keep money refunded, product cost and recoverable inventory value distinct, and avoid counting the same loss twice. A return-reason survey supplies none of those financial figures automatically.

This keeps prioritization grounded in the actual problem: the affected products, the number of observed returns, the experience customers describe and the cost you can substantiate.

Start with one investigation

Install OrderSurvey to collect relevant feedback, and use the setup walkthrough to publish the intended blocks. The segmentation guide explains supported audience rules.

Free includes 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 and configured low-NPS webhook alerts. Compare the plans or book setup help to verify a focused survey before collecting customer answers.

Frequently asked questions