What to ask on your Shopify thank-you page (and what to leave out)

The thank-you page is the most valuable unused real estate in most stores. The buyer is still there, still in the mindset of the purchase, and — for once — has nothing else to do. Whatever you ask them here will get a better answer than the same question in an email three days later.

Which is exactly why it is worth being careful about what you ask. You get one short window per order. This is a guide to spending it well, regardless of which tool you use to run the survey.

Start from a decision, not a question

The test for any survey question is simple: if the answers come back overwhelmingly one way, what will I do differently on Monday?

If you cannot answer that, cut the question. “How satisfied were you overall?” usually fails the test — a 4.2 average tells you nothing you can act on. “How was checkout and payment?” passes, because a bad score points at a specific screen you can go and fix.

Write the list of things you are willing to change first. Then write questions that tell you which one to change.

Keep it to three questions

Every question you add costs you responses on all the others. There is no exact threshold, but the shape of the trade-off is consistent: a survey that looks like it will take ten seconds gets answered, and one that looks like a form does not.

A good default for a post-purchase widget:

1. One rating per journey stage you actually control. Finding the product, the product page itself, checkout. Three stars-rows read as one glance, not three questions.
2. One follow-up that only appears when something went wrong. Low score in, specific reason out.
3. One optional open box. For the thing you did not think to ask.

That is it. Resist the temptation to bolt on demographics, NPS, a newsletter opt-in and a referral prompt — each is defensible on its own and fatal in aggregate.

Rate stages, not the whole experience

A single overall rating is a thermometer. It tells you the store feels warm or cold, and leaves you guessing about which room the draught is coming from.

Breaking the same 1–5 scale across the stages of the journey costs the buyer almost no extra effort — they are still just tapping stars — and turns the result into a map. “Overall 4.4, checkout 3.1” is a week’s work with a clear target. “Overall 4.1” is a meeting.

Pick stages that correspond to something you can change: search and filtering, product information, checkout, delivery promise, account creation. Skip anything you have no lever over.

Make bad scores cheap to explain

Most people will not write you a paragraph. They will tap a chip.

When a rating comes in low, show a short list of plausible reasons — shipping cost, too many steps, confusing options, payment problem, slow page — and let one tap do the work. You lose some nuance versus free text, and you gain something more useful in practice: counts. Fifteen taps on “shipping cost” this week versus three last week is a signal you can watch over time, which a pile of prose is not.

Keep the list under about six items, phrase them in the buyer’s language rather than yours, and always leave a way to write something instead.

The one question your analytics cannot answer

“How did you hear about us?”

Attribution tooling is good at the parts of the journey with a URL and bad at everything else. Word of mouth, a podcast mention, a friend’s screenshot, a physical shop, an aggregator someone read weeks ago — none of it survives a last-click model, and increasingly it does not survive privacy settings either.

Self-reported attribution is famously imprecise. Buyers misremember, and the option they see first gets picked more often. It is still worth asking, because the alternative is not precision — it is a blank. Read it as a directional supplement to your analytics, not a replacement, and keep the option order stable so week-over-week comparisons mean something.

Questions to leave out

Anything you already know. Order value, items purchased, shipping method: it is on the order. Asking wastes the buyer’s patience and yours.

“Would you recommend us to a friend?” as your only question. NPS is a boardroom metric — a good longitudinal number, a poor diagnostic. On its own it cannot tell you what to fix.

Product quality questions. The buyer has not received the product yet. Anything about how the item performs belongs in a review request sent after delivery, which is a different tool doing a different job.

Leading questions. “How much did you love our new checkout?” produces answers that make you feel good and mean nothing. Neutral wording, symmetrical scales, no adjectives on the response options.

Sensitive or personal fields. Never ask a post-purchase widget for anything you would not want sitting in a feedback record: no addresses, no birthdays, nothing about health or finances. You already have the contact details from the order.

About incentives

Offering a small discount for feedback lifts response rates and skews them slightly positive — people who just accepted a gift rate a little kinder. If you use one, use it consistently rather than switching it on during bad weeks, or your trend line will measure your promo calendar instead of your store.

The cleaner version: give the reward for responding, never for responding well, and never make it conditional on a rating.

Then actually close the loop

Feedback you do not act on is worse than no feedback, because it costs you the goodwill of asking. Two habits make the difference:

Reply to the bad ones. A buyer who rates checkout 2 and gets a real human reply the same week is more loyal than one who sailed through happily. This is the highest-return use of a post-purchase survey and almost nobody does it.

Watch the trend, not the day. Individual ratings are noisy. Weekly averages by stage, with your release dates marked, will show you exactly which change helped — and which of your improvements made things quietly worse.

A survey you can copy

If you want to stop thinking about it and just start collecting:

• “How was your shopping experience?” — 1–5 stars on Overall, Finding what I wanted, Checkout & payment
• On any rating of 1–3: “What went wrong?” → shipping cost · confusing options · too many steps · payment error · slow to load · something else
• Optional: “How did you hear about us?” → a short, stable list plus “other”
• Optional: a free-text box, unlabelled and unpressured

Three taps, ten seconds, and a map of your store’s friction inside a fortnight.

That is the survey Poby ships by default on the Shopify thank-you page — but the shape matters more than the tool. Ask few questions, ask them at the right moment, and be ready to change something when the answers come in.