What is a how did you hear about us survey?

It is a single question, usually shown on the order confirmation page, that asks a new customer where they first heard about the store. The answers give self-reported attribution, which catches channels click tracking cannot see, such as podcasts, word of mouth and AI assistant answers. It works best as a check on click-based attribution rather than a replacement.

What does HDYHAU stand for?

HDYHAU stands for how did you hear about us. In e-commerce it usually refers to a post-purchase attribution survey that asks customers which channel first introduced them to the brand. Answers are typically collected on the Thank you page with a short list of channels and an Other option.

Where should a post-purchase survey go on Shopify?

On the Thank you page, which Shopify describes as a one-time page customers see after checkout, and optionally the Order status page. Since Shopify's 26 August 2026 deadline for upgrading those pages, surveys should be added as app blocks through the checkout and accounts editor rather than pasted into the old Additional Scripts box.

What answer options should a HDYHAU survey have?

Build them from a few weeks of open-ended answers, then list the channels customers actually name, such as a friend, a podcast, Instagram, TikTok, YouTube, Google search and ChatGPT or another AI assistant, with Other and a text box last. Randomise the order of everything except Other, because Pew Research Center finds people in self-administered surveys favour options near the top of a list.

Should I trust survey attribution or click attribution?

Neither on its own. Clicks are precise but blind to podcasts, word of mouth and recommendations that were read but not clicked, while surveys capture those but rely on memory and only reach customers who answer. Compare survey shares with a first-touch click model channel by channel, and use a holdout test where they disagree on a channel you spend heavily on.

How Did You Hear About Us Survey: A Guide for Stores

How to run a how did you hear about us survey after checkout: question wording, answer options, placement, response rates and reconciling it with clicks.

Click tracking only sees what arrives as a click. The podcast mention, the friend at dinner, the ChatGPT answer read on a phone and the link pasted into a group chat all reach your store as direct traffic or branded search, if they leave a trace at all. A how did you hear about us survey, asked once after checkout, is the cheapest way to hear about them, and the easiest to misread.

The short answer: ask one question after purchase, then check it against clicks, with Arktis on the click side

A how did you hear about us survey, often shortened to HDYHAU, is a single question on the order confirmation page asking a new customer where they first heard of you, with a short list of answers and an Other box. It captures self-reported attribution: what the customer remembers, which includes channels no tracker can see. It is biased in its own ways, so it works as a second opinion on click-based attribution rather than a replacement for it.

Arktis does not run surveys; you run the survey in a Shopify post-purchase app or on your own confirmation page. For the click side of the comparison, Arktis is the clear winner for an online store: it records first-touch and last-touch sources per customer, flags dark social visits from shared links and counts sessions and conversions from AI assistants, which are exactly the categories a survey tends to surface. Plans are $49, $149 and $349 a month, with a 7-day free trial on Growth.

Channel	Click-based tracking	Post-purchase survey
Paid social and search	Strong, down to the campaign	Channel level only
Podcasts and radio	Only with a code or vanity URL	Strong
Word of mouth	Invisible	Strong
AI assistants	Partial, when the click keeps its source	Catches the answer read and acted on later
Links shared in messaging apps	Partial, as dark social	Captured if the list offers it
Organic search	Strong	May be named when it was only the last step

Key takeaways

Put one question on the Thank you page, asked of first-time buyers, with the word first in it. Build the answer list from an open-ended pilot, randomise the order, and keep Other last with a text box. Treat response rate as a number to report alongside every result, because the people who answer are not a random sample. Then compare the survey's shares with first-touch click data channel by channel: where they agree, trust both; where the survey is much higher, you have found a channel clicks cannot see; where clicks are much higher, customers may be forgetting an ad.

What a survey catches that click tracking misses

Click-based attribution depends on the visit carrying a source: an ad click identifier, a UTM tag or a referrer. Anything that influences a purchase without a click in the same browser falls outside it. A podcast host reading your name, a recommendation from a colleague and a product mentioned in a ChatGPT answer that someone reads on their phone and acts on at their desk the next day all arrive as direct traffic or a branded search.

Some of that leaves partial traces. ChatGPT adds utm_source=chatgpt.com to the links it cites, so a clicked recommendation can be tracked, and Arktis files deep-link visits with no referrer and no UTM tag as dark social, which is where a link pasted into WhatsApp tends to land. But neither tells you about the person who heard your name and typed it in later. The survey is the only instrument that asks them.

How to write the question

Ask how the customer first heard about you, not why they bought or what made them buy today. The word first points the answer at discovery, which is the part tracking misses, rather than the retargeting ad or search that closed the sale. One question is enough; a conditional follow-up, such as which podcast, can come after for customers who choose that answer.

Pew Research Center's guidance on survey questions is a good free reference here, and two of its findings apply directly. First, the list you offer shapes the answers you get. In one Pew poll, 58 percent chose the economy when it was offered as an option, against 35 percent who volunteered it in an open-ended version. Second, in self-administered surveys people tend to pick items near the top of a list, which Pew calls a primacy effect, and it often randomises answer order to counter it.

So build the list from evidence. Run the question open-ended for the first few weeks, as Pew describes researchers doing in pilot studies, group the free-text answers, and turn the common ones into options. Name channels the way customers think of them: a friend or family member, a podcast, Instagram, TikTok, YouTube, Google search, ChatGPT or another AI assistant, an article or review, then Other with a text box. Randomise everything except Other, which stays last. Avoid umbrella options like social media or online, which feel easy to pick and tell you nothing.

Where to put it

The Thank you page is the natural home. Shopify describes it as a one-time page shown after checkout that the customer cannot access again, which means the question is asked right after the purchase, and only once. The Order status page, which customers revisit to track an order, is a second placement that some apps also support.

For Shopify stores the route changed this year. Shopify set 26 August 2026 as the deadline to upgrade and replace customisations on the Thank you and Order status pages, and stores not upgraded by then were upgraded automatically, so surveys that relied on the old Additional Scripts box have to be replaced. The supported way is an app that adds a block to those pages through the checkout and accounts editor. Several survey apps in the Shopify App Store are built for exactly this question, including Fairing, KnoCommerce and Zigpoll. If you sell through Stripe or a custom checkout, the same question can sit on your own confirmation page.

Ask first-time buyers only, or at least tag answers by new and returning customer. A repeat buyer asked how they heard about you is answering a different question.

Response rates, honestly

Survey vendors publish their own response-rate figures, and they are worth reading as claims. KnoCommerce's homepage, for example, advertises a 45 percent average response rate on its post-purchase surveys, without a published method behind it. Your own rate will depend on your audience, the device mix and how prominent the question is, so measure it rather than assuming it.

The bigger issue is who answers. Respondents are the customers willing to click one more thing after paying, and there is no reason to assume they discovered you the same way as the ones who closed the tab. Report the response rate next to every chart, compare shares among respondents rather than scaling them up to all orders, and be wary of any channel with only a handful of answers in a month.

Reconciling the survey with click-based attribution

The survey asks about first discovery, so compare it with a first-touch click model rather than last-touch. Our explainer on first-touch vs last-touch attribution covers why the two disagree.

Take an example store with 400 first orders in a month and 200 survey answers. Suppose 24 percent of respondents say a podcast, while first-touch clicks give podcasts 2 percent through a vanity URL. Suppose 20 percent say Instagram, while first-touch clicks credit Meta ads with 34 percent. Neither source is wrong. The podcast gap is the survey finding a channel clicks cannot see; the Meta gap is customers forgetting an ad they clicked, or remembering the organic post instead.

Turn the comparison into a ratio per channel, survey share divided by click share, and track it month to month rather than judging a single month. A stable ratio lets you use click data for day-to-day campaign decisions and apply the survey as a correction for the channels clicks undercount. A ratio that jumps is a prompt to investigate. When the two methods disagree about a channel you are spending real money on, settle it with a holdout test, as described in our guide to incrementality testing.

AI assistants deserve their own line in that comparison. Compare the share of customers who answer ChatGPT or another AI assistant with the AI referrer sessions and conversions in your analytics; the gap between them is roughly the recommendations that were read but not clicked. Our guide to tracking ChatGPT traffic covers the click side.

Limits, including ours

A survey measures memory, and memory is selective. Customers compress a journey of several touches into one answer, and some will name Google when Google was only where they typed your name. Pew notes that people also shade answers toward what seems acceptable. The survey reaches only buyers, so it says nothing about the channels that attract browsers who never convert, and it will never tell you which ad set or creative worked.

Arktis is ours, and it has no on-site survey tool, so it will not ask the question or store the answers. What it gives you is the other half: Shopify orders matched to the visitor on email, or on the UTM tags in the landing URL against a session from the previous 24 hours, conversions counted per UTM campaign and per ad platform from matched orders, next to the Meta spend it syncs per campaign, a comparison of first-touch, last-touch, linear, time-decay and position-based credit per ad platform, dark social detection and an AI referrer report. You put the survey export next to those numbers yourself.

To see the click side on your own store, start the 7-day Growth trial, or compare the tiers on the pricing page.

Sources

Pew Research Center: Writing survey questions, open and closed questions, the 58 and 35 percent example, primacy effects and randomisation, accessed 24 September 2026

Shopify Help: Upgrading and replacing your Thank you and Order status pages, the 26 August 2026 deadline and automatic upgrade, accessed 24 September 2026

Shopify Help: Customization options for Thank you and Order status pages, app blocks through the checkout and accounts editor, accessed 24 September 2026

KnoCommerce, the vendor's advertised 45 percent average response rate, accessed 24 September 2026

Fairing on the Shopify App Store, post-purchase attribution surveys on Thank you and Order status pages, accessed 24 September 2026

Zigpoll on the Shopify App Store, post-purchase surveys, accessed 24 September 2026

OpenAI Help: Publishers and developers FAQ, ChatGPT adds utm_source=chatgpt.com to referral URLs, accessed 24 September 2026