How Did You Hear About Us Survey: Attribution Done Right

A how did you hear about us survey is the cheapest piece of market research you will ever run. One question, asked at the right moment, and you learn something your analytics increasingly cannot tell you: where your customers actually came from. Most stores still get it wrong. They bury it in a five-field form, offer options nobody picks, and never reconcile the answers against anything. Done right, it is the closest thing ecommerce has to a free attribution tool.

This guide is the done-right version. What the survey is really for, why it matters more than it did five years ago, the exact question to ask, the answer options that earn their place, and the one habit that stops the data lying to you. There is a copy-and-run template near the end, and an honest account of the job this survey will never do.

What a “how did you hear about us” survey actually tells you

Strip away the form-builder marketing and this is an attribution-focused post-purchase survey. It asks the buyer to name, in their own words or from a short list, the channel that first put you on their radar. That is first-touch attribution, self-reported, collected at the one moment the customer is paying full attention.

That framing matters because it sets the limits. The survey tells you what the customer remembers and is willing to say. It does not tell you the mathematically true first touch, and it never speaks for the people who did not buy. Treat it as a compass, not a GPS, and it becomes one of the most useful questions in your store. Treat it as gospel and it will send you to the wrong place with total confidence.

Why it matters more than it used to

There was a decade when you could half-ignore this question, because your ad platforms and Google Analytics claimed to know exactly where every customer came from. That decade is over, and two things ended it.

The first is privacy. Since Apple introduced App Tracking Transparency with iOS 14.5 in April 2021, iPhone users have to actively agree before an app can track them across other companies’ apps and sites. Most of the tracking your ad pixels rely on now needs that permission. AppsFlyer’s data three years on found that as of early 2024 only about 46 per cent of UK users opt in, a shade under the 50 per cent global figure and ahead of the United States on 44. Put plainly, for something close to half of iPhone traffic, the pixel is guessing.

The second is dark social. A large share of the best marketing you do never leaves a trace in analytics, because it travels through private channels: a screenshot in a group chat, a link pasted into WhatsApp, a recommendation over dinner. Alexis Madrigal named this “dark social” in a 2012 essay for The Atlantic after noticing that a huge chunk of his magazine’s traffic arrived with no referrer and got dumped into the “direct” bucket. More than a decade later, most stores still have a suspiciously large “direct / none” slice in their reports, and most of it is word of mouth they cannot see.

Between the two, your platform analytics has real blind spots, and they sit exactly where the highest-intent demand tends to come from. The buyer is often the only witness who can see the whole journey. Asking them is not a nice-to-have any more. It is how you find the channels your dashboards are quietly hiding.

Where and when to ask

Timing is most of the result. Ask at the peak of attention, which for an ecommerce store is the order-confirmation or thank-you page, the instant the order is placed. The customer has just decided, the memory is fresh, and they are already looking at the screen, so the friction of answering one more question is close to nothing. Wait a week and email it cold and they honestly will not remember whether it was the ad or the friend that tipped them over.

Keep it to a single question in that moment. The response-rate maths backs this up: Survicate’s 2025 analysis of 4,332 surveys found that short microsurveys clear roughly 16 per cent median response, while surveys of seven or more questions drop under 7. Every extra field you bolt on costs you answers from the exact people whose journey you most want to understand.

If you are not an ecommerce store the principle holds, only the moment moves. For a service business or a lead-gen form, ask on the step straight after someone books, enquires or signs up. For an event, ask on the registration form. The rule is always the same: catch the discovery memory while it is still warm.

The one question, and the answer options that earn their place

The question itself is boring, and should stay boring. “How did you hear about us?” works. Do not dress it up, do not make it double-barrelled, do not lead the witness. The craft is almost entirely in the answer options.

Open text, a list, or both

You have three ways to collect the answer, and they are not equal.

An open text box captures answers you never thought to offer, which is the whole point of the exercise. It also costs you responses. Pew Research, which runs this comparison at proper scale, finds that open-ended questions carry far higher item non-response than closed ones: closed questions typically lose 1 to 2 per cent of answers, open-ended ones average around 18. People skip the ones that make them type.

A fixed list is easy to answer and easy to analyse, but it can only ever return the options you already imagined, and it quietly pushes people towards your guesses.

The version that wins for most stores is a short list plus an escape hatch: five to eight named channels, then “Other” with a free-text field. Most people click a channel in a second, and the small number who found you somewhere you did not predict can tell you so. That “Other” box is where the surprising, valuable answers live, so read it every week. Treat the size of that “Other” slice as a signal in itself: if it climbs past fifteen or twenty per cent, your list is missing a channel your customers actually use. Promote the recurring write-in to its own option next month and watch “Other” shrink.

How to build the option list

Start from your own spend and work outwards. Your list should map to decisions you can actually make, so lead with the channels you invest in, then add the routes your analytics can never see.

Keep the whole thing to five to eight visible options. Longer than that and it stops being scannable, so people either bail or click the first plausible line. If a channel is not getting real budget or attention, it does not need its own row. Roll it into “Other” and let the text box catch it.

Randomise the order

One detail changes your data, and it is easy to miss. In surveys people read rather than hear, respondents lean towards the options at the top of the list. Pew calls this a primacy effect, and it means a fixed list hands free credit to whatever channel you happened to put first. Randomise or rotate the order of your channel options between respondents. It does not remove the bias, but it spreads it evenly instead of parking it permanently on one channel. Leave any genuine scale alone, of course. This applies to unordered channel lists, not to ordered ones like a 0-to-10 rating.

The follow-up that doubles the value

If you are willing to spend one more question, and only sometimes you should, ask what actually tipped them over: “Was there a particular reason you bought today?” Attribution tells you the channel. This tells you the trigger, the review, the sold-out restock, the friend’s nudge. It is the difference between knowing someone saw your TikTok and knowing what in the TikTok made them reach for their card. Keep it open text, keep it optional, and only add it if your response rate can carry the extra step.

The mistake that makes the data lie

The single biggest error is treating the survey as the truth and reallocating budget straight off the back of it. Self-reported attribution is one witness, not the verdict.

Remember what each source is actually measuring. Your survey captures a customer’s memory of the first thing that put you on their radar. Your platform analytics captures the last click before the sale. Those are different questions, and they will disagree, which is a feature, not a fault. The job is to read them together.

Take a worked example. Say you sell skincare and spend across Meta, Google, a few creators and email. Your thank-you-page survey comes back: 34 per cent “Instagram”, 22 per cent “a friend or family member”, 18 per cent “Google”, 14 per cent “a creator”, the rest “Other”. Your analytics tells a different story: Google claims most of the conversions, Meta claims a healthy slice, and “direct / none” is suspiciously fat. Both are right about different things. Here is how to reconcile them rather than pick a side.

The pattern to internalise: where the two agree, act. Where they diverge, that is a flag to look closer, not a number to obey. The survey earns its keep precisely in the gaps your analytics cannot fill, and it should be held loosely everywhere else. Self-reported channel data is also a genuine voice-of-customer signal, so tag the free-text answers into themes and let them inform more than just your media split.

One more filter sharpens all of this: separate new customers from repeat ones. Acquisition attribution only means something from a first-time buyer. A loyal returning customer answering “how did you hear about us” is recalling a discovery from a year ago, or simply telling you they are already a fan, and lumping them in muddies the read on which channels are winning you genuinely new people. If your app can tell a first order from a repeat one, judge your acquisition channels on the first-order answers alone.

A template you can copy

Two versions. Pick one, put it on the thank-you page, and resist the urge to add a third question.

The one-question version, for the highest response rate:

How did you hear about us? Instagram / TikTok / Google / A friend or family member / A creator or influencer / Podcast / Other (please tell us) (options shown in random order, with a text box on “Other”)

The two-question version, when you can afford a little drop-off:

1. How did you hear about us? (same option list as above, random order, “Other” free text)

2. Was there a particular reason you bought today? (open text, optional)

Swap the channel list for the five to eight channels you actually spend on. That is the entire survey. It costs a customer about ten seconds and, kept this short, will comfortably outperform the bloated forms it replaces. To put it live, any of the best Shopify survey apps, such as Fairing, KnoCommerce or Zigpoll, will drop this onto your thank-you page and join the answers to real order data.

What a “how did you hear about us” survey can’t tell you

Be clear-eyed about the limit, because it is a hard one. This survey can only ask people who already bought. It is a read on the past, from the small fraction who said yes. It is superb at explaining where your existing demand came from. It is no use at all for the question that comes earlier: will this thing sell in the first place?

You cannot survey your way to that answer, because the people you would need to ask have not been given anything to react to yet. A new product, a new pack, a price you have not tested, an ad you have not run, a name you are still arguing about internally: none of them have customers to interview. That job belongs before you spend, not after, and it means testing the idea against your target audience before you commit stock or budget.

This is the one place our own tool fits, so I will name it plainly. With TestFeed you can put a product, a pack, an in-context price, an ad or a claim in front of simulated shoppers, and get back a purchase-intent read, their reasons in their own words, and a clear next move, in days rather than weeks. It is a pre-spend, directional signal, not a sales forecast, and it does not judge taste, texture or smell. Use it to decide which idea is worth launching, then point a how did you hear about us survey at the customers it brings you. One looks forward before you spend. The other looks back once the money is in.

Frequently asked questions

How do you ask ‘how did you hear about us’?

Ask one plain question at the point of purchase, on the order-confirmation or thank-you page while the decision is fresh. Offer a short list of five to eight channels you actually spend on, add an ‘Other’ box with a free-text field, and randomise the order of the options so the top choice does not soak up lazy clicks. Keep it to that single question if you want the highest response rate.

What are the best ‘how did you hear about us’ answer options?

The channels you actually invest in, plus the ones your analytics can never see. Start with your paid and search channels, add social and creators, then include the untrackable routes: a friend or family member, a podcast, an event, TV or radio, packaging or a flyer. Always finish with ‘Other’ and a text box, and keep the whole list to five to eight options so it stays scannable.

Is ‘how did you hear about us’ survey data accurate?

It is directional, not exact. People name the last thing they remember, which is not always the first touch that started the journey, and the sample skews towards people who finished the purchase. Use it as one witness, not the verdict: put it next to your platform analytics and act where the two agree, investigate where they diverge.

Where should you put a ‘how did you hear about us’ survey?

For an ecommerce store, the order-confirmation or thank-you page, the instant the order is placed and attention is highest. For a service business or lead-gen form, the step straight after someone books, signs up or enquires. Avoid emailing it cold days later, when memory of the first touch has faded and the response rate drops.

Why ask customers when I already have analytics?

Because your analytics no longer sees the whole journey. Since Apple’s App Tracking Transparency, only about 46 per cent of UK iOS users consent to app tracking, and a large share of referrals from private messages and word of mouth land in your ‘direct’ bucket with no source attached. Asking the buyer directly is one of the few honest reads left on where demand actually comes from.

The survey to put live this week

Add one question to your thank-you page: how did you hear about us, with five to eight channels you actually fund, “Other” with a text box, and the order randomised. Read the “Other” answers every week. Once a month, sit the channel breakdown next to your analytics and look for agreement and disagreement rather than a single winner. Spend where both point the same way, dig where they part company, and never move real budget on the survey alone. Do that and you will recover the attribution your pixels have been quietly losing, for the price of about ten seconds of each customer’s time.