Post-purchase survey questions for ad research

You have had "How did you hear about us?" on the thank-you page for a year. Eleven options in a dropdown. Instagram wins every month, TikTok is second, "Friend" is third. That is the whole finding.

Thousands of buyers answered. Nobody on the team could write a single ad from it.

The survey is fine. The question is doing a job it was never built for.

Short answer: A post-purchase survey for ad research needs two questions doing two different jobs: a closed "where did you first hear about us?" for a rough channel cross-check, and one open "what almost stopped you from buying?" for copy. The open answer is the valuable one, because buyers describe the objection in words your ads never used.

The takeaways

• A dropdown can only return words you wrote. Eleven channel options give you a share-of-voice chart and zero hooks. One optional text box under it is where the copy comes from.
• "What almost stopped you?" beats "why did you buy?" Ask why they bought and you get "good quality". Ask what nearly stopped them and you get the objection, with the detail that makes it usable in an ad.
• Buyers are the survivors. Every answer comes from someone whose objection lost. The objection that beat the people who bounced is not in the data at all.

Why does the "how did you hear about us" dropdown produce no ad copy?

The "how did you hear about us" dropdown produces no ad copy because you wrote every possible answer before a buyer saw it. A closed question hands your own vocabulary back to you. "Instagram" tells you where somebody scrolled. It says nothing about what they were looking for when they stopped.

Keep it anyway. A one-tap channel signal is cheap and nearly everyone answers it. Two changes make it better. Add an "Other" option with a text field, because what people type there ("my physio mentioned you", "a YouTube review of your competitor") names the channels you did not know to list. And ask where they first heard about you. A buyer who saw a TikTok in March and clicked a retargeting ad in May answers whichever touch they remember, so name the moment you want.

Which question turns buyers into ad copy?

The question that turns buyers into ad copy is "What almost stopped you from ordering?", placed as one optional text box under the channel dropdown. It catches the objection at the moment the buyer has just beaten it, while the detail is still fresh: the size they were unsure about, the delivery time they assumed, the subscription they feared they could not cancel.

"Why did you buy?" gets polite, vague answers. "Good reviews." "Looked nice." Nothing you could put in a headline.

If you want a second open question, use "What were you using before this?". It hands you the before-state, the raw material for a hook aimed at people who have not switched yet.

Dropdown first, text box second. The tap gets nearly everyone. The text box gets the people with something to say.

How far can you trust the channel answer?

You can trust the channel answer as a direction and never as a count. People misremember. They credit the touch that felt memorable, a creator, a friend, a podcast, and forget the three paid impressions that came before it. Retargeting ads get named rarely, because nobody remembers the reminder.

So read the dropdown against Ads Manager over time. If Meta claims a rising share of purchases while the survey share of Instagram and Facebook stays flat for two months, ask why. If both move together, that is weak confirmation. Neither proves cause.

Survey vendors like to sell this as the attribution answer the pixel lost. It is a second opinion from a witness with a bad memory. Whether the ads caused the sales takes a holdout; the affordable versions are in incrementality testing for ads.

How many answers do you need before a pattern is worth testing?

You need fewer answers than a survey vendor would suggest, because the survey's job here is to produce hypotheses, and the ad account decides which ones hold. My own habit: once the same objection turns up three times, in different words from different buyers, it earns an ad.

Different words matters more than the count. Three people writing "shipping" is one thin signal. "Thought it would come from China", "needed it before Saturday" and "didn't want to pay 6 euro to send it back" are three distinct fears about delivery, and each one is a different line of copy.

Stop collecting a category when ten new answers produce no new phrasing. That is saturation, the only honest stopping rule for open text.

How do you sort the answers into angles?

Sort the answers into three buckets, and each bucket feeds a different part of the ad. Pain is the before-state ("my dog kept slipping out of every harness"), and it writes hooks. Objection is the near-miss ("thought it would run small"), and it writes the body and the proof. Trigger is what made today the day ("he got out again on Sunday"), and it writes urgency and the offer.

Keep each answer verbatim. The phrasing is the asset. Tidy "kept slipping out" into "escape-proof design" and you have thrown away the reason the answer was worth collecting.

Objections deserve the most attention: a buyer who names one is describing what many non-buyers silently decided. Answering one in an ad without sounding defensive is its own craft, covered in handling objections in ad copy.

What can a buyer survey never tell you?

A buyer survey can never tell you why the other visitors left. Every respondent is someone whose objection lost. The people whose objection won closed the tab, and their reasons appear nowhere in your data.

No wording fixes that gap, and no sample size does either. It is why public sources complement your own survey: competitor reviews, Reddit threads and app-store complaints include people who never bought from anyone.

Adscalr does not ingest your survey responses. What it does is the public half: Audience Intelligence searches five named voice-of-customer sources (Reddit, Amazon reviews, the App Store, Google Play and custom forums), captures 2 to 4 phrase markers per quote along with emotion and register, and ranks the pain map by intensity. Your thank-you page tells you why buyers bought. The public corpus tells you what everyone else is still stuck on.