If you run paid ads on a Shopify store, you've probably added a post-purchase survey asking "How did you hear about us?" And if you have, you've probably also felt the quiet disappointment of looking at the results.

"Instagram" — 38%. "Google" — 22%. "TikTok" — 17%.

Great. Now what?

You still don't know which Instagram ad to scale, which creator actually drove sales, or whether "Instagram" even means your ads at all. The pie chart looks like data, but you can't make a single budget decision with it.

The problem: "Instagram" is one layer too shallow

Here's what's actually happening. When a customer taps "Instagram," they could mean any of these:

  • A paid ad you're running
  • An organic reel from your own account
  • A post from a creator you paid
  • A friend's story that reshared your product

Those are four completely different budget decisions — scale the ad, post more organically, pay the creator again, or lean into word-of-mouth. And your survey just dumped all four into one bucket labeled "Instagram."

So the channel breakdown isn't just incomplete. It's actively misleading: it looks like an answer, which stops you from asking the real question — why did they actually buy?

Why static surveys can't get you there

The reason almost every post-purchase survey stops at "Instagram" is structural: it's a multiple-choice form. The customer taps an option, the survey ends. The moment they say something interesting, there's no way to ask the obvious next question.

A human interviewer would never stop there. They'd say: "Oh, Instagram — was that an ad, or someone you follow?" But a static form can't. It only knows the questions you predicted in advance.

The fix: ask one follow-up

You don't need a better pie chart. You need one more question.

"Instagram" → "Was it an ad, or someone you follow?" → "A creator I follow" → "What about their post convinced you to buy?" → "A before/after reel — the transformation looked real, and your site reviews backed it up."

Look at the difference. "Instagram" became: a specific creator, a before/after reel, reinforced by on-site reviews. That's not a data point — it's a decision. Pay that creator again. Make more before/after content. Keep the reviews prominent on the product page.

That second layer is where every actionable insight lives. The first answer tells you the channel; the follow-up tells you the reason — and you can only spend money against reasons.

What to do about it

Option 1 — do it manually. On your next 20 orders, when someone answers your survey, follow up by email and just ask the one question: "When you said [X], what specifically convinced you?" It's tedious, but even by hand it'll teach you more than your pie chart has all year.

Option 2 — automate the follow-up. This is the gap we built BetterFeedback to close: an AI post-purchase survey for Shopify that asks the follow-up questions automatically, like a real customer interview, then groups the reasons by source, product, and campaign. There's a free tier, and a live demo you can click through without installing.

Either way, the lesson is the same: stop collecting channels. Start collecting reasons.

FAQ

What's wrong with "How did you hear about us?" surveys? Nothing — the question is fine. The problem is that a single multiple-choice answer like "Instagram" is too shallow to act on. It bundles paid ads, organic posts, paid creators, and word-of-mouth into one bucket, so you can't tell what to scale.

How do I make a post-purchase survey actually useful for attribution? Ask a follow-up. After the channel answer, ask whether it was an ad or organic, which specific content, and what convinced them to buy. That turns "Instagram" into a reason you can spend against.

Can static survey tools ask follow-up questions? No. Multiple-choice forms only ask what you scripted in advance. Asking a contextual follow-up requires AI that adapts to each answer in real time.

Is this just for big brands? No — it matters most for smaller Shopify merchants running paid ads, where every wasted dollar counts and you don't have an analytics team to triangulate attribution another way.