If I were in your shoes, I'd start by asking every new customer two specific questions right on the order confirmation page: 'How did you first hear about us?' and 'How long did you know about us before your first purchase?'
I'd start there because that combination gets you more than just a channel name. Jeremiah Prummer made this point on The Bottom Line: Ecommerce Tactics for Profitable Growth, and it’s shaped my thinking. The first question gives you the channel, but the second one gives you the consideration window. This is critical. It helps you understand the difference between a quick conversion from a Meta ad and a slow-burn conversion from a podcast mention that took six months to pay off. Without that context, you might cut the channel that’s building all the upstream value.
In week one, I’d make the 'How did you hear about us?' question a simple, open text field. Don’t start with multiple-choice. As Matt Bahr has pointed out on shows like Honest Ecommerce, you need to hear the answers in the customer's own words first. They'll mention channels and creators you haven't even thought of. It will be messy, but it’s the most honest signal you can get.
After about a month of collecting responses, I'd analyze the free-text answers and standardize the top 80% into a multiple-choice list, keeping an 'Other (please specify)' option. This is your core attribution question. At the same time, I'd start splitting the logic for new and returning customers, a great tip Kurt Elster brought up. For a repeat buyer, I’d ask something different, like 'What brought you back to our site today?'
The main thing I'd ignore is trying to get perfect, one-to-one matches between my survey data and my platform analytics. That's not the point. Dimitri O, from Loop Earplugs, mentioned on Ecommerce Playbook that their survey data showed massive influence from word-of-mouth and influencers that their digital attribution was completely missing. The goal of collecting this zero-party data is to get a blended, more honest view of your marketing, not to perfectly validate last-click reports.
The trap I'd absolutely avoid is viewing the survey as the single source of truth that replaces everything else. A customer's memory isn't perfect. They might say they found you on Google, forgetting they heard about you on a podcast two weeks prior. What you’re gathering are memory-based signals that give you directional clues. Use these clues from post-purchase surveys to ask better questions of your analytics data, challenge your assumptions about what's working, and gain a much richer understanding of your customer's journey.
