How a post-purchase survey works
Immediately after someone buys or submits an enquiry, you ask a single question: how did you first hear about us? The answer is optional, the options are short, and there is a free-text box for anything that does not fit. Because the customer has already committed, the question costs you nothing in conversions, and the moment of purchase is when memory of the journey is freshest.
The output is not attribution in the technical sense. It is self-reported, so it records what the customer believes rather than what a cookie observed. That is precisely its value: it can name a friend’s recommendation, a WhatsApp group, a podcast, a shop sign or a Facebook group that no tag on your site could ever see.
Why post-purchase surveys matter
Tracked channels are only the visible part of demand. Conversations, messaging apps, forwarded links and offline word of mouth arrive as direct traffic with no source attached, and every analytics system quietly credits them to whatever tagged click happened to be nearby. A survey gives you an independent reading that no platform controls.
It is also the cheapest measurement available. Where a proper experiment needs volume, budget and patience, this needs one field on a thank-you page. For small businesses, and for markets like Nepal where a great deal of discovery happens in Facebook groups, on TikTok and through family recommendation, it often explains a gap that months of tag debugging never would.
Where it goes wrong
Long lists are the first failure. Give people many similar options and they pick the first plausible one, which is usually whichever platform is listed at the top. Keep the list short, put the options in a rotating or neutral order, and always allow a free-text answer.
The second failure is wording that invites the wrong memory. “How did you find our website?” gets the answer “Google”, because almost everyone arrives through a search box regardless of what prompted them. Ask how they first heard of you instead. The third is over-reading small numbers: a handful of responses in a week is a hint, not a finding, and monthly totals with the free-text answers read properly are far more useful than a weekly pie chart.
How to get it right
Ask one question, keep it optional, and place it after the transaction is complete so nothing is at risk. Store the answer against the order so it can be compared with what the ad platforms claimed for the same customers. Review the free text every month; the phrases people use are often better campaign copy than anything written in a planning meeting.
Then use it as one input among several rather than a verdict. Read it alongside your platform reports and your own sales records, in the way described under triangulation, and treat a persistent disagreement as a question worth investigating rather than proof that either side is lying. Setting it up is usually a small change to the checkout or enquiry flow, which sits naturally with the rest of a tracking and reporting setup.
