How did you hear about us survey: post-purchase attribution
Run a post-purchase "how did you hear about us" survey beside click attribution: what each sees, how to code answers and how to compare them by order.
By DATALYR · · 4 min read
Contents
What a post-purchase survey measures
A "how did you hear about us" survey asks the buyer, right after checkout, which channel brought them to you. It measures what the customer remembers, which makes it the best tool you have for podcasts, creators, word of mouth and other channels that rarely produce a trackable click.
Click attribution measures something else: which ad click or tagged link preceded the order. It is precise about paid clicks and blind to anything that happened off-screen. The two methods answer different questions, so the useful setup runs both and compares them order by order.
Neither is the truth on its own. A buyer who saw five Instagram ads and then searched your brand name may answer "Instagram" while the click data credits branded search. Both records are correct about what they observed.
Design the question so answers can be counted
Use a single-choice question with a short, fixed list of channels and an "Other" option with a text box. Free text alone is hard to tally, and a list longer than eight or nine options gets skimmed.
Name channels the way customers think about them. "Instagram or Facebook" is easier to answer than "Meta Advantage+". Add a follow-up only where it pays off, such as asking which podcast after someone picks "Podcast".
On Shopify, survey apps usually place the question on the Thank you page as an app block. Shopify's checkout UI extensions can render on the Thank you page and list a post-purchase survey as an example use. Blocks on the information, shipping and payment steps are limited to Shopify Plus.
- Keep one fixed option list and change it rarely, so months stay comparable.
- Randomise option order if your survey app supports it, to reduce first-option bias.
- Store the order ID with every answer. Without it you cannot compare against click data.
- Code "Other" answers weekly into existing options or a new one.
Worked example: survey answers next to click data
Illustrative example. A store takes 400 orders in a month and 260 buyers answer the survey, a 65% response rate. Answers: Instagram or Facebook 104 (40%), Podcast 52 (20%), Friend or family 39 (15%), Google 39 (15%), TikTok 26 (10%). That adds up to 260.
Click attribution for the same 400 orders: Meta 150, Google 90, and 160 with no paid click. Podcasts produce no click, so they never appear in the click report. In the survey they are one buyer in five.
If respondents are representative of all buyers, which is an assumption you should test, podcasts influenced about 20% of 400, or roughly 80 orders. That does not move 80 orders out of the Meta or Google columns. It tells you the unattributed 160 contains a real podcast effect worth funding and testing.
One month, two lenses (illustrative)
Survey: Podcast
52 of 260 answers
20% of respondents name a channel with no clicks.
Clicks: Podcast
0 of 400 orders
No tracked link, so no click credit.
Survey: Instagram or Facebook
104 of 260 answers
40% of respondents remember Meta placements.
Clicks: Meta
150 of 400 orders
37.5% of orders follow a Meta ad click.
Compare the two by order, not by total
Join survey answers to your order-level attribution on the order ID. Then look at the agreement table: how many orders the survey and the click data credit to the same channel, and where they differ.
Disagreements are the useful part. Orders with a Meta click where the buyer answers "Friend" suggest word of mouth that ads then converted. Orders with no click where the buyer answers "TikTok" point to view-driven demand that click tracking cannot see.
Track response rate per channel too. If buyers from one channel answer less often, the survey under-counts that channel. A falling response rate after a theme or checkout change usually means the block stopped rendering.
Use each signal for the decision it suits
Use click attribution for daily ad decisions: which campaign, ad set and creative brought paid orders and revenue. It is fast, granular and tied to spend.
Use the survey for channel-level budget questions, especially for channels without clicks. When the survey keeps naming a channel your click data cannot see, run a holdout or geo test before moving large budgets.
DATALYR handles the click side for Shopify: each paid order is linked to the storefront visit and ad click that preceded it, and revenue comes from Shopify transactions rather than browser values. Keep your survey app for the self-reported side, and compare the two on order ID.
Related articles.
Measurement ·
Incrementality: measure the lift attribution cannot prove
Use a worked holdout example to distinguish attributed sales from incremental lift, define a test, and interpret the result before changing spend.
Attribution ·
Multi-touch attribution: connect and allocate a customer journey
Follow a three-touch purchase, compare fractional credit, and check identity, eligibility, and revenue reconciliation before trusting a multi-touch report.
Attribution ·
Marketing attribution models: choose the rule for your question
Compare first-touch, last-touch, linear, time-decay, and data-driven attribution while keeping eligibility, revenue, and causal limits explicit.
