How to find out where your customers really come from

Your ad platforms each claim the same order, and neither can see a friend’s recommendation. Asking the customer is the only way to close that gap.

Your ad dashboards add up to more sales than you actually made. Meta claims an order, Google claims the same order, and your Shopify revenue sits well below the two of them combined. Nobody is lying — each platform can only see its own slice, and both count the same customer.

A post-purchase survey asks the person instead. It is the cheapest attribution you will ever run, and the only one that can see a channel where no click happens at all.

Why click tracking stopped being enough

Three things broke it at roughly the same time:

• Tracking got blocked. After iOS App Tracking Transparency, most people on an iPhone never opted in, so platforms model their results rather than measure them.
• Buying moved off-click. Someone watches a TikTok, thinks about it for a week, then types your name into Google. The purchase gets filed under Google search, and the video that actually did the work gets nothing.
• Word of mouth leaves no trail at all. A friend recommends you in a message you will never see. In every dashboard that order appears as “direct”.

Which is why the channel many stores grow on — people telling each other — is invisible in exactly the reports they use to decide where to spend.

What a survey sees that analytics cannot

Ask “How did you first hear about us?” on the thank-you page and you get three things no tracking pixel can give you:

• The first touch, not the last click. People name where they discovered you, which is usually not where they clicked from.
• Channels with no click. A friend, a podcast, a physical shop, a sign at a market.
• Their words. Answers typed into “Other” tell you about places you did not know you were being talked about.

How to read the answers honestly

It is a second opinion, not the truth

Memory is imperfect and people under-report ads — nobody likes to think an ad worked on them. Treat survey results as one of two instruments, and get interested when the two disagree.

Watch the gap, not the number

The useful signal is the difference. If 20% of customers say TikTok while your reports attribute 4% of revenue there, TikTok is doing more than you are crediting it for — and probably deserves more budget than its dashboard argues for.

Read it as a share, not a count

Only a portion of customers answer, so raw counts mean little. Compare the shares across months instead, and change one thing at a time so the movement means something.

Keep the option list stable

Change the options and you have broken your own trend line. Add a channel only when you genuinely start spending there, and leave the rest alone.

Getting enough answers to trust

A survey that 7% of people finish will mislead you, and the people who do finish are the enthusiastic and the annoyed — the two groups least like your average customer. Three things lift that rate:

• Ask on the thank-you page. The customer is already there and has nothing else to do. Emailed surveys are answered mostly by people with a complaint.
• Ask one thing at a time. One question per screen, tap to answer. Attribution first, while attention is highest.
• Give something back. Stores that thank people get answers again next time. Stores that take the answer and go quiet watch their response rate fall with every order.

That last point is the reason Twicebow exists. The gift is set aside the moment the customer answers and opens when the package arrives — so the thank-you lands at the best moment of the purchase, and the next survey gets answered too.

What to do with what you learn

The answers are only worth collecting if a decision changes. In practice they tend to change three:

• Where the budget goes. A channel that customers keep naming but dashboards keep ignoring is usually underfunded.
• What the ads say. “Why did you choose us” answers are customer-written headlines. Use their phrasing, not yours.
• What you fix next. “What almost stopped you” names the objection that did stop everyone who left without buying.

See the questions worth asking for the wording of each of these.