Should a Shopify Brand Send DSP Traffic to Amazon or to Its Own Site?

The four variables that actually decide this, not intuition

Margin after fees: Amazon's referral fee (typically in the 8-15% range depending on category, and worth checking your specific category's current rate rather than assuming) comes off every Amazon-destination sale; a Shopify sale carries payment processing fees, usually smaller, but no marketplace cut. Customer ownership: an Amazon sale gives you very little first-party data — no email, limited retargeting ability outside Amazon's own ecosystem; a Shopify sale gives you the full customer relationship to build on. Conversion rate: Amazon's native one-click purchase flow and stored payment details often convert cold traffic better than a Shopify checkout a new visitor has never used before. Measurement rigor: Amazon-destination traffic gets native, structured attribution reporting; Shopify-destination traffic needs your own analytics built and maintained.

A worked comparison at the same spend level

Take $10,000 of DSP spend split two ways to compare directly. Sent to Amazon: assume a 3.1% conversion rate on cold prospecting traffic (typical for Amazon's high-trust checkout with new visitors) at a $52 average order value, generating roughly 597 orders for $31,044 in gross Amazon sales — after a 12% referral fee, $27,319 net. Sent to Shopify: assume a 1.8% conversion rate (typical for a new-visitor checkout without stored payment details) at the same $52 AOV, generating roughly 346 orders for $17,992 in gross Shopify sales — after roughly 3% payment processing, $17,452 net, but with the full customer email and purchase history captured for future remarketing that Amazon never provides.

On pure net revenue from this one campaign, Amazon wins clearly. On long-term value — the ability to email, retarget and build lifetime value from the 346 Shopify customers you now own outright — the comparison changes, and the honest answer depends on how much you weight a smaller number of owned customers against a larger number of one-time Amazon transactions.

What the returning-customer picture looks like a year later

Extend the worked example out: if 20% of the 346 Shopify customers from that one campaign make a second purchase within 12 months at the same $52 AOV, that's roughly 69 additional orders — $3,588 in additional revenue that never shows up in the original campaign's ROAS at all, because it happened outside the attribution window entirely. Amazon's version of that same repeat behaviour largely accrues to Amazon's own platform metrics, not to anything the brand can see, measure, or take credit for in its own reporting. That asymmetry is the real argument for weighting toward owned-site traffic in categories where repeat purchase is genuinely likely.

Why conversion rate alone shouldn't decide it

The conversion-rate gap in the worked example is real and typical, but it compounds differently depending on repeat-purchase behaviour. A category with a strong subscription or repeat-purchase pattern makes the Shopify math better over time than a single-transaction comparison shows, because each Shopify customer generates additional lifetime revenue an Amazon customer largely doesn't — Amazon retains that repeat relationship on its own platform, not yours. A single-purchase, low-repeat category makes Amazon's higher one-time conversion rate the more straightforwardly attractive option, since there's less lifetime value to capture either way.

The common mistake, including ours

The mistake is deciding this once, based on a single campaign's numbers, and treating it as a permanent allocation rather than an ongoing test. We've seen a brand lock in a 100% Amazon-destination DSP strategy after one early test showed a stronger single-campaign ROAS, without ever testing whether the lower-converting but customer-owning Shopify path would have paid back better over a 12-month lifetime-value window. The single-campaign comparison genuinely favoured Amazon; the full lifetime-value comparison, which nobody ran, might well have favoured a more balanced split.

What the honest answer looks like when neither side clearly wins

For a lot of brands, the worked comparison above won't produce a clean winner — the numbers will land close enough that the decision genuinely comes down to strategic priorities rather than arithmetic. A brand prioritising near-term cash flow and simplicity may reasonably lean Amazon-heavy despite the lost customer-ownership value. A brand building toward a long-term direct-to-consumer brand with its own retention engine may reasonably accept a lower near-term net revenue on the Shopify side for the compounding value of an owned list. Neither is the objectively correct answer — the framework here is for making the trade-off deliberately, not for producing a single number that settles it.

A practical way to test both without over-committing

Run a genuine, even split — say 50/50 — for at least one full quarter, tracking net revenue after fees for the Amazon-destination portion and both net revenue and repeat-purchase rate for the Shopify-destination portion over a longer window than the campaign itself, since Shopify's real value is partly in the second and third purchase, not just the first. Let that data, not a single-campaign snapshot, inform the permanent split — and revisit the split periodically as fee structures, conversion rates or your own repeat-purchase economics change.