Cashback marketing is a promotion model in which the customer pays full price and the brand returns part of the money after the purchase — as a bank transfer, card credit or account balance — instead of lowering the price at checkout. The difference sounds cosmetic but is structural: cashback changes when the cost hits your P&L, which buyers you actually pay for, and — most importantly — what happens to the price customers remember.
This guide covers the three mechanics behind merchant cashback, the psychology that makes it convert differently from an equivalent discount, what it really costs compared to coupons, and the situations where a plain discount is still the better tool.
How cashback works for merchants
Nearly every cashback program a brand can run uses one of three mechanics, and they have very different economics.
Portal and affiliate cashback is the model behind cashback sites and browser extensions: the shopper clicks through, the retailer pays an affiliate commission on the tracked sale, and the site passes part of that commission back as cashback — how much reaches the shopper varies by platform. For the merchant it is cost-per-acquisition: you pay only on completed sales, but commissions sit at the upper end of publisher rates because they fund the visible rebate. The standard critique is last-click attribution: because the cashback click is often the final touch before checkout, portals can be credited for orders that other channels actually drove.
Direct brand cashback promotions are claim-based: buy the product, submit a claim, receive the reward. The agencies that run these campaigns openly model partial redemption — public cost calculators treat a 20–40% redemption rate as standard for claim-based cashback, with roughly 85% of submitted claims being valid. Unclaimed rewards ("breakage") are baked into the economics, which is both the appeal and the trap, as covered below.
Card-linked and API-driven cashback removes the claim entirely: the payment network or the brand's own system detects a qualifying transaction and the money is applied automatically. Redemption is effectively 100%, the merchant is billed on actual redemptions, and breakage economics disappear — it behaves like pure performance spend.
Why cashback converts differently from a discount
The mechanism has a name: Richard Thaler's hedonic framing, from his 1985 Marketing Science paper on mental accounting — part of the work behind his 2017 Nobel. The relevant prediction is the "silver lining" principle: a small gain presented separately from a larger loss is valued more than an equal-sized reduction of that loss. Five euros back after paying €50 registers as a win; €5 off a €50 price registers as a slightly smaller loss. Same money, different accounting in the customer's head.
Two honesty notes. This is a framing prediction, not a guaranteed uplift — replication reviews find textbook mental-accounting effects often shrink under scrutiny, so treat it as a hypothesis to test in your category, not a law. And it only works if the gain actually feels like a gain: a rebate that arrives six weeks later as a voucher does not.
The second mechanism is reference price. A discount rewrites what the product "costs" in the customer's memory; run it often enough and full price starts to look like a markup. Cashback leaves the headline price intact — the customer paid full price, and the reward sits in a separate mental account.
What cashback actually costs
A discount costs you on every unit sold, including all the units customers would have bought at full price. Subsidizing that would-have-paid-anyway demand is the silent bulk of most discount budgets.
Claim-based cashback costs you only on redemptions — but breakage is not free money. Gilpatric's 2009 Marketing Science analysis shows a rebate can only profitably exploit non-redemption if the offer is substantially larger than the equivalent discount, because it must compensate customers for the hassle and delay of claiming. A pre-digital-era University of Florida study (2004) found 60% of non-redeemers simply procrastinated or forgot — behavior your mechanic design directly controls.
Automatic cashback costs you on every qualifying sale, like a discount — but it is paid post-purchase, tied to a verified transaction, and measurable per redemption, which makes it budgetable as performance spend rather than blanket margin erosion.
For portal cashback, add the attribution tax: before treating the channel as incremental, audit what share of commissioned orders had another paid touch earlier in the journey.
Cashback and price integrity
In the EU, discounting is also a regulatory question. The Omnibus Directive requires any announced price reduction to display the lowest price charged in the prior 30 days, and the CJEU confirmed in 2024 that the advertised percentage must be calculated against that 30-day-lowest price. Frequent discounting therefore leaves a legal paper trail on your pricing. Cashback is structured differently — a post-purchase reward with the headline price unchanged — but do not assume blanket exemption: whether a given mechanic counts as an announced price reduction depends on how you frame it and on the member state.
In the US, rebate and cashback offers fall under FTC truth-in-advertising standards: material terms must be clear and conspicuous, FTC settlements have required delivery within the stated window (or 30 days if none is stated), and states including California and New York require disclosing the pre-rebate price.
When cashback is the wrong choice
You already discount habitually. If your audience has been trained to wait for sales, layering cashback on top adds cost without restoring price integrity. Fix the discount cadence first.
Your margin can't fund a reward worth noticing. Cashback comes out of the same margin a discount would — and claim-based cashback, per the rebate literature, needs to be larger than the equivalent discount to work. Low-margin, low-basket categories rarely have that room.
Your mechanic creates friction. Receipt uploads, claim portals and six-week payouts recreate the mail-in rebate experience, and the resentment lands on your brand, not on the agency running the campaign. If you can't pay quickly and transparently, don't run the promotion.
You need conversion pressure at checkout, today. A discount removes the price barrier in the moment of decision. A cashback promise is a weaker instrument for a purely price-driven impulse purchase — that is the honest trade for protecting your reference price.
Accrual vs instant payout
The last design choice is when the money moves. Accrual models bank the reward into a points balance or pending ledger and pay out at a threshold or on a schedule — cheaper to float, but the delay is exactly the hassle the rebate literature says erodes the perceived gain. Instant payout sends real money as soon as the transaction is verified, which is what makes the "separate gain" framing actually land.
The infrastructure for instant payout is now commodity. TikJoy's Cashback API, for example, sends cashback to any consumer or creator with a single API call over SEPA or local payment rails, on whatever trigger you define — a verified purchase, a review, or, in the JoyBack model, a social action like sharing a video, with no purchase involved at all.
The decision rule is short: use a discount when you need immediate conversion and your price image can absorb it; use cashback when the headline price is an asset worth protecting, your margin can fund a reward customers actually notice, and you can pay it out fast enough to feel like a gain instead of a chore.