Prime Day Advertising, Explained With Real Numbers

What "Prime Day advertising" actually means

"Prime Day ad" covers four different Amazon products, not one: Sponsored Products (keyword and product targeting on search and product pages), Sponsored Brands (headline search placements with brand video or logo), Sponsored Display (retargeting on and off Amazon), and Amazon DSP (programmatic display and video, bought through a managed seat, running on and off Amazon). None of these is a special Prime Day format. They're the same auctions and the same targeting tools you use every week, dropped into a two-day (sometimes longer) traffic spike where everyone raises budgets on the same keywords and audiences at once.

Amazon's own Prime Day guidance for 2026 reports that advertisers running Sponsored Products, Sponsored Brands, and display together saw sales grow 137% faster than the category average, and that Sponsored Brands plus display alone added a 39% lift over category growth in 2025. Those are Amazon's aggregate figures, not a controlled study, so treat them as directional rather than a promise: more formats working together beats one format alone, which matches what we see across our own advertiser book.

How to size a Prime Day ad budget — a worked example

Budgeting for Prime Day is a math problem before it's a strategy problem. You need a CPM, a CPC, and a CPA to turn a sales target into a spend number. Here's what that looks like on a real book, not a rate card. Across 30 of our managed advertisers in July 2026 — a normal month, not a Prime Day month, which matters for the comparison — the portfolio ran 78.4 million impressions at a $4.00 CPM and a blended $1.42 CPC, converting into 57,137 attributed purchases at a blended $5.49 CPA, for a 6.04x return on ad spend across the whole book, not cherry-picked from the best line item.

Use those as planning inputs, not promises. If your target is 1,000 incremental purchases during the event and your account's historical CPA sits near a $5.49 blended figure — most accounts won't match it exactly, category and competition move it — your working budget is roughly $5,490, before you add the bid inflation Prime Day itself causes. That inflation is the part sellers underestimate: the same keyword that costs $1.42 in June routinely costs more during the event because everyone raises budgets at once. Build the CPA-based number first, then layer on headroom for bid pressure, not the other way around.

One number worth flagging because it gets misquoted constantly: the $0.41 CPC figure that circulates in Prime Day roundups is an online-video number, not a blended account average. It's real, but it describes one placement type inside a much bigger book. Plan a whole account's budget off a video-only CPC and you'll underfund search and product targeting badly.

The mistakes that sink Prime Day campaigns

Almost all of them are pacing mistakes, not targeting mistakes.

• Raising budgets the morning of the event. Start increasing budgets three to five days out, in 20-30% steps, so campaigns have time to relearn before the peak. Wait until the day itself and your best campaigns go "budget limited" during the exact hours that matter.
• Trusting last-click ROAS to judge display and DSP. Sponsored Products will always look better than display in a last-click report, because last click hands all the credit to whichever ad the shopper touched last. That doesn't mean display did nothing — it means last click can't see what it did. We've made this mistake ourselves early on: reading a DSP line item as underperforming against Sponsored Products, when the two were never measuring the same thing.
• Not separating new-to-brand performance from repeat purchases. Prime Day pulls in deal-seekers who've never bought from you. Across our July 2026 book, 20.1% of attributed purchases were new-to-brand — worth tracking on its own, because a campaign that looks average on blended ROAS might be doing the best new-customer acquisition work in your whole account.
• Skipping negative targeting. High-traffic events without a negative product or keyword list waste budget fast on competitor ASINs and mismatched searches — cheap to fix, expensive to ignore.

What to do when Prime Day ad performance looks wrong

If a campaign isn't spending, check budget-limited status first. It's the most common cause during Prime Day and the easiest to fix — raise the daily budget or the campaign simply stops competing during the hours it matters most.

If it's spending but ROAS looks worse than a normal week, check whether you're comparing like for like. Prime Day traffic includes a wave of new-to-brand shoppers who convert at different rates than repeat buyers, and blending the two without separating them will make good campaigns look bad on paper.

If sales are up but you can't tell whether ads caused it — versus the event itself, a coupon, or organic lift — that's an attribution problem, not a strategy problem, and no amount of dashboard-staring fixes it. Last-click attribution was never built to answer "would this sale have happened anyway," and it never could. Holdout tests and matched controls can. That's a heavier lift than most sellers need for one event, but it's the only honest way to know whether display and DSP spend added incremental sales or just took credit for sales that were coming regardless.

Where DSP fits during Prime Day

DSP sits above sponsored ads in the funnel: retargeting people who viewed your Store or product pages before the event, building awareness with shoppers who haven't searched for you yet, and holding attention after Prime Day when search volume drops but intent doesn't disappear overnight. Run well, it works alongside Sponsored Products and Sponsored Brands rather than against them. The mistake, covered above, is measuring it with the same last-click report and concluding it underperforms.

Reconciling DSP and sponsored ads in Amazon Marketing Cloud, rather than two separate dashboards, is the only way to see whether a shopper who saw a DSP ad and later converted through a Sponsored Products click is one sale or two — most advertisers' reporting counts it twice. We run this exact reconciliation for Prime Day campaigns at brands including HexClad, Ridge, and Beardbrand.