Amazon Advertising Revenue: What Each Quarter Means for Sellers

Updated for Q3 2026, reported October 29, 2026. Next update: early February 2027, when Amazon reports Q4.

Amazon reports these numbers every quarter, and each report says something about the auction you bid into. This page tracks them, explains what they can and can't tell you about your own costs, and shows how to tell whether a change in your CPC or TACoS is the market or your account.

Quick Answer

Amazon's advertising revenue was $19.6 billion in Q3 2026, up 12% year over year excluding currency effects, down from 26% in Q2. Most of that slowdown is calendar: Prime Day moved from Q3 into Q2 this year. Amazon expects ad growth back in the high teens in Q4. Measured against an estimate of total marketplace sales, ad spend is still rising as a share of what shoppers buy, but more slowly than earlier this year. For sellers, that points to a pricier auction this November than last, so set peak bids from your margin, not from October CPCs.

Latest Quarter: What Amazon Reported for Q3 2026

Advertising services revenue was $19,637 million, up 11% as reported in US dollars and 12% excluding foreign exchange (ex-FX), which strips out the effect of currency swings on sales made in euros, pounds, yen and other currencies. We use ex-FX throughout because Amazon uses it in its own guidance (Amazon Q3 2026 earnings release).

Why Prime Day makes Q3 look softer than the auction was

Prime Day is Amazon's largest ad-spend event outside Q4. Sellers raise budgets and bids for it, so whichever quarter contains Prime Day gets a large one-time lift in ad revenue. In 2025, Prime Day ran entirely in Q3. In 2026, it ran in Q2 for most of Amazon's largest countries, including the US. So Q3 2026 is being compared with a Q3 2025 that included Prime Day spend, and Q2 2026 got the reverse effect.

On the earnings call, CEO Andy Jassy put the impact of the shift at about 850 basis points (8.5 percentage points) on Q3's year-over-year ad growth. Amazon named two smaller timing factors as well: there were no NBA games on Prime Video in Q3, and Q3 2026 laps a change Amazon made to its search engine marketing in Q3 2025 that sent more traffic to Amazon's own search pages.

Reading Q2 and Q3 together removes most of the distortion. Combined, they grew about 18% over the same six months of 2025. That is the cleaner read of where ad demand sits right now.

What Amazon expects for Q4 2026

Amazon expects year-over-year ad revenue growth in the high teens in Q4. Applied to Q4 2025's $21.3 billion, 17% to 19% growth implies roughly $24.9 billion to $25.4 billion. That range is our arithmetic, not an Amazon figure. NBA on Prime Video returns for its second season in Q4, so part of that growth comes from streaming inventory that Sponsored Products advertisers never bid against.

A Q4 plan built on October CPCs is built on the wrong baseline

October usually runs at lower CPCs than late November, because many advertisers hold budget for Black Friday and Cyber Monday. Third-party benchmarks also show Amazon click prices flat to down year over year in Q3 2026; Pacvue's Q3 2026 benchmark, for example, shows Sponsored Products CPC down 3% across its client base (Pacvue). If you set November bid targets off October or Q3 CPCs, you are pricing peak against a quieter market. That can leave campaigns underbid just as traffic peaks, or lead an automation to read a market-wide price rise as falling efficiency and cut spend on campaigns that are working.

Amazon Advertising Revenue by Quarter

Each report adds a row. The last column compares ad growth with growth in paid units, the number of items sold on Amazon. When ad revenue grows faster than units, Amazon is collecting more ad revenue per item sold. That is not the same as CPC, because ad revenue also includes Prime Video, DSP and other inventory outside the search auction, but it is the closest signal Amazon's own filings give.

Source: Amazon quarterly earnings releases (SEC Form 8-K, Exhibit 99.1). The gap column is our calculation. Q3 2026's 0 points is mostly Prime Day timing, not a change in direction.

What Amazon's Filings Actually Disclose

Amazon's supplemental financial tables split net sales into seven lines. Three matter to sellers, and two of them are easy to misread.

• Online stores is Amazon's own first-party retail. Amazon records it gross, meaning the full price the shopper paid, because Amazon owns the inventory on these sales. It works as a rough proxy for first-party sales volume, though it also includes digital purchases like Kindle books.
• Third-party seller services is the opposite case. On a marketplace sale Amazon only books its own cut: referral fees, FBA fulfillment, shipping and storage. The $46.5 billion Amazon reported for Q3 2026 is what it collected from sellers, not what shoppers spent with them. Third-party sellers accounted for 61% of paid units in the quarter.
• Advertising services covers sponsored ads, display, DSP inventory on other sites, Twitch, Sponsored TV and Prime Video ads, including live sports. Amazon does not break it out by format.

Amazon does not publish gross merchandise value (GMV), the total value of everything sold. So any ratio of ad spend to marketplace sales has to be estimated.

Ad Spend as a Share of Sales: A Market-Wide TACoS Estimate

TACoS, or total advertising cost of sales, is ad spend divided by total sales, not just ad-attributed sales. Most sellers track it for their own account. Amazon's filings let you estimate the same ratio for the whole marketplace, which gives you a yardstick for your own drift.

The obvious shortcut is wrong. Dividing ad revenue by Online stores plus third-party seller services gives 17.1% for Q3 2026, but the denominator counts seller fees instead of seller sales, so it misses most of the marketplace. The fix is to convert seller fees back into an estimate of seller sales by dividing by Amazon's take rate, the share of each sale it keeps in fees:

Estimated GMV = Online stores + (Third-party seller services ÷ take rate)

The take rate is an assumption. We use a 30% to 40% range to cover referral, fulfillment and storage fees together; your own settlement reports give your actual blended rate. Here is how the range plays out for Q3 2026:

The level carries about a point of uncertainty either way, so treat it as a range. The trend is more reliable, because the take-rate assumption shifts every quarter by roughly the same proportion. At a 35% take rate:

Source: Amazon Q3 2026 earnings release, supplemental tables; implied TACoS is our model.

What moved this quarter

Read directionally, the ratio is still rising. But Q2 and Q3 tell different stories on their own, because Prime Day moved spend and sales between them. Combining the two quarters gives the cleaner read: implied market TACoS rose from 9.14% to 9.73%, about 59 basis points year over year. That is still an increase, but slower than the roughly 80 basis points Q2 alone suggested. In Q3 itself, ad revenue grew 12% ex-FX and seller fees grew 10%, close to the same pace, while Amazon's own retail sales grew only 2% as Prime Day shoppers moved to June.

Why the ratio overstates the pressure on sellers

The numerator is not all seller money. Prime Video ads, NFL and NBA inventory, and DSP placements on other sites all sit inside Amazon's advertising line, and most of that inventory barely existed two years ago. Some of the rise in this ratio is Amazon selling new kinds of ads, not existing sellers paying more for the same clicks. Amazon does not break out the split, so we can't say how much.

If your own TACoS has drifted and you want to bring it down without cutting sales, our guide on how to lower TACoS covers the levers.

What Ad Revenue Can and Can't Tell You About Your CPCs

For sponsored ads, Amazon earns revenue when a shopper clicks, so that slice of revenue is roughly clicks multiplied by CPC. Display and video ads are mostly sold on impressions instead. That gives ad revenue three ways to grow:

• More clicks, because more shoppers are searching, which happens every Q4.
• Higher CPCs, because advertisers are bidding more for the same placements.
• Inventory outside the search auction, such as Prime Video and live sports.

Amazon does not say how much of any quarter's growth comes from each. So the revenue line can't tell you what your category's CPC will do. What it does support is a planning assumption: when Amazon guides ad growth up, more budget is likely chasing the same search placements, and that tends to push CPCs up. Treat it as a reason to set your response rules before the auction moves, not as a forecast.

Market or Account: Two Reasons Your CPC Can Rise

Your CPC is set by your bid competing against other advertisers' bids for the same placement. So when CPC rises, the rise can come from one of two places:

• The market: other advertisers raise budgets and bids, and the same click costs more for everyone.
• Your account: your own settings shift spend into more expensive clicks, such as broad match pulling in pricier queries or placement adjustments pushing more spend to top of search.

The two look the same in a CPC report, but they call for opposite responses.

The four steps below separate the two. The same logic works for TACoS: compare your year-over-year change with the market estimate above, and if you moved much more than the market did, look at your own account first.

Step one: compare CPC change across all campaigns

Pull week-over-week CPC for every campaign and rank them by change. If nearly all of them rise together, you are most likely looking at the market. If two or three campaigns carry most of the increase, look at your own account first.

Step two: check what changed in the campaigns that moved most

For those outliers, compare placement share (top of search, product pages, rest of search) and match-type share against the prior four weeks. A shift toward broad match or top-of-search placements will raise CPC even if no competitor changed anything, and you can fix it by tightening targeting before you touch bids.

Step three: work out a CPC ceiling from your margin, then cap your bids to it

For each campaign, work out the most a click can cost and still be profitable: average order value times target ACoS times conversion rate. A product with a $40 average order, a 25% target ACoS and a 10% conversion rate can afford about $1.00 per click.

Amazon has no separate CPC cap. In Sponsored Products, your bid is your maximum CPC, and you often pay less. Two settings raise that maximum: placement adjustments, which multiply your bid by up to 900% for a placement, and the "dynamic bids, up and down" strategy, which can raise your bid by up to 100% for top of search and 50% for other placements. So the ceiling has to live in your bid. Check that your bid, multiplied by your largest placement adjustment and by any dynamic uplift, stays at or under the ceiling. If it doesn't, lower the bid or the adjustment.

The ceiling works in both directions. A campaign whose CPC rose but still sits under its ceiling is still profitable at peak traffic, so don't cut it. A campaign whose maximum possible CPC sits above its ceiling needs its bid or placement adjustment brought down, which keeps a market-wide rise from turning into overpaying.

Step four: repeat steps one and two every week through peak

The cause of a CPC rise can change from one week to the next. Black Friday week may be a market-wide surge, and the week after Cyber Monday may be your own broad match drifting into gift queries. A single check in early November only tells you about early November. This is also the step that tends to slip, because it needs placement and search-term data for every campaign during the busiest weeks of the year.

Qore Runs the Weekly Market-or-Account Check for You

Step four is where Qore fits. Qore, the workflow layer from Trellis, turns a review you do by hand into one that runs on a schedule. Set up as a weekly read, it runs steps one and two across your portfolio and returns a ranked list of campaigns where the CPC rise traces back to your own account. Proposed target and budget changes wait for your approval by default, so an automation does not read a market-wide price rise as an efficiency problem and cut a campaign that is working. Qore is in open beta, and you can sign up here.

Qinetix, our ads automation platform, can execute the changes you approve and keep a change log, so a later review can separate "we were outbid" from "we moved." The limit: Trellis cannot forecast auction prices and has no view into other advertisers' budgets. It measures the move in your own account quickly enough to act on it.

How to Use This Page Each Quarter

Three habits make Amazon's numbers useful rather than noisy:

• Compare year over year, not quarter over quarter. Prime Day, NBA scheduling and other calendar shifts move spend between quarters. When an event moves, combine the affected quarters.
• Quote the trend, not the level. The market TACoS estimate carries a point of uncertainty. Its direction and pace are the reliable part.
• Check your own account against it. When Amazon reports, compare its ad growth and the market TACoS change with what you saw in your own CPCs and TACoS over the same quarter. That gap is what tells you whether the change was the market or you.

Amazon reports Q4 2026 in early February 2027, and we will update this page then.