Retail Media Benchmarks 2026: Amazon vs Walmart vs Instacart ROAS, CPC and ACoS

Retail media benchmarks in 2026 cluster tightly by platform: Amazon averages 29% ACoS and 2.5x-6.0x ROAS depending on category, Walmart Connect runs roughly 55% lower CPCs than Amazon on matched CPG terms, and Instacart delivers 4.8x-6.7x ROAS on sponsored products. Retail media as a whole has held about 6.1x ROAS for five straight quarters.

Those are the numbers. The harder question is what they are worth, because almost every figure a retail media network publishes about itself is measured by that network, on its own data, using last-click logic. We will get to that.

US retail media ad spend is forecast at $71.09 billion in 2026, up roughly 18% year over year (eMarketer, H1 2026 forecast). Amazon holds close to 69% of it. The top three networks - Amazon, Walmart and Target - control about 74% of US retail media spend between them, and eMarketer projects Amazon and Walmart alone will absorb 89% of the net-new money entering the channel this year. For a brand deciding where to start, that concentration is the single most useful fact in this post: there are 150+ retail media networks, and three of them matter.

NAMEX 2026 Retail Media Benchmark Table

Figures below are pulled from platform reporting and third-party benchmark sets published through Q1-Q3 2026. Where sources disagree we have given the range rather than pick a favourite.

Amazon: the default, and the most expensive place to be average

Amazon ACoS averages around 29% across categories in 2026. Competitive categories - supplements, electronics, anything with a thousand near-identical listings - run 40-50%. Defensive or genuinely differentiated catalogues can hold 15-20%.

ACoS is only meaningful against your break-even. If your contribution margin is 35%, a 32% ACoS is a rounding error away from losing money on every incremental unit, and no amount of bid tuning fixes that. Set target ACoS from margin first, then work backwards into keyword strategy.

CTR on Sponsored Products sits in the 0.5%-1.2% band depending on category. Above 1.2% is genuinely good. Below 0.3% is almost always a relevance or imagery problem rather than a bidding problem - you are winning impressions on queries your product does not answer.

Walmart Connect: the arbitrage is real but closing

Walmart Connect CPCs run about 55% lower than Amazon on equivalent category searches. CPG-specific CPCs land $0.40-$0.80 against Amazon's $1.30-$1.50 for comparable products. Walmart's conversion rate on matched categories is also higher - roughly 17% versus Amazon's 10.3% - because the shopper base skews grocery and household replenishment, where intent is narrow and repeat.

That combination is why Walmart Connect grew ad revenue 46% year over year to roughly $6.4B. It is also why the gap is closing: every quarter of 43-46% growth is another cohort of advertisers bidding the CPCs up. The arbitrage window on Walmart is a 2026 phenomenon, not a permanent structural advantage.

Expect 1.5x-2.5x ROAS through the first 60-90 days while the platform collects conversion data, then 4x-5x at maturity. Judging Walmart on week-three numbers is the most common mistake we see.

Instacart: the highest ROAS, on the smallest addressable base

Instacart posts the strongest headline efficiency of the three: 4.8x-6.7x ROAS, conversion rates of 18-25%, and CPCs of just $0.25-$0.85. CTR runs 5.2-8.1%, an order of magnitude above Amazon, because the ad unit sits inside an active shopping list rather than alongside a search result.

The constraint is reach, not efficiency. Instacart is a grocery and consumables channel. If your product is not in a basket someone is already assembling, the numbers above do not apply to you. Where it does fit, it is usually the highest-return line in the plan and the first place we would move budget.

The part the benchmarks do not tell you

Every ROAS figure above is platform-reported, and platform attribution is built to flatter the platform. Closed data environments, self-reported conversions and last-click logic all push the same direction. Published 2026 analyses put the gap between last-click ROAS and incremental ROAS at 30-60% - meaning a reported 6x is plausibly a 2.5x-4x in reality.

It cuts the other way too. Criteo measured organic clicks rising 56% in AMER within two weeks of retail media activation, which platform reporting does not capture at all. So the reported number is simultaneously too generous about what the ad caused and too stingy about what the campaign did.

The only way out is a holdout. Suppress the channel in a matched set of geos or audiences for 2-4 weeks, run everywhere else, and measure the gap. That number is your planning input. The dashboard number is a bidding signal. Treat them as different things - we go deeper on this in our comparison of last-click, data-driven and MMM attribution.

If your retail media reporting has never been checked against a holdout, that is the highest-value hour you can spend this quarter. We will do it with you on a free 30-minute media audit - bring your last 90 days of platform reporting and we will show you where the incrementality gap most likely sits.

How to sequence your first 90 days

Start where you already have distribution and velocity. Sponsored Products on the one or two networks where your products actually sell, nothing else. Display, video and off-site retail media are efficient only once you have on-site conversion data to target from.

Set target ACoS from contribution margin before you set a single bid. Run 60-90 days without judging ROAS, because that is the data-collection period on every platform here. Then run your first holdout, and only then decide whether to expand networks or deepen the two you have.

Retail media pricing does not sit in isolation either - CPCs here compete for the same budget as paid search and CTV, and the relative cost per outcome moves every quarter. Our 2026 digital ad spend by channel benchmarks put the retail media number in context against the rest of the mix, and our 2026 CTV CPM benchmarks cover the upper-funnel channel most retail media brands under-invest in.

We plan and buy retail media alongside programmatic, CTV and paid search across 13 DSPs - see how the channels connect on our programmatic advertising and paid search pages.

Frequently asked questions

What is a good ROAS for retail media in 2026?
Cross-platform, retail media has held around 6.1x for five consecutive quarters, but that is a last-click figure. Platform-specific ranges run 2.5x-6.0x on Amazon, 4x-5x on mature Walmart Connect campaigns and 4.8x-6.7x on Instacart. Judge against your own break-even, not the average.

Is Walmart Connect actually cheaper than Amazon?
Yes, currently. CPCs run roughly 55% lower on matched CPG categories, and conversion rates are higher at about 17% versus 10.3%. But Walmart's ad revenue is growing 46% year over year, which means competitive pressure - and CPCs - are rising fast.

How much should I budget to start?
Enough to get through 60-90 days of data collection on one network without pausing. Ad platforms optimise on conversion volume, and a budget that stops before the platform has learned anything buys you nothing you can use.

Why is my reported ROAS so much higher than my actual revenue lift?
Because retail media attribution is last-click inside a closed environment. Published 2026 analyses put incremental ROAS 30-60% below reported ROAS. A geo holdout test is the only reliable way to size the gap in your own account.

Should I run retail media if I only sell direct-to-consumer?
Generally no, not as a primary channel. These networks monetise their own checkout. Without distribution on the retailer, you are paying retail media CPCs for traffic that has to leave the platform to buy - which is what programmatic and paid social already do more cheaply.

Where to go from here

If you want a second opinion on which network fits your margin profile and catalogue, book a 30-minute intro call and we will go through it directly. If you would rather we look at the numbers first, request a free media audit and we will come back with where your retail media spend is and is not working.

Sources

Industry figures in this article are drawn from the organisations below. Campaign-level benchmarks reflect North American Media Experts client data.

• Walmart Connect
• Instacart