How to Improve ROAS: The 2026 Playbook for Paid Social, Google, and Retail Media

Your ROAS slipped, spend feels like it’s leaking, and you’re not sure which lever to pull first. That’s the worst position to be in, because the wrong fix wastes another two weeks of budget. Most guides hand you a list of tips and hope one sticks. This one doesn’t. It’s the same diagnostic order we use inside client accounts, from paid social to Google to retail media. You’ll learn how to calculate the ROAS you actually need, how to find what’s dragging performance down, and how to fix it in a sequence that protects spend while you work. We’ve used this exact process to take brands from unprofitable to 8x and beyond. Let’s get into it.

Quick answer: to improve ROAS, fix it in order. Validate tracking first, cut wasted spend and audience overlap, refresh creative, tighten the landing page and offer, then scale carefully. On Shopping, add review stars, promotions, and a clean feed. Measure account-level ROAS, not the campaign card. The list below is the full version.

What is a good ROAS, and how do you calculate breakeven ROAS?

A “good” ROAS isn’t a single number. Instead, it depends on your margin. For example, a brand with 70% margin can stay profitable at a much lower ROAS than a brand running on 25%.

The number that matters is your breakeven ROAS, and the math is simple:

Breakeven ROAS = 1 / gross margin

If your gross margin is 40%, your breakeven ROAS is 1 / 0.40, which is 2.5. Anything above 2.5 makes money before overheads. Anything below it loses money, even if the platform dashboard looks busy.

Once you know your breakeven number, you know whether you have a profitability problem or a scaling problem. In short, that single calculation changes the whole conversation. Run your own margins through our break-even ROAS calculator if you want the exact figure for your store.

ROAS benchmarks by channel in 2026

Benchmarks are only a starting point, and they swing hard by channel. Here’s where returns typically land in 2026. Still, treat these as directional, then judge yourself against your own breakeven, not the average.

Those are third-party averages. For a first-party view, our own performance benchmark report breaks down real results across 44 client engagements, with a median documented ROAS of 7.35x and a range from 5.12x to 30x. Read it as a ceiling rather than a promise, since it aggregates published wins, but it shows the shape of what a properly run account can do.

Why is your ROAS dropping?

The five classic reasons ROAS falls

Most ROAS drops trace back to a short list of causes, and they rarely arrive alone. Five are the classics:

1. Broken or partial tracking. iOS opt-outs, a misfiring pixel, or missing Conversions API data means the platform optimizes against noise.
2. Audience overlap. Two or three ad sets bidding against each other pushes up your own costs.
3. Creative fatigue. Frequency climbs, click-through rate falls, and cost per acquisition creeps up week over week.
4. Landing page friction. Traffic is fine, but slow pages, weak offers, and clunky checkouts kill the conversion.
5. Margin erosion under the surface. Discounts, rising shipping costs, and refunds quietly eat the profit while ROAS still reads okay.

Four newer causes hitting Shopping and Performance Max in 2026

In 2026, four newer causes hit Shopping and Performance Max accounts especially hard, and most audits still miss them:

6. No review stars on your Shopping ads. If a competitor’s listing shows 4.6 stars and 800 reviews and yours shows nothing, they win the click before price even matters. Missing stars quietly caps your click-through rate and your ROAS.
7. No promotion annotations. Competitors running merchant promotions get a “Special offer” tag, and increasingly get exclusive deals surfaced inside AI results. A plain listing sitting next to a “20% off” listing loses.
8. A feed that AI can’t read. Google now serves products across AI Mode and Gemini, and it builds those answers from your Merchant Center feed. Thin titles and fluffy descriptions get skipped, so you lose free and paid visibility at the same time.
9. AI Max spending on the wrong traffic. AI Max and broad match expansion can pull in loose queries that don’t convert. If you measure it only at the campaign level, it looks fine while account-level ROAS slips.

Fixing the wrong one first is the most common and most expensive mistake. In other words, the order you work in matters as much as the fixes themselves.

How to fix low ROAS on paid social (in the right order)

Work top to bottom. Above all, don’t skip ahead to creative because it’s the fun part. If tracking is broken, better creative just feeds bad data faster.

Step 1, validate tracking before you touch anything

A low ROAS only means something if the revenue behind it is real. Before you change a single ad, confirm the basics:

• Browser pixel and Conversions API are both firing, with events deduplicated by event ID.
• Purchase is set as your top priority event so opted-out iOS conversions still report.
• Event match quality is strong, using hashed email and phone data where you can.
• Your attribution window is consistent across every channel and your finance numbers.

Half the “low ROAS” accounts we audit don’t have a performance problem. Instead, they have a measurement problem. So clean the data first.

Step 2, kill audience overlap and fix the budget split

Once the data is trustworthy, look at where the money goes. Overlapping audiences make you bid against yourself. Therefore, consolidate ad sets, cut the overlap, and check that budget follows performance, not habit. In addition, a steady weekly reallocation of 10% to 15% from your weakest campaigns into your strongest ones compounds fast without shocking the algorithm.

Step 3, test creative like a system, not a guess

Creative is the biggest lever on paid social, and it’s the one most brands treat casually. As a result, accounts that test fewer than five new creatives a month tend to watch cost per acquisition climb. Build a testing cadence instead:

• Ship a small batch of new angles every week, not one big swing every quarter.
• Set a frequency alert at 4.0 so fatigue gets caught before ROAS slips, not after.
• Lead with the hook. On muted autoplay, bold on-screen text in the first three seconds decides whether anyone stays.
• When click-through rate on a prospecting ad sits below 1%, the problem is usually the offer or the hook, not the audience.

Step 4, fix the landing page and the offer

Ad clicks are wasted if the page doesn’t close. First, check page speed, mobile checkout, and how fast your value and price register. This is where conversion rate optimization does more for ROAS than any bid change, because you’re lifting the return on traffic you’re already paying for.

A worked example, diagnosing a 0.53 ROAS account

Say an account reads 0.53 ROAS. On a 40% margin, breakeven is 2.5, so this account is deep underwater. Here’s the order we’d run it:

• Tracking: the pixel fires but the Conversions API is off, and iOS purchases aren’t reporting. Real revenue is understated. Fix the API, and reported ROAS often jumps before anything else changes.
• Waste: a broad campaign is eating 60% of spend with no winning products surfaced. Segment by margin and demand.
• Creative: one tired video carries the whole account at frequency 6. Refresh the angles.
• Offer: the landing page buries the price and the guarantee. Surface both.

In other words, four ordered fixes, not forty random tweaks. That’s how a 0.53 becomes profitable.

Step 5, scale without resetting the learning phase

Once the account is profitable, scaling is its own skill. Push too fast and you throw campaigns back into the learning phase and tank the ROAS you just fixed. Do it in two moves:

• Vertical scaling first. Raise budget on proven campaigns in steps of 20% to 30%, not overnight jumps.
• Horizontal scaling next. Once vertical scaling slows, launch new campaigns for fresh audiences so you expand reach without re-entering learning on your winners.

On bidding, start new campaigns on maximize conversions or lowest cost through the first week or two of learning, then move to a target ROAS or cost cap once you have real conversion data to set the target from.

How to improve ROAS in Google Ads and Performance Max

Search and Performance Max reward structure and clean signal. The biggest wins usually come from the feed and the campaign build, not the bid. Clean product titles, segment by margin and demand, and give Performance Max the guardrails it needs instead of letting it spend blind.

Before you touch bids, run a waste pass. In mature accounts, removing bad spend lifts ROAS faster than chasing more volume. Pull these reports and cut what spends without converting:

• Search terms: add negatives for irrelevant and low-intent queries.
• Placements: exclude junk apps and sites draining Display and PMax budget.
• Geographies: trim regions that click but never buy.
• Devices: adjust for the device that converts worst for you.
• Time of day: pull budget from dayparts that burn spend with no return.

We break the full method down in our guide to Performance Max asset group structure for ecommerce. On the Shopping side, a clean, intent-matched feed lifts click-through and return at the same time, which is often the fastest ROAS gain available in a Google account.

Why your Shopping ROAS is losing to competitors (reviews, offers, and feed)

Paid social fixes solve one half of the problem. However, if you sell on Google Shopping or Performance Max, the other half is the listing itself. This is where a lot of accounts quietly bleed ROAS, because the ad settings look fine and the feed is the real weak link. In the end, three things decide who wins the click, and most brands only control one of them.

Your competitors have review stars and you don’t

Open a Shopping results page for one of your products. If a rival listing shows 4.6 stars and 900 reviews and yours shows only a price, the shopper’s eye lands on the rival before they read a word. Star ratings lift click-through and conversion rate, and they now feed AI shopping answers too, where Google shows pricing, availability, and reviews side by side. Fewer reviews doesn’t just cost you clicks anymore. It makes the AI less likely to surface you at all.

Here’s the part most people miss. Review stars on Shopping ads are not automatic. Instead, you have to switch them on.

• Enroll in the Product Ratings program inside Google Merchant Center.
• Reach at least 50 reviews across all products. Each product needs roughly 3 reviews before its own stars show.
• Submit a product review feed at least once a month, or connect a Google-approved reviews partner (Trustpilot, Yotpo, Feefo, and others) that submits for you.
• Give every product a GTIN. Google matches reviews to products by GTIN first. Without it, many reviews never attach.

After you cross the threshold, expect up to two weeks of onboarding and 7 to 10 days for new reviews to sync. So the sooner you collect and submit, the sooner stars appear. If a competitor shows 800 reviews and you show 40, closing that gap is one of the highest-return moves you have, and it costs zero ad budget.

Their offer promotion extensions are beating your plain listings

Next to review stars, promotions are the other visible edge. A listing with a “Special offer” tag pulls more clicks than the identical product with no tag. In 2026 this matters more than ever, because Google’s Direct Offers now surface promotions inside AI Mode and the Gemini app at the moment someone shows buying intent, and the system can bundle several of your offers and pick the best one per shopper.

If you’re not running merchant promotions, you’re invisible in that layer while competitors are not. To fix it:

• Set up merchant promotions in Google Merchant Center so eligible products earn the “Special offer” annotation.
• Upload more than one promotion type (percentage off, free shipping, coupon) and let the system choose per shopper.
• Keep promotion dates and price data accurate, because time-sensitive offers need frequent feed updates to stay live.

One caution worth knowing. Google sometimes shows a promotion annotation instead of review stars when both compete for the same space, so test and watch which drives more clicks for your listings rather than assuming both always show together.

Show a sale price cut-off, and give first-time buyers a reason to choose you

Two more offer levers move ROAS, and most stores ignore both.

The first is the sale price cut-off. When you add a sale_price alongside your regular price in the feed, Google shows the original price with a line through it, the lower price in green, and a “Sale” badge. That strikethrough does real work, because shoppers see the saving at a glance and click. A few rules earn it:

• The discount has to be more than 5% and less than 90% of the original price.
• The original price needs to have been the real price for at least 30 days in the last 180, so you can’t fake a “was” price.
• Show both prices on the landing page, and add a sale_price_effective_date to put a clock on the offer.
• Drop a price below its 60-day average and Google can also add a “Price drop” badge automatically, which is free urgency you don’t have to design.

The second lever is first-time customer offers. A first-order discount like “15% off your first order” pulls in exactly the buyers you want, the new ones. Set it up as a merchant promotion and exclude existing customers, so you’re not handing money to people who would have bought anyway. Then take it further with the New Customer Acquisition goal in Performance Max, on both Google and Microsoft. It lets you bid higher for net-new customers, or chase them exclusively, and assign a higher value to a first purchase than to a repeat one. That’s how you stop paying prospecting prices for customers you already had.

Your product feed is not built for AI shopping

This is the deepest issue and the one with the longest payback. Google now surfaces products across the Shopping tab, AI Mode, and Gemini, and it builds those conversational answers from your Merchant Center feed. Free listings can appear there with no bid, based purely on data quality. Paid coverage through Performance Max and AI Max for Shopping runs on the same feed. One dataset, many surfaces.

The rule for 2026 is simple. Feed quality is now campaign performance. In practice that means:

• Write titles and descriptions as clear, factual product data, not marketing fluff. Gemini paraphrases your description into the ad explainer it shows shoppers, so vague copy produces vague, low-converting ads.
• Fill the newer Merchant Center attributes built for conversational search, like answers to common product questions, compatible accessories, and substitutes. These help you match natural-language queries competitors skip.
• Update price and inventory often. For promotions, bundles, and AI Max Shopping, sub-hourly price and stock updates are becoming the baseline. Stale data gets your products dropped from time-sensitive placements.

A clean, structured, AI-ready feed lifts you across every surface at once, paid and free. That’s the compounding advantage most accounts are leaving on the table.

The new AI metrics you need to watch in 2026

The old dashboard view of ROAS hides where performance is actually won or lost now. As AI Max and Performance Max take over more of the delivery, these are the numbers that tell you the truth:

• Account-level incrementality, not campaign ROAS. AI Max and broad match expansion often report a healthy campaign ROAS while pulling traffic that was already yours. Independent 2026 testing found most advertisers saw neutral or negative results at the account level even when single campaigns looked strong. Judge AI features on whether total account revenue and ROAS rose, not on the campaign card.
• The Final URL Expansion landing pages report. AI Max can send clicks to pages you didn’t choose. Check the landing pages report to see where it’s routing traffic, and exclude the pages that convert poorly.
• Search terms in AI Max. The search terms and asset reports show which expanded queries you’re now paying for. Add negatives once you have about two weeks of data, not before, so the system can learn first.
• Asset-level performance in Performance Max. PMax now reports impressions, clicks, click-through rate, and conversion contribution for each headline, image, and video, plus which asset combinations win together. Cut the weak assets and feed the system more of what converts.

Watch these and you’ll catch an AI-driven ROAS drop weeks before it shows up in a monthly report.

How to improve ROAS in commerce media and retail media campaigns

Retail media is where a lot of budget is moving, and where almost nobody has written a real ROAS playbook. On Amazon Ads, Walmart Connect, and other retail networks, the rules shift. You’re bidding at the point of purchase, so intent is high, but so is competition on price.

A few things move ROAS most on retail media:

• Protect your price. Unauthorized resellers undercutting you erodes conversion and drags return. If your Shopping or retail media ROAS is soft, price erosion is underdiagnosed. Our guide on MAP pricing vs MSRP covers how to catch it.
• Win the product page first. On retail media, the listing is the landing page. Weak images and thin copy cap your return no matter how good the bid is.
• Separate branded from non-branded. Branded retail media terms look like heroes and quietly claim credit. Split them so you can read true incremental return.

For D2C brands moving into marketplaces, the channel mix matters as much as the bids. We cover that shift in the rise of D2C brands.

How to lift ROAS across Meta, Google, and Bing (cross-channel targeting)

Running each channel in a silo is where a lot of ROAS quietly leaks. The channels don’t know about each other, so they claim the same conversions and chase the same people. Fixing that overlap is often a bigger win than any single campaign tweak.

Start with the double-counting trap. Someone sees your Meta ad, visits, leaves, then searches your brand later. Performance Max grabs that conversion and reports a strong ROAS. Meta claims it too. Now two platforms take credit for one sale, and both look better than they are. To cut through it:

• On Meta Advantage+ Shopping, set the existing customer budget cap and upload your customer list as an exclusion, refreshed weekly. Without the cap, Meta drifts toward existing buyers because they convert easily, which flatters platform ROAS while hiding weak new-customer acquisition.
• On Performance Max, exclude your first-party customer lists when the goal is growth, so it stops harvesting conversions your other channels already earned.
• Judge every channel on incrementality-adjusted ROAS, not the number each platform reports about itself.

Then there’s Bing, which most brands skip and shouldn’t. Microsoft Advertising reaches Bing search, the Microsoft Audience Network across MSN and Outlook, the Edge browser, and LinkedIn profile targeting. The audience skews older and more professional, and CPCs are often lower than Google, so the same budget can return more. You don’t have to rebuild anything. Import your Google campaigns, including Performance Max with New Customer Acquisition goals, into Microsoft, then adjust bids and negatives for the different audience instead of mirroring blindly. This is one of the fastest incremental ROAS gains we add for ecommerce clients, because it captures cheaper conversions that Google-only advertisers never see.

ROAS vs profit: why a rising ROAS can still lose money

Here’s the trap. ROAS can climb while your bank balance shrinks. It happens when discounts deepen, shipping costs rise, or refund rates creep up underneath a healthy looking number. ROAS measures revenue against ad spend. But it says nothing about margin.

That’s why we push clients toward POAS, profit on ad spend. POAS folds in cost of goods, shipping, and returns, so you’re optimizing for money kept, not revenue booked. If you only watch ROAS, you can scale yourself straight into a loss. Watch profit, and the decisions get honest.

How we took a skincare brand to 8.2x ROAS in 38 days

Skincare is one of the hardest categories to run paid ads in. High CPCs, strict health and beauty ad rules, and constant disapprovals. Drought Secret, a regulated eczema skincare brand, came to us with inconsistent returns and an ad account that couldn’t scale.

We rebuilt it in the same order you just read. Tracking first, then waste, then creative, then offer. We segmented audiences by eczema trigger, wrote pain-point-led messaging, and cleaned the Shopping feed around sensitive-skin intent. The result was 8.2x ROAS in 38 days, and it kept climbing past 9x as we scaled. You can read the full breakdown in our eczema skincare case study, and see how we approach regulated beauty accounts on our skincare marketing page.

The point isn’t the number. Rather, it’s that a disciplined, ordered process beats guesswork in even the toughest category.

Want us to audit your paid accounts?

If your ROAS is stuck and you’d rather not spend another month guessing, we’ll find the leak for you. Our team runs full paid account audits across Google, Meta, Bing, and retail media as part of our ecommerce PPC management and PPC agency services. Request your free account audit and we’ll pull the data before the call.

Frequently asked questions

What is a good ROAS? It depends on your margin. Calculate breakeven ROAS as 1 divided by your gross margin, then aim above it. Many ecommerce brands target around 4x.

What is the difference between ROAS and ROI? ROAS measures revenue against ad spend only. ROI measures profit against total cost. ROI is the stricter, more honest business metric.

How do you calculate breakeven ROAS? Divide 1 by your gross margin. A 40% margin gives a breakeven ROAS of 2.5. Above that you profit, below it you lose money.

Why is my ROAS suddenly dropping? Usually broken tracking, audience overlap, creative fatigue, landing page friction, or hidden margin loss from discounts and refunds. Check them in that order.

What is a good ROAS for Facebook ads? There’s no fixed number. Compare it to your breakeven ROAS. A 3x on a high-margin product can beat a 5x on a thin-margin one.

Why don’t my Google Shopping ads show star ratings? You need the Product Ratings program enabled in Merchant Center, at least 50 reviews across products, GTINs on items, and a review feed submitted monthly.

What is AI Max in Google Ads? It’s Google’s AI layer for Search and Shopping campaigns using search themes, URL expansion, and AI copy. Judge it on account-level ROAS, not campaign ROAS.

How do I show a sale price on Google Shopping? Add the sale_price attribute alongside your regular price. The discount must be between 5% and 90%, with both prices shown on your product page.

Is Bing worth running alongside Google Ads? Often yes. Microsoft Advertising has lower CPCs and a different audience. Import your Google campaigns, then adjust bids and negatives for it.

Final thoughts

A stuck ROAS is almost never one big problem. Instead, it’s a few small ones in the wrong order. Start with tracking, because a wrong number can’t guide a right decision. Then cut the waste, tighten the budget split, and treat creative and offer as the real growth levers they are. Above all, keep one eye on profit, not just the return the dashboard shows you. Do that consistently and profitable scaling stops feeling like luck. If you’d rather hand the diagnosis to a team that runs this process every day, we’re one message away.