Google Ads ROAS benchmarks for ecommerce in 2026 (real numbers by vertical)

Most agencies will not show you real ROAS benchmarks. Because real benchmarks make it obvious when an account is underperforming - and that is uncomfortable for the agency running it.

Here are the real numbers.

These come from ZenoX's portfolio across 200+ ecom accounts and €200M+ in managed ad spend. These are blended account ROAS figures - total revenue attributed to Google Ads divided by total Google Ads spend, including branded traffic, PMax, Standard Shopping, and Search.

They are ranges, not targets. Your position in the range depends on feed quality, tracking accuracy, campaign structure, and vertical competitive density. We will explain what drives each one.

Jewelry

3.5-5x

Fashion

2.8-4x

Home Decor

3-4.5x

Beauty

3.5-5x

Jewelry: 3.5-5x blended ROAS

Jewelry is the highest-ROAS vertical we run consistently. Three reasons.

High AOV. Average order values of €150-500+ mean the conversion value is large relative to cost per click. A click that costs €2 that converts at €300 AOV is a very different math from a click that costs €2 that converts at €35 AOV.

High purchase intent on specific queries. Jewelry searches are specific. "14k solid gold necklace 18 inch womens" tells you exactly what the shopper wants and at what price point. Conversion rates on matched queries run 3-6%. Broad fashion queries like "blue summer dress" convert at 0.8-1.5%.

Feed structure advantage. Jewelry feeds have natural structured attributes - metal type, karat, gram weight, length, gemstone. These specs go directly into titles and create precise matching. The accounts that use them well outperform the accounts running brand-first titles by 40-60% in conversion rate on the same impressions.

The gap between 3.5x and 5x on jewelry accounts is almost entirely feed quality and campaign structure. The cut CPA 38% jewelry case study shows exactly what moves a jewelry account from the bottom of the range to the top. The same account, from 1.8x ROAS to 2.6x, through feed and structure changes only.

Below 3x ROAS on a jewelry account usually means one of three things: client-side tracking losing 30-40% of conversions (the most common), titles matching to gold-plated intent on solid gold products, or a one-PMax-for-everything setup averaging your solid gold Champions with your plated tail.

Fashion: 2.8-4x blended ROAS

Fashion is the most competitive vertical we run. Auction density is high. CPC is elevated. Margins are compressed by the sheer number of DTC brands bidding on the same generic queries.

The benchmark range is 2.8-4x. The gap between 2.8x and 4x is almost entirely feed quality and creative cadence.

Feed quality in fashion: titles need to lead with fabric type, cut, gender, and size availability. "Womens Linen Shirt, Regular Fit, Blue, S-XXL, Breathable" beats "Blue Linen Shirt" on purchase-intent matching. Fashion feeds are notoriously poorly optimised because brands invest in marketing copy ("our best-selling summer essential") instead of spec copy.

Creative cadence in fashion: PMax asset groups for fashion need to refresh every 6 weeks at minimum. Seasonal catalog drops each need their own asset pack. Stale creative tanks CTR, which tanks Smart Bidding's ability to scale, which shows up as ROAS decline that looks like a bid problem.

Fashion is also the vertical where seasonal structure matters most. Q4 (gifting) requires a different campaign structure, different ROAS targets, and different budget allocation than Q2. Accounts running a flat structure year-round miss the seasonal swing and underperform both ways.

The accounts at 4x on fashion are the ones with clean feeds, fresh creative on the right cadence, seasonal structure, and server-side tracking. The accounts at 2.8x usually have at least two of those four broken.

	3x ROAS (average)	4x ROAS (top-tier)
Tracking	Client-side pixel (loses 25-35%)	Server-side (full conversion signal)
Feed titles	Marketing copy, brand-first	Spec-first, search-intent matched
Asset refresh	Quarterly	Every 6 weeks per catalog drop
Seasonal structure	Flat ROAS target year-round	Q4 budget uplift + adjusted tROAS
Campaign structure	One PMax	Margin-tier splits

Home decor: 3-4.5x blended ROAS

Home decor has wide seasonal variance. The annual ROAS curve looks like: flat in H1, spike in October-November, step-down in January.

The accounts that hit 4.5x are running seasonal structure. Higher budgets in Q4, adjusted tROAS targets for gifting windows, asset packs specific to the "holiday home" aesthetic (not the same lifestyle images that run in March). The accounts that miss the Q4 window end up at 3x for the year because they averaged the flat H1 with a Q4 they were not structured for.

Home decor also has a strong Demand Gen component. Furniture and home decor buyers research visually - Pinterest pins, YouTube room reveals, Instagram styling. Demand Gen campaigns on YouTube and Discover capture this research phase and feed warmed-up buyers into Shopping. Accounts without Demand Gen run cold traffic at Shopping prices.

The vertical also benefits from room-type custom labels. Accounts that split home decor by room (living, bedroom, kitchen, outdoor) can match seasonal peaks better - outdoor products spike in spring, bedroom in January, kitchen year-round. Without the label, PMax averages across room types and misses the peaks.

AOV in home decor varies widely - from $40 decorative objects to $600 statement pieces. Accounts that run a single PMax for the full price range end up with Smart Bidding averaging across AOVs that have very different ROAS math. Margin-tier splits by AOV band are the structural fix.

Beauty: 3.5-5x blended ROAS

Beauty has one of the highest ROAS ceilings in ecom Google Ads. The reason: repeat purchase rates. Beauty buyers who convert once tend to come back. The lifetime value per acquired customer is high. Smart Bidding learns this over time and bids more aggressively on audiences with high repeat-purchase signals.

The top of the range (5x+) is achievable in beauty but it requires two things: a clean customer list uploaded to Google Ads for value-based bidding tuning, and a properly installed server-side tracking setup that captures repeat purchases without double-counting.

Beauty is also the fastest asset-decay vertical. Trend half-life in beauty is 4-6 weeks. An asset pack that converted at 3% CTR in March can drop to 0.8% by May because the trend moved. Accounts that refresh beauty assets every 2-3 weeks consistently outperform the accounts that refresh quarterly.

The range floor (3.5x) is usually an account with decent structure but stale creative and no repeat-purchase signal feeding Smart Bidding. The ceiling (5x) is an account with fresh creative, customer-list value-based bidding, server-side tracking, and a product line with real repurchase dynamics.

Pet: 3-4x blended ROAS (unusually consistent year-round)

Pet is the most consistent vertical by month across the year. Pet owners do not stop buying when the season changes. Food, health products, accessories - all run relatively flat demand 12 months a year.

This consistency is unusual. Most verticals have seasonal ROAS variance of 30-50%. Pet accounts swing less than 20% month-to-month outside of major promotions.

The 3-4x range in pet is primarily limited by competitive density. The Amazon effect is real in pet - a significant fraction of pet product searches end on Amazon because buyers trust the Prime delivery. Google Shopping pet accounts are competing with that default. The accounts that win are the ones with niche products Amazon does not carry, lifestyle brands with a clear community (the dog breed community is particularly loyal), or health products with proprietary formulas.

Pet accounts also benefit from subscription/recurring product labeling. If your feed has subscription products (monthly flea treatment, recurring food orders), custom label them separately. Smart Bidding should be bidding higher on subscription conversions because the LTV is higher. Without the label, it treats a one-time purchase and a subscription start as equivalent conversions.

Furniture: 4-6x blended ROAS (highest ceiling)

Furniture has the highest ROAS ceiling we see across any vertical. Two reasons.

High AOV filters low-intent clicks naturally. A $1,200 dining table search is inherently higher-intent than a $30 phone case search. The buyer is already in a considered-purchase mindset. Conversion rates on the click are lower (0.5-1.5%), but the conversion value is so high that the ROAS math works out better than verticals with higher conversion rates at lower AOV.

Long consideration filters waste. Furniture buyers research for weeks before buying. Performance Max serve to these buyers on YouTube and Discover during the research phase. By the time they click a Shopping ad, they have already decided on the category and the price point. The click is close to a purchase decision.

The accounts that hit 6x on furniture are the ones with AOV-band custom labels (splitting $200-500 pieces from $1000+ statement pieces), Demand Gen campaigns that capture the research phase, and clean product images with multiple room-context lifestyle shots.

The floor (4x) is usually a furniture account with decent structure but no research-phase Demand Gen, all products in one PMax regardless of price point, and product-only images without lifestyle context that buyers need for large-purchase confidence.

What these benchmarks actually mean for your account

The benchmarks are a diagnostic tool, not a target-setter.

If your account is at the floor of your vertical's range: the structural issues are the first place to look. Feed quality, tracking accuracy, campaign structure. These typically explain 70-80% of the gap between floor and ceiling performance.

If your account is above the ceiling of your vertical's range: you are either capturing a lot of branded traffic inside PMax (which inflates blended ROAS without reflecting real efficiency) or you have genuinely exceptional feed and structure. Verify by looking at non-branded ROAS separately.

If your account is below the floor of your vertical's range: something structural is broken. Client-side tracking losing 30%+ of conversions is the most common culprit - it makes your real ROAS look lower than it is. After that: one-PMax setup averaging across too many products, or titles matching to wrong-intent queries and dragging conversion rate down.

Furniture

4-6x

Pet

3-4x

Beauty

3.5-5x

Fashion (floor)

2.8x

The full benchmark deep-dive

For a full breakdown of ROAS benchmarks by vertical, spend tier, and campaign type, the Google Ads eCommerce Benchmarks 2026 page has the complete data set - including quarterly trends, PMax vs Standard Shopping splits, and branded vs non-branded ROAS by vertical.

The free Google Ads eCom Lab community is also where operators share real account data anonymously. 740+ ecom operators across every vertical, posting actual account performance. If you want to see where your numbers sit relative to a real peer group, that is the place.

What to do if your ROAS is below benchmark

The fastest path to ROAS improvement in 2026 is not a bid change. It is fixing the inputs Smart Bidding bids on.

Three moves in order:

1. Install server-side tracking and validate it against your Shopify back-end. This single fix recovers 25-35% of reported conversion volume on most Shopify stores with client-side-only pixels. Your real ROAS improves immediately when the algorithm bids on the full picture.

2. Rewrite the top 80 SKU titles with specs from your search-term report. This kills the wrong-intent traffic that drags down conversion rate and inflates CPA.

3. Split Performance Max by margin tier. This stops Smart Bidding from averaging your best products with your worst.

The 5-lever scaling system covers the full sequence. The feed optimization guide covers the feed side in detail.

If you want to see where your account sits against your vertical's benchmark and what the gap is driven by, drop the store URL and we will audit it live. We pull the account data in the first 10 minutes and tell you on the spot whether the gap is tracking, feed, structure, or something else.

ROAS benchmarks are honest diagnostics. Use them as such.