Note
Updated May 2026. ROAS is the most-quoted paid media metric and the most-misused. This guide covers the formula, the math behind why a 4x ROAS on a 25%-margin product loses money, real benchmark ranges by platform (Meta, Google, TikTok), how Smart Bidding thresholds depend on conversion volume, and what changed in 2026 with Meta's Andromeda rollout dragging reported ROAS by 15-40%.
ROAS is one of the most-quoted metrics in paid media and one of the most-misused. Operators throw around "we hit 4x ROAS" without specifying which ROAS, on which spend window, attributed how. This guide covers what return on ad spend actually means, how to calculate it correctly, and the benchmark numbers worth remembering for Meta, Google, and TikTok in 2026.
What is ROAS
Note
ROAS definition. ROAS (Return on Ad Spend) is the revenue generated per dollar of advertising spend, calculated as revenue attributed to ads divided by ad spend. A ROAS of 4 means every 1 USD of ad spend produced 4 USD of attributed revenue. Sometimes written as a ratio (4:1), sometimes as a multiple (4x), sometimes as a percentage (400%). The 2026 convention in operator forums is to write it as a multiple: "we are running at 3.2x ROAS."
How to calculate ROAS
The formula is simple:
ROAS = Revenue from ads / Ad spend
Where it gets messy is what counts as revenue and what counts as spend.
What counts as revenue
Three flavors:
- Reported revenue from the ad platform. What Meta, Google, or TikTok says they drove. This is attribution-model dependent and tends to overstate. Meta's default attribution as of 2026 is 7-day click + 1-day view; Google Ads default click-through window is 30 days.
- Tracked revenue from your own analytics. What GA4 (formerly Universal Analytics, sunset July 2023) or Adobe Analytics says ads drove. More conservative; depends on your tracking setup. Google renamed GA4 conversions to "Key Events" in March 2024 to reduce confusion with Google Ads conversions.
- Total business revenue divided by ad spend (MER). Marketing Efficiency Ratio. The most honest top-down view but does not tell you which ads worked.
What counts as spend
Almost always the gross media spend on the platform. Some operators include agency fees and tooling; we do not, because most public benchmarks exclude them.
A worked example
Meta says you drove 50,000 USD in revenue. Your CRM says you drove 35,000 USD. You spent 12,000 USD on Meta ads.
- Platform-reported ROAS: 50,000 / 12,000 = 4.17x
- CRM-tracked ROAS: 35,000 / 12,000 = 2.92x
Both are real. They tell different stories. Operators who only report platform ROAS to their CFO eventually get caught.
ROAS vs ROI vs MER vs POAS
| Metric | What it measures | When to use |
|---|---|---|
| ROAS | Revenue per ad dollar (channel-level) | Tactical channel decisions |
| ROI | Profit per dollar invested (after costs) | Whether the business is profitable |
| MER | Total business revenue / total marketing spend | Holistic view; sanity check on platform claims |
| POAS | Profit per ad dollar (ROAS adjusted for gross margin) | DTC operators with margin variance across SKUs |
| CAC | Cost to acquire one new customer | Subscription, SaaS, repeat-purchase models |
| LTV/CAC | Lifetime value vs acquisition cost | Companies with repeat revenue |
Most operators conflate ROAS and ROI. ROAS is a top-line number; it does not consider product cost, fulfillment, agency fees, or platform fees. A 4x ROAS on a 25%-margin product produces almost no profit: 4 USD revenue x 25% margin = 1 USD gross profit, against 1 USD ad spend. Breakeven, before fulfillment, overhead, or platform fees.
The breakeven ROAS formula
Breakeven ROAS = 1 / Contribution Margin
Where contribution margin = (revenue minus COGS minus variable selling costs) divided by revenue.
| Margin | Breakeven ROAS |
|---|---|
| 80% | 1.25x |
| 60% | 1.67x |
| 40% | 2.50x |
| 25% | 4.00x |
| 15% | 6.67x |
Above breakeven you make money; below it you lose money on every order.
Good ROAS benchmarks by platform
These are the rough operating ranges we see across paid media accounts in 2026 across 1,000+ Hyper customer accounts and 10M+ USD/month managed ad spend. Real numbers depend on margins, vertical, funnel stage, and attribution.
Meta Ads ROAS benchmarks
Direct response e-commerce; subscription
- Best for
- Direct response e-commerce; subscription
- Pricing
- N/A
Pros
- Prospecting (cold): 1.5x to 2.5x is healthy
- Retargeting: 4x to 8x is typical (lower volume)
- Blended Meta ROAS: 2x to 3.5x is the operating range for healthy DTC
- Above 5x usually means underspending or attribution leak (Meta over-attributing)
- Advantage+ Shopping campaigns typically run 0.3x to 0.5x above blended Meta ROAS
Cons
- Numbers vary 3-5x by vertical (jewelry differs from apparel differs from supplements)
- Andromeda rollout in 2026 dragged reported ROAS by 15-40% across most accounts. See dedicated post
Google Ads ROAS benchmarks
Search demand capture; Performance Max for catalog
- Best for
- Search demand capture; Performance Max for catalog
- Pricing
- N/A
Pros
- Branded search: 8x to 25x (cherry-picking your own demand)
- Non-brand search: 2x to 5x for healthy accounts
- Performance Max (launched 2021, replaced Smart Shopping by 2023): 2x to 4x typical, depends on shopping share
- YouTube and Display: 1x to 2x is normal (top-funnel role)
Cons
- Branded ROAS distorts blended numbers; always look at non-brand separately
- Performance Max cannibalizes branded search by default; brand exclusion list is mandatory in 2026
TikTok Ads ROAS benchmarks
Discovery; trend-driven products
- Best for
- Discovery; trend-driven products
- Pricing
- N/A
Pros
- Prospecting: 1x to 2.5x typical
- TikTok Shop integrated: 2x to 4x
- Lower than Meta for most accounts in 2026, but improving since TikTok Shop US launch in late 2023
- Better view-through values than click-only attribution captures
Cons
- Heavy creative refresh required; one winning creative does not last more than 2-3 weeks
- Smart Performance Campaigns (TikTok's analog to Performance Max) launched 2024, still maturing
Smart Bidding ROAS thresholds and minimums
Each platform's automated bid strategies have published volume floors. Below these floors, the strategy starves data and runs sub-optimal:
| Platform | Strategy | Min volume to enable |
|---|---|---|
| Google Ads | Target ROAS | 50+ conversions in last 30 days, ideally with values |
| Google Ads | Target CPA | 15+ conversions in last 30 days |
| Google Ads | Maximize Conversions | no published floor; performs best above 30/30 days |
| Meta Ads | Highest volume + value optimization | ad set needs 50+ conversions in 7 days to exit Learning Phase |
| Meta Ads | Cost cap or bid cap | no published floor but variance high on low volume |
| TikTok Ads | Cost Cap (lowest cost with cap) | 20+ conversions in last 7 days typical |
If your account doesn't hit these volumes per ad set or per campaign, switching to a target-ROAS strategy will collapse spend rather than optimize ROAS upward.
What changes ROAS the most
In rough order of impact for a typical paid account:
- Creative. A winning creative can lift ROAS 50-200% over a baseline creative on the same audience.
- Audience targeting. Right-fit audience vs broad audience changes ROAS 30-100%.
- Landing page. A page that converts at 4% vs 2% doubles ROAS at the same CPC.
- Bidding strategy. Manual vs target ROAS vs maximize conversions changes ROAS in different directions depending on data volume.
- Account structure. Combining low-intent and high-intent traffic into one campaign hides which is producing.
- Tracking quality. Enhanced Conversions on Google Ads recovers 5-10% (per Google's published case studies); Conversion API on Meta recovers 30-50% of iOS opt-out signal.
Why ROAS dropped in 2026 (Andromeda)
Most accounts saw reported Meta ROAS drop 15-40% in March-April 2026. The cause is Meta's Andromeda delivery system rollout, which started December 2025 and reached most accounts by mid-March. Andromeda is Meta's next-generation ML system for ads ranking and retrieval, replacing the prior infrastructure across Facebook, Instagram, and Audience Network.
The reported drop is mostly attribution conservatism, not real revenue loss. To verify in your account, triangulate three numbers over the last 30 days:
- Meta Ads Manager reported ROAS
- GA4 Meta-attributed revenue
- Total business revenue divided by total ad spend (MER)
If only Meta dropped while GA4 and total business revenue stayed flat, the loss is reporting noise. Cutting Meta budgets based on Meta's reported ROAS alone has been the single most expensive mistake operators made in March-April 2026, because the cuts removed real revenue on top of the reporting noise.
Full diagnostic at /blog/meta-andromeda-issues-may-2026 and /blog/meta-ads-roas-dropping-2026.
Common ROAS mistakes
ROAS pitfalls to avoid
Do this
Recommended: Report platform ROAS and tracked ROAS side-by-side. Look at branded and non-brand search separately. Adjust for product margin before celebrating (calculate POAS). Compare prospecting and retargeting as separate buckets, not blended. Triangulate with MER monthly. Verify Smart Bidding minimums (15+/50+ conversions per 30 days) before switching strategies.
Avoid
Recommended: Reporting only platform ROAS to leadership. Comparing branded ROAS to non-brand ROAS (apples to oranges). Treating retargeting wins as net-new revenue. Optimizing toward a target ROAS that ignores volume. Conflating ROAS with profit. Switching to Target ROAS bidding before the account hits the 50-conversions-in-30-days floor. Cutting budgets in March-April 2026 based on reported ROAS alone (Andromeda reporting regression).
Autonomous marketing
Grow your business faster with AI agents
- Automates Google, Meta + 5 more platforms
- Handles your SEO end to end
- Improves website conversions
- Runs social media for you
How Hyper helps
Hyper's analytics layer pulls platform ROAS, tracked ROAS, and MER into the same view so the operator can see the gap. The agent then tunes campaigns toward the metric that matters for the business, not just the metric the platform optimizes against. Across 1,000+ customer accounts and 10M+ USD/month managed ad spend, Hyper has caught the Andromeda reporting-vs-real divergence reliably enough that customer ROAS through the rollout window has held within normal seasonal variance. Real numbers in the case study at /blog/ai-marketing-case-study.
Autonomous marketing
Grow your business faster with AI agents
- Automates Google, Meta + 5 more platforms
- Handles your SEO end to end
- Improves website conversions
- Runs social media for you
Frequently asked questions
Q: What is a good ROAS?
It depends on margin. A 4x ROAS on a 70%-margin product is great; a 4x ROAS on a 25%-margin product loses money after fulfillment and overhead. Calculate your breakeven ROAS first (1 / contribution margin), then aim 1.5-2x above it for new acquisition. For a 60% margin business, breakeven ROAS is 1.67x and a healthy target is 2.5-3.3x.
Q: Why do Meta and my analytics show different ROAS numbers?
Meta uses click and view-through attribution with a 7-day click + 1-day view default window. GA4 uses last-non-direct click attribution by default. iOS ATT opt-out (70-80% of iOS users since 2021 launch) and signal loss widens the gap further. Both are real; neither is wrong; they answer different questions. Triangulate against total business revenue (Shopify, CRM) for the holistic view.
Q: Should I optimize for ROAS or volume?
ROAS in isolation is a trap; you can hit a high ROAS by spending almost nothing. Volume matters. Most healthy accounts pick a ROAS floor and then maximize spend up to the floor. The floor is breakeven ROAS plus a profit buffer (typically 1.5-2x breakeven for new customer acquisition).
Q: Is target ROAS bidding worth using?
Yes once the account has 50+ conversions in the last 30 days for Google Ads Target ROAS, or stable Learning Phase exits on Meta. Below those floors, target ROAS bidding starves volume. Maximize conversions with a CPA cap is usually safer for new campaigns until the volume hits the threshold.
Q: What is the difference between ROAS and POAS?
POAS (Profit on Ad Spend) is ROAS adjusted for product gross margin. POAS is more honest because it reflects the dollars actually contributed to the business. Top-tier DTC operators run on POAS, not ROAS. Tools like Lifetimely and Triple Whale's Sonar feature surface POAS by SKU.
Q: Why did my Meta ROAS drop in 2026?
Most likely the Andromeda rollout. Meta's new ML delivery system launched December 2025 and reached most accounts by mid-March 2026. Andromeda's attribution is more conservative than the prior system; reported Meta ROAS dropped 15-40% on most accounts even when real business performance was unchanged. Triangulate against GA4 and total business revenue before assuming the drop is real, and read /blog/meta-andromeda-issues-may-2026 for the full diagnostic.
Q: What is Marketing Efficiency Ratio (MER)?
MER is total business revenue divided by total marketing spend across all channels. It's the holistic sanity check on platform-reported ROAS. If Meta says 4x and Google says 5x but your MER is 2x, the platforms are double-counting attribution or you have a real performance gap that platform numbers are hiding. MER is most useful as a 28-day or 90-day rolling number to smooth seasonality.
Q: Can I trust Meta's reported ROAS in May 2026?
Less than usual. The Andromeda rollout has produced reporting regressions of 15-40% across most accounts. Meta hasn't formally acknowledged this; their stance is 'expected behavior during rollout.' The fix is not to trust platform ROAS in isolation. Pull GA4-attributed Meta revenue and total business revenue into the same view weekly, and treat reported ROAS as one of three signals rather than the truth.