The essentials
- ROAS: revenue generated divided by ad spend. A ROAS of 4 means €4 of sales for every €1 invested.
- Before optimising: check that measurement is accurate (pixel, conversions API, attribution). A false ROAS leads to bad decisions.
- Highest-impact levers: offer and landing page, ad creative, exclusion of existing customers, budget concentrated on profitable campaigns.
- 2026 benchmarks: median e-commerce ROAS on Meta between 2.5 and 4; the break-even point is calculated from your margin, not from a universal figure.
The aim of ROAS (Return On Ad Spend) is to measure the direct profitability of social ads campaigns. It is calculated by dividing the revenue attributed to advertising by the spend. A falling ROAS rarely has a single cause: saturated audience, worn-out creative, slow landing page, faulty measurement. Here is the method for diagnosing it, then the ten levers that improve ROAS, ranked by the impact observed on SME and e-commerce accounts.
Setting the right ROAS target
A ROAS is only good or bad relative to your margin. The break-even point is calculated as follows: minimum ROAS = 1 / gross margin. With a gross margin of 40%, a ROAS of 2.5 just covers the advertising cost; you need to aim higher to make a profit, unless customer lifetime value (see LTV) justifies accepting a first sale at a loss.
| Platform (2026) | Median e-commerce ROAS | Note |
|---|---|---|
| Meta Ads (Facebook, Instagram) | 2.5 to 4 | The most mature; catalogue and retargeting pull the average up |
| TikTok Ads | 1.5 to 3 | Strong on discovery, weaker on direct conversion |
| Pinterest Ads | 2 to 3.5 | Good for decoration, fashion and homeware |
| LinkedIn Ads (B2B) | Not relevant as a direct ROAS | Measure cost per qualified lead and pipeline |
Step zero: check the measurement
Before any adjustment, check that the ROAS shown is reliable. Three checks: the pixel and the conversions API (Meta) or the Insight Tag (LinkedIn) are indeed sending purchases with their value; the attribution window (7 days after click, 1 day after view by default on Meta) is known and consistent with the buying cycle; the platform's figures are reconciled with GA4 and with actual sales. A gap of 20 to 40% between Meta ROAS and GA4 ROAS is normal (different attribution); a larger gap signals a tracking problem. The detail is in conversion tracking in SMA.
The ten levers, by impact
| # | Lever | Typical effect on ROAS | Effort |
|---|---|---|---|
| 1 | Offer and landing page (speed, clarity, proof, price) | +20 to +60% | Medium to high |
| 2 | New creative every 2 to 4 weeks, short video formats and UGC | +15 to +40% | Moderate |
| 3 | Excluding recent customers and converters | +10 to +25% | Poor |
| 4 | Concentrating the budget: cutting ad sets below the break-even point | +10 to +30% | Poor |
| 5 | Segmented retargeting (basket, product viewed, 7/30/90 days) with distinct messages | +10 to +20% | Moderate |
| 6 | Broad audiences with Advantage+ targeting or 1 to 3% lookalike audiences | +5 to +20% | Poor |
| 7 | Value-based bidding (minimum ROAS) rather than volume | +5 to +15% | Poor |
| 8 | Clean product catalogue (titles, images, prices, stock) | +5 to +15% | Moderate |
| 9 | Automatic placements and suitable formats (9:16 for Reels and Stories) | +5 to +10% | Poor |
| 10 | Scheduling: reinforce high-conversion periods, cut back the quiet ones | +5 to +10% | Poor |
Levers 1 and 2 in detail
Offer and landing page. A page that takes more than 3 seconds to load on mobile, with no visible price and no social proof, loses half of its paid visitors. Test: speed (PageSpeed), consistency between the ad's promise and the first screen, customer reviews, delivery costs announced early, a short funnel. A specific offer (a bundle, free delivery above a threshold, a first-order deal) converts better than a general discount and protects your margin.
Creative. On Meta, ad fatigue sets in within 2 to 4 weeks: frequency above 3, falling click-through rate, rising CPM. Renew with different angles (problem, proof, demonstration, comparison) rather than colour variants. In 2026, videos of 10 to 20 seconds and phone-shot content (UGC) achieve costs per purchase 20 to 40% lower than studio visuals in most consumer sectors. The criteria are in validating your ad creative.
Structure and budget
- Simplify: 2 to 4 campaigns (prospecting, retargeting, catalogue, and possibly brand) are enough; fragmentation stops the algorithm learning. See structuring your Meta Ads account.
- Let it learn: an ad set needs around 50 conversions per week to leave the learning phase; do not change budget and targeting every day.
- Decide on marginal ROAS: increasing the budget of a profitable campaign lowers its ROAS; the goal is total profit, not the highest ROAS. See what budget to plan for.
- Exclude: customers from the last 30 or 60 days in prospecting, employees, and visitors who have already converted in retargeting.
Tracking and deciding each week
A weekly table by campaign with spend, attributed revenue, ROAS, cost per purchase, frequency, click-through rate and CPM lets you attribute a drop to its cause: rising CPM (auction, season), falling click-through rate (creative), falling conversion rate (page, offer, traffic quality). The complementary metrics are in the SMA KPIs ; the reconciliation with GA4 in GA4 and SMA.
How GreenRed helps
Rather than juggling several tools, GreenRed's return on investment module brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.
Frequently asked questions
What is a good ROAS for social ads?
One that exceeds 1 divided by your gross margin. With a 50% margin, a ROAS of 2 covers the spend; aim for 3 or more to make a profit. The 2026 e-commerce medians sit between 2.5 and 4 on Meta, but a ROAS of 2 can be excellent on a high-margin product or one that is bought again often.
Why does my ROAS fall when I increase the budget?
Because the algorithm goes looking for less responsive audiences once the most likely ones are exhausted: this is diminishing returns. Increase in steps of 20% every 3 to 4 days and track total profit rather than average ROAS; as long as the marginal ROAS stays above the break-even point, the increase is justified.
Meta's ROAS and GA4's ROAS are different, which should I believe?
The two measure different things: Meta attributes sales after a click and after a view within its window; GA4 attributes to the last click or according to a data-driven model. Use Meta to optimise campaigns against each other and GA4 to arbitrate between channels. A gap of 20 to 40% is normal.
Should margin be included in ROAS?
Classic ROAS does not include it. Some accounts track a margin ROAS (gross margin generated divided by spend) or a POAS (Profit On Ad Spend), closer to economic reality, especially when products have very different margins. That means sending the margin as the conversion value.