How to improve Facebook ads ROAS.

The short answer

Return on ad spend (ROAS) on Meta is determined by four levers: creative quality, audience match, attribution setup, and post-click funnel conversion. Fix them in that order. Most brands assume the problem is creative when it's actually attribution or funnel. The diagnostic is to compare Meta-reported ROAS to GA4 or back-end revenue: a large gap signals attribution, a closely-matching low number signals funnel, and a low number on both signals creative or audience.

The diagnostic, in 60 seconds

Before you change anything, compare your Meta-reported ROAS to your back-end revenue (Shopify, Stripe, your CRM). Three patterns emerge:

• Meta says 4.2x, back-end says 1.5x. Attribution problem. Meta is over-claiming. Fix attribution before touching creative.
• Meta says 1.2x, back-end says 1.3x. Funnel problem. Conversions are happening, just not enough. Fix the post-click experience.
• Meta says 0.8x, back-end says 0.9x. Creative or audience problem. The ads aren't winning enough buyers. Fix creative first.

Skip this diagnostic and you'll spend a quarter rebuilding creative when the actual problem is your Conversion API or your Klaviyo flow.

Lever 1 · Attribution

Since iOS 14.5 (April 2021), Meta has under-attributed by default. The Conversion API (CAPI) closes most of that gap by sending purchase events server-side. If your CAPI integration is partial or missing, Meta sees fewer conversions than actually happened. Optimisation suffers, audiences degrade, and ROAS reports look worse than reality.

The fix: full Conversion API with event match quality above 8/10. Use Meta's Event Match Quality report to see your current state. Most accounts under £50k a month spend have an EMQ score below 7 and don't know it.

Secondary: review your attribution window. Default 7-day click + 1-day view is right for most e-commerce. Shorter windows under-report; longer windows over-report.

Lever 2 · Funnel and post-click experience

ROAS is ad spend divided by revenue. If your conversion rate is 1.2% when it should be 2.8%, you need to spend 2.3x as much to hit the same revenue. The ad creative is irrelevant if the landing page converts at 1.2%.

Common funnel breaks for Meta-driven traffic:

• Mobile load speed above 3 seconds. Meta traffic is 80 to 90% mobile; speed kills conversion.
• Hero section that doesn't restate the ad's promise.
• Single-product page with no upsell or cross-sell.
• Checkout asking for too much information (postcode, phone, marketing consent, terms acceptance, captcha).
• No Apple Pay or Google Pay on mobile checkout.
• Klaviyo abandoned cart flow that fires once at 24 hours instead of three times across the first 48 hours.

Lever 3 · Creative

Once attribution and funnel are correct, creative is usually the biggest lever left. Three rules:

• Volume beats polish. Four creative tests per month at decent quality beats one polished campaign per quarter. Meta's algorithm needs variation to learn.
• Static plus motion. Mix static images, short product videos, and UGC-style talking-head clips. Don't pick one format.
• Hook in the first 1.5 seconds. Mobile feed scrolling is fast; either the first frame stops the thumb or the ad is wasted spend.

On creative production: Mira shoots photo and video in-house as an add-on to Paid Social retainers because the wave cadence is hard to sustain on freelance pipelines.

Lever 4 · Audience

Meta's Advantage+ Shopping campaigns are now the default recommendation for e-commerce, and in most cases they outperform manually-targeted campaigns. The audience work has shifted: instead of building lookalike layers, you upload high-quality first-party data and let Meta's algorithm find matches.

Where manual audience work still wins:

• Retargeting with custom audiences (90-day site visitors, 180-day purchasers, 7-day cart abandoners).
• Cold prospecting on niche categories where Advantage+ struggles to find the audience.
• Geo-targeted regional campaigns where Advantage+ over-spends in low-value markets.

The order to fix things in

1. Run the diagnostic (Meta vs back-end ROAS comparison).
2. Fix attribution if Meta's number is materially higher than back-end. CAPI, EMQ above 8/10.
3. Audit the funnel. Mobile speed, checkout friction, post-purchase flow.
4. Test creative at higher cadence. 4 to 6 fresh assets per month, mixed static and motion.
5. Refine audience structure. Advantage+ for prospecting, manual retargeting for the back end.

"Four out of five accounts have an attribution problem they don't know about. They keep replacing creative against numbers that aren't real."

A pattern across the last 12 months of Meta audits we've run

What ROAS to aim for

Depends on your gross margin and customer lifetime value. Rules of thumb for UK e-commerce in 2026:

• 40 to 60% gross margin: target blended ROAS of 3.0x or higher.
• 60 to 80% gross margin: target 2.0x or higher.
• Subscription / repeat-purchase model: aim for break-even on first purchase, profit on LTV. Higher first-purchase ROAS is a bonus.

Originally published October 2024. Refreshed May 2026 with Advantage+ recommendations and current EMQ benchmarks.