Ecommerce Performance Marketing in Australia: What to Measure Beyond ROAS

Ecommerce performance marketing means measuring tracked revenue and contribution margin against total marketing spend, not platform-reported ROAS. Meta and Google both claim credit for the same sale, attribution windows inflate results, and a 4:1 ROAS can still be a loss once COGS, shipping, discounts and returns are factored in. Marketing Efficiency Ratio (MER) and contribution margin are the defensible measures.

If you run paid media for an Australian ecommerce store, you've probably seen a dashboard showing 4:1 or 5:1 ROAS while the bank balance tells a flatter story. That gap isn't a mystery. It's the predictable result of platforms measuring themselves, attribution windows built to flatter ad spend, and nobody subtracting the cost of the product that was actually sold.

This article breaks down why platform ROAS can't be trusted on its own, what Marketing Efficiency Ratio (MER) and contribution margin actually show, and the 90-day sequence we use under the 3P Framework, Profile, Plan, Perform, to rebuild measurement until every dollar of spend can be traced to a real sale.

Key Takeaways

• Platform-reported ROAS double-counts sales across Meta, Google and organic search, and inflates results with view-through windows and modelled conversions.

• Marketing Efficiency Ratio (MER), total revenue divided by total marketing spend, is the one number that survives iOS14.5 and cookie loss because it doesn't depend on a pixel.

• A 4:1 ROAS can still be a loss once COGS, shipping, payment fees, discounting and returns are subtracted from the sale.

• Blended ROAS hides the difference between acquiring a new customer and retargeting someone who already bought, which is why new and returning customer ROAS need separate reporting.

• Defensible ecommerce measurement requires server-side tracking, clean GA4 events, refund data fed back into ad platforms, and reporting built against tracked revenue, not platform metrics.

At a Glance: ROAS, MER and Contribution Margin

Why Platform ROAS Overstates Ecommerce Performance

Platform ROAS overstates performance because Meta and Google both claim credit for the same sale, count view-through impressions as conversions, and let branded search cannibalise demand another channel already created. A customer who saw a Meta ad, searched your brand name on Google, then bought, shows up as a win in two ad accounts, while your bank balance moved once.

Double counting happens because every platform's pixel is built to prove its own value. Meta's ads manager and Google Ads both run their own attribution logic, and neither subtracts the sale the other platform is also claiming. Run a genuine multi-touch or data-driven model across both accounts and combine it with total store revenue, and the two numbers rarely add up. We see this constantly across our performance marketing work: two dashboards, two 4:1 or 5:1 ROAS figures, and a store that only sold one unit.

Attribution windows compound the problem. Meta's default is a 7-day click and 1-day view window. Google Ads commonly defaults to a 30-day click window. A sale that happens three weeks after someone clicked a search ad, having also browsed five other sites and read two comparison articles in between, still gets full credit assigned to that one click. Longer windows don't measure influence more accurately. They just assign more credit to whichever platform's window reaches furthest back.

Modelled conversions make this worse again. Since Apple's iOS14.5 changes and the broader move away from third-party cookies, platforms increasingly fill attribution gaps with statistical modelling rather than verified, trackable events. A modelled conversion is, by definition, a platform's best guess dressed up as a measured outcome. It's not fraud. It's also not proof.

Branded search cannibalisation is the quiet one. An Australian retailer's brand name searches convert at a high rate regardless of ad spend, because the customer already decided to buy and is just finding the checkout link. Running a brand search campaign on top of that demand captures credit for a sale that organic search, or word of mouth, was already going to deliver. We call this search term policing: checking whether a brand campaign is paying for demand your organic listing would have captured for free. Traffic and impressions are vanity traffic if they don't convert to tracked revenue, and this is exactly the kind of number that inflates a platform ROAS report without adding a single incremental sale.

Marketing Efficiency Ratio (MER): The Number That Survives Cookie Loss

Marketing Efficiency Ratio (MER) is total revenue divided by total marketing spend across every channel, for a set period. It needs no pixel, no attribution model and no platform login, which is why it keeps working after iOS14.5 and browser cookie restrictions broke channel-level tracking.

The calculation is deliberately blunt: take total store revenue for the month, divide it by total marketing spend across Meta, Google, affiliates, email tools and anything else with a cost attached. If revenue was $300,000 and total marketing spend was $60,000, MER is 5:1. Nothing about that number depends on whose pixel fired or which platform's model gets the credit.

Setting a target MER starts with contribution margin, not an industry benchmark pulled from a blog post. If your average contribution margin, before marketing spend, sits around a third of revenue, you need a MER of roughly 3:1 just to break even on marketing, before covering overheads, salaries and profit. Brands with thinner margins after COGS, shipping and returns need a higher MER to stay profitable. Brands with strong margins, low return rates and high repeat purchase have more room to invest in growth even at a lower MER.

This is why we push MER to the top of every ecommerce performance marketing report, straight from the account data, rather than leading with channel-level ROAS. MER can't be gamed by attribution window settings, can't be inflated by modelled conversions, and doesn't shift when Apple or Google change a privacy policy. Channel ROAS will always need a caveat. MER rarely does.

Contribution Margin, Not Revenue: The Number Platform ROAS Ignores

Contribution margin is what's left of a sale after the cost of goods, shipping, payment fees, discounting and returns are subtracted, and then marketing spend is subtracted again. It's the only number that tells you whether a profitable-looking ROAS is actually making money, because ROAS only ever measures revenue against ad spend, never true profitability.

Here's a hypothetical example to show the mechanics. It's illustrative only, not a real client's numbers. Imagine a dress selling for $80, with a platform dashboard reporting a comfortable 4:1 ROAS, meaning $20 of ad spend produced that $80 sale. On the surface, that's a strong result.

Now work through the full cost stack. Cost of goods sits around $28. Shipping to the customer costs around $9. Payment processing takes roughly $2. A site-wide discount averaging 15 per cent effectively drops the realised sale price to about $68 rather than $80. A returns rate of 12 per cent means just over one in eight of those sales eventually reverses, with the product, and the margin, coming straight back out of the ledger. Once COGS, shipping, payment fees, the discount and the returns rate are subtracted from that $68 realised average, and the $20 of ad spend is subtracted again, the contribution margin left over can sit close to zero, or turn negative, even though the ROAS dashboard still reads a comfortable 4:1.

This is precisely why we won't scale a campaign off a ROAS screenshot. A store can grow revenue every month and lose more money with every extra sale, and the platform will keep reporting a great number the whole way through. The only defensible test is contribution margin against total spend, calculated at the product or SKU level where COGS and discounting actually vary.

New vs Returning Customer ROAS: Stop Flattering Yourself With Retargeting

Blended ROAS hides the difference between winning a new customer and re-engaging someone who already bought from you. Retargeting to past purchasers or warm site visitors commonly reports ROAS figures of 10:1 or higher because warm audiences convert easily, while genuine cold acquisition might sit closer to 2:1. Averaging the two into one number flatters the account and hides where growth is actually happening.

Separate the two and the picture changes fast. New customer ROAS should be judged against customer acquisition cost (CAC) versus first-order contribution margin, not against a blended average. If acquiring a new customer costs $45 and the first order only returns $30 in contribution margin, that's a loss on order one, and it's only a sound strategy if repeat purchase rate and lifetime value genuinely recover it within a reasonable window. Returning customer ROAS should be judged against genuine incremental lift, using holdout groups where budget allows, because a 10:1 retargeting number often includes customers who were going to buy again regardless of whether the ad ever ran.

This split matters more in Australia than the sheer market size might suggest, because acquisition costs on Meta and Google have climbed as more local retailers compete for the same finite audience. An account that only reports blended ROAS can look healthy for months while quietly bleeding on every new customer it wins, propped up by a retargeting list that will eventually stop growing if acquisition stalls.

The Measurement Stack That Makes This Possible

The measurement stack that makes contribution margin reporting possible combines server-side tracking, clean GA4 events, offline and refund data fed back into ad platforms, and a single reporting layer built against revenue rather than platform-reported clicks. Without this stack, every number upstream is a guess dressed up as data.

Server-side tracking, through tools like Google's server-side tagging or the Conversions API equivalents on Meta, captures events that browser-based pixels increasingly miss due to ad blockers, browser privacy settings and app-based checkout flows. GA4 event integrity means checking that purchase events aren't firing twice on page refresh, that transaction IDs are deduplicated, and that revenue values match what actually settled, including GST treatment, rather than a pre-discount subtotal.

Refund and return data has to flow back into the picture too. A sale that gets returned three weeks later should reduce reported revenue in the platforms and in GA4, not sit there forever as phantom performance. Most Australian ecommerce accounts we review have never connected this loop, which means every return inflates historical ROAS permanently.

Finally, all of it needs to land in one reporting layer, typically Looker Studio, that pulls order data straight from Shopify or WooCommerce and blends it with ad spend across every channel. This is where MER, contribution margin and new versus returning customer performance get reported side by side, against tracked revenue, not traffic. It's also where our performance marketing work starts for every ecommerce client, because you can't plan budget allocation you can't prove.

A 90-Day Sequence to Fix Ecommerce Measurement: Profile, Plan, Perform

Fixing ecommerce measurement takes roughly 90 days under the 3P Framework: 30 days to Profile, rebuilding tracking and defining targets, 30 days to Plan, reallocating budget against contribution margin, and 30 days to Perform, scaling only what the account data proves is working.

Days 1-30: Profile

Profile means auditing every tracking gap before touching the media budget. We rebuild GA4 event integrity, connect server-side tracking, reconcile refund data, and set a target MER based on the store's actual contribution margin, not an industry rule of thumb. Nothing gets a recommendation until we can see which keyword, ad or product line produced a sale, straight from the account data.

Days 31-60: Plan

Plan means reallocating budget against what Profile revealed. Channels and campaigns get split into new customer acquisition and retention, with separate targets for each. Product lines with thin contribution margin, once discounting and returns are counted, get deprioritised in paid media even if their ROAS looks fine, because ROAS was never measuring profit in the first place.

Days 61-90: Perform

Perform means scaling with accountable execution: increasing spend only where contribution margin and MER both support it, cutting channels that only ever looked good on a platform dashboard, and reporting monthly against tracked revenue rather than clicks or impressions. This is also the point where the reporting layer, not a slide deck, becomes the source of truth for every future budget decision.

What 250+ Client Accounts Taught Us About Tracked Revenue

Across more than 250 client accounts, the pattern repeats regardless of industry: traffic and rankings can climb every month while the bank balance stays flat, and the only fix is rebuilding measurement before scaling spend. This is why 3P Digital reports against tracked revenue and qualified leads, never traffic or impressions alone, and why every engagement starts with Profile.

The discipline isn't unique to paid media. An automotive dealership group we worked with had SEO spend running with no clear line back to actual service bookings or vehicle sales. Applying the same contribution-margin thinking to organic, tying every ranking improvement back to booked revenue rather than position tracking, produced a 46:1 return on that SEO investment within 12 months, our best recorded result and one that held up when we checked it against the dealership's own booking data, not a case study slide.

A medical equipment retailer's 325-product catalogue told a similar story from the organic side. The copy was written for casual browsers, not for how clinicians, aged-care procurement teams and home-care buyers actually search. We rewrote and republished 324 of 325 product pages in a single pass, grounded in real Search Console search language rather than guesswork, with every existing page snapshotted first for full rollback. The result was 192 product pages reaching page one of Google, 497 search terms ranking in the top 10, and organic traffic rising 21 per cent around the sweep. That's exactly the kind of organic revenue contribution that belongs in an ecommerce brand's MER calculation, not treated as a separate SEO vanity metric sitting outside the money pages that actually convert. Our ecommerce SEO work is built around that same principle: rank the pages that produce a tracked sale, not the pages that produce a screenshot.

A national recruitment firm gave us a cleaner version of the same acquisition-efficiency lesson that applies directly to new customer ROAS. They were burning budget on job board listings with no compounding return. Shifting spend from rented job board space to owned organic and content assets, built around actual candidate and client search intent through our broader SEO approach, generated 574 leads at a 63.5 per cent lower cost per lead than the previous strategy. The channel was different, but the principle was identical to separating new customer ROAS from retargeting flattery: measure the true cost of winning something new, not the comfortable number a rented platform hands you.

We'd rather tell an ecommerce client the truth about what isn't working than hand over a ROAS report that looks busy. That position costs us clients who want to be told good news. It's also a large part of why our retention sits at 98 per cent across those 250-plus accounts, because when the numbers are genuinely working, month to month with no lock-in gives a client no reason to leave.

If your reported ROAS doesn't match your bank balance, that's not a coincidence, it's the account data trying to tell you something. Book a tracking and profitability review with 3P Digital to see true contribution margin by channel, no pitch, 15 minutes, and find out what your numbers actually say once we stop trusting the pixel.

Frequently Asked Questions

What is a good ROAS for ecommerce businesses in Australia?

There's no universal good ROAS, because it depends entirely on contribution margin, not an industry average. A store with 60 per cent margin after COGS can profit at a lower ROAS than a store with 20 per cent margin, so the right benchmark is always worked out from your own cost stack, not copied from a blog post.

Is MER a better metric than ROAS?

MER is more reliable than platform ROAS because it measures total revenue against total spend without depending on any platform's pixel or attribution model. It doesn't replace channel-level reporting entirely, but it's the number that should sit at the top of any board or cash-flow report, because it survives tracking loss that breaks channel ROAS.

Why does my Meta or Google Ads ROAS not match my actual sales?

This usually comes down to double counting across channels, generous attribution windows crediting sales that happened weeks after a click, and modelled conversions filling gaps left by iOS privacy changes and cookie restrictions. Reconciling ad platform revenue against actual store revenue, including refunds, almost always reveals a gap.

How do I calculate contribution margin for an online store?

Start with the sale price, subtract cost of goods sold, shipping cost, payment processing fees, the average discount applied, and an allowance for the store's return rate. Subtract marketing spend last. What remains is contribution margin, and it should be calculated per product or SKU, since COGS and discounting vary significantly across a catalogue.

Should I separate new customer ROAS from returning customer ROAS?

Yes. Blended ROAS hides genuine acquisition performance behind easy retargeting wins. New customer campaigns should be judged against CAC versus first-order contribution margin and repeat purchase rate, while returning customer campaigns should be judged against incremental lift, not just presence in front of people who were likely to buy again anyway.

References

• Australian Competition and Consumer Commission (ACCC), Digital Platforms Inquiry Final Report: https://www.accc.gov.au/by-industry/digital-platforms-and-services/digital-platforms-inquiry

• Australian Bureau of Statistics (ABS), Retail Trade, Australia: https://www.abs.gov.au/statistics/industry/retail-and-wholesale-trade/retail-trade-australia

• Google Ads Help, About attribution models: https://support.google.com/google-ads/answer/6259715

• Meta Business Help Centre, About attribution settings: https://www.facebook.com/business/help/458681590974355