A $20 cost per purchase can be brilliant or disastrous. If that customer buys a $90 product at a 70% gross margin, you have room to scale. If they buy a discounted $45 product, return it two weeks later, and never purchase again, your ad account is quietly burning cash.
That is the problem with most Meta reporting. It answers whether the platform can claim a conversion, not whether the business made more money. Founders asking what ad metrics actually predict revenue need to look past the tidy dashboard numbers and into the economics underneath them.
The metric most agencies avoid: contribution after acquisition
Revenue is not profit. ROAS is not profit. And a lower CPA is not automatically a win.
The commercial question is simple: after product cost, shipping, fulfilment, transaction fees, discounts, returns and ad spend, does each new order leave enough contribution to fund growth?
Start with contribution margin per first order. If your average order value is $100 and your variable costs are $55, you have $45 of contribution before advertising. A $30 new-customer acquisition cost leaves $15. A $50 acquisition cost loses $5 on the first transaction.
Neither result is automatically good or bad. It depends on whether that customer repurchases, how quickly they repurchase, and whether cash flow can carry the gap. A consumable brand with a strong 60-day repeat rate can rationally acquire customers at a first-order loss. A one-off product brand usually cannot.
This is why the best advertising target is not a generic ROAS goal copied from another Shopify store. It is a target based on your actual margin, repeat purchase behaviour and cash position.
What ad metrics actually predict revenue?
There is no single metric that predicts revenue in isolation. There is a small set of numbers that, read together, tells you whether Meta is creating profitable demand or simply collecting credit for sales that might have happened anyway.
1. New customer revenue and new customer CAC
This is the clearest starting point. Existing customers are cheaper to convert, more likely to recognise your brand, and often already in your email or SMS flows. Retargeting them can make a Meta account look efficient while new customer growth stalls.
Track revenue from first-time purchasers separately from returning-customer revenue, then calculate new customer CAC: prospecting spend divided by genuinely new customers acquired.
A healthy blended ROAS can hide a deteriorating new customer CAC. That matters because a brand cannot scale indefinitely by repeatedly selling to the same pool of buyers. If prospecting efficiency is slipping, you need to know before your retention engine runs out of people to retain.
Meta attribution is useful here, but Shopify customer data is the source of truth. Compare period-on-period trends in first-time orders, not just what Ads Manager says it delivered.
2. Blended CAC
Blended CAC measures total acquisition spend across channels against all new customers. It is less flattering than platform-level CPA and far more useful for running a business.
Why? Because customers do not move through a tidy, single-channel funnel. They see a Meta ad, search your brand later, click an email, ask a mate, then buy on mobile. Every platform wants the credit. Your bank account does not care who claims it.
Use Meta’s reported CPA to diagnose campaign performance. Use blended CAC to decide whether you can spend more. If Meta CPA improves but blended CAC worsens, something is off. You may be cannibalising organic demand, over-investing in retargeting, or seeing attribution inflate while total customer growth softens.
3. MER, or blended revenue efficiency
Marketing efficiency ratio is total revenue divided by total marketing spend. Some brands use total paid media spend; others include creative production, agency fees and retention marketing. The definition matters less than using the same definition consistently.
MER is not a magic number. It can be distorted by a major organic launch, wholesale revenue, seasonal demand or a viral creator post. But it is one of the best guardrails against channel-level nonsense.
If Meta reports a 4.0 ROAS while total marketing spend rises 40% and store revenue barely moves, the account is not performing as well as the report suggests. Conversely, a lower reported Meta ROAS can be acceptable when total revenue, new customer volume and contribution are rising profitably.
The aim is not to worship MER. It is to make sure paid social is expanding the pie, not fighting over slices it did not create.
4. Conversion rate by traffic temperature
Site conversion rate does not predict revenue on its own, but it tells you where the constraint lives.
A weak cold-traffic conversion rate can signal a poor offer, mismatched creative, slow product pages, shipping sticker shock or an audience that is too broad for the message. A strong retargeting conversion rate with a poor prospecting conversion rate often means your brand is closing people once they are warmed up, but your acquisition message lacks bite.
Do not judge every campaign against the same conversion-rate benchmark. Cold traffic should convert differently from branded search, email traffic and repeat customers. The useful question is whether cold traffic is converting well enough at your target CAC to create contribution.
5. Average order value and discount dependence
AOV is often treated as a merchandising metric. It is also an acquisition lever.
Lifting AOV through bundles, quantity breaks, post-purchase upsells or better product pairing gives you more room to bid for customers. A $10 increase in AOV can be worth more than weeks of micro-optimising audiences.
But watch how the lift happens. If conversion only improves when you run a 25% discount, you may be purchasing revenue at the cost of margin and future customer expectations. Discount-led ROAS can look brilliant right until the promotion ends.
Track AOV after discounts, not before them, and pair it with contribution margin. Revenue that relies on margin destruction is not scale.
6. Repeat purchase rate and payback period
For brands with genuine repeat behaviour, the most valuable number is often payback period: how long it takes for gross profit from a newly acquired customer to repay the acquisition cost.
A 12-month lifetime value projection is easy to make look attractive. Cash does not arrive over 12 months when Meta charges your card today. Track 30-, 60- and 90-day cohort revenue instead. You want to know whether customers acquired this month behave as well as customers acquired three months ago.
If new cohorts are buying once and disappearing, a high historical LTV is irrelevant. Your acquisition model has changed.
Metrics that help diagnose but do not predict revenue
CTR, CPM, CPC, frequency and thumb-stop rate matter. They are not rubbish. They tell you where to investigate.
A rising CPM may indicate tougher auction conditions or a narrower audience. A falling CTR can point to creative fatigue or a message that no longer earns attention. A cheap CPC may show that an ad is entertaining people who have no intention of buying.
The mistake is treating these as performance outcomes. They are diagnostic inputs. A creative with an average CTR can outperform a high-CTR ad if it attracts buyers with higher AOV and stronger repeat purchase behaviour.
The same applies to frequency. High frequency is not automatically bad. For a short promotion, a warm audience or a high-consideration product, repetition can be necessary. It becomes a problem when spend rises, incremental conversions flatten and creative is not refreshed.
Build a revenue scorecard that cannot hide the truth
Your weekly scorecard should connect media activity to business outcomes. At minimum, track total revenue, new customer revenue, new customer count, blended CAC, MER, AOV, contribution after acquisition, and cohort repurchase performance where relevant.
Then use Meta metrics to explain movement, not to replace those numbers. If new customer CAC rises, look at creative performance, prospecting conversion rate, CPM, landing-page behaviour and audience allocation. If MER falls, check whether total revenue failed to keep pace with spend, whether discounting inflated demand, or whether a channel is claiming too much credit.
This approach is less convenient than a screenshot showing green ROAS. It is also how you stop making decisions on platform theatre.
The real test is incremental growth
The cleanest question is not: did Meta report purchases? It is: did increasing Meta spend create more profitable revenue than the business would have produced without it?
You will not get a perfect answer every day. Attribution windows, seasonality, stock availability and creative volatility all complicate the picture. But you can get closer by monitoring blended results, holding out audiences where practical, watching geographic or budget changes, and comparing new-customer cohorts over time.
For founder-led Shopify brands, that level of clarity is not a luxury. It is the difference between scaling a real acquisition engine and feeding an ad account that is good at taking credit.
The next time someone sends you a glowing Meta report, ask one question before you celebrate: did we acquire more profitable new customers, at a payback period this business can actually afford?