You're probably here because the ROAS number in Meta Ads Manager looks fine, but the business doesn't feel fine. Revenue is growing, maybe even up year-over-year, but cash in the bank isn't. Your agency sends weekly reports showing a healthy 3.2x blended ROAS. Your accountant keeps asking why margins are compressing. Both of them are looking at real numbers. Only one of them is telling you the truth.

Blended ROAS has become the default metric in DTC because it's easy to report, visible in every ads dashboard, and simple enough that founders without a marketing background can understand it. That doesn't make it correct. It makes it convenient. Most of the time, the ROAS number your agency is optimising toward isn't the number that determines whether your business is actually growing, breaking even, or quietly bleeding money.

This post is the argument for retiring blended ROAS as your primary growth metric and replacing it with four numbers that actually tell you what's happening. It's opinionated on purpose. If you run ads and you measure success by the ROAS number Meta shows you, you're measuring the wrong thing, and everything downstream of that measurement (budget decisions, creative tests, campaign structure) is being optimised toward a number that doesn't map to profit.

Let's walk through why ROAS became the default, the four specific ways it misleads, what to track instead, and how to actually use those new metrics without building a full analytics team.

#Why ROAS Became the Default Metric

Before attacking the metric, it's worth being honest about why it caught on.

ROAS is easy to calculate. Revenue divided by ad spend. Anyone can do it in their head. It shows up prominently in every ad platform dashboard. It's the number agencies report on because clients ask for it. It's the number founders ask about because it's the number they were taught to ask about. And in the early days of Facebook Ads, when tracking was reliable and the market was less competitive, ROAS correlated reasonably well with profitability.

None of those conditions still hold. iOS 14.5 broke the tracking foundation in 2021. The market got more competitive, driving up acquisition costs. Platform attribution models shifted toward modelled conversions that over-credit paid channels. And the biggest issue, which was always there, became impossible to ignore: ROAS measures revenue, not profit, and revenue and profit are not the same thing.

Agencies still report ROAS because that's what clients ask for. Clients ask for it because that's what they've always asked for. It's the industry's shared language even though the language is increasingly imprecise. The first step to fixing this is naming it clearly: ROAS is a legacy metric from an earlier era of paid media, and it's doing more harm than good in 2026.

#The Four Problems With Blended ROAS

There are four specific ways blended ROAS misleads founders. You can probably recognise at least two of them in your own business.

#Problem 1: Repeat Customers Inflate the Number

The most common version of this failure looks like the business has been running for two or three years, has a decent repeat customer base, and Meta shows a blended 4x ROAS. The founder feels good. The business is actually stalling.

Here's what's happening. The 4x blended number includes revenue from existing customers who would have bought anyway, plus revenue from retargeting campaigns that show ads to people who were going to convert regardless. Strip those out, and new-customer ROAS on the same campaigns might be 1.6x or 1.8x. The paid channel isn't acquiring profitably. It's taking credit on the acquisition side for revenue that the brand itself earned through retention.

The more mature your customer base, the worse this gets. A brand with strong repeat behaviour can have a 4x blended ROAS while losing money on every new customer acquisition, because the repeat customers in the numerator are covering for the losses on new customer acquisition. You can grow revenue for a while this way. You can't grow the customer base.

The fix is to split new-customer ROAS from repeat-customer ROAS and make budget decisions based on the former. Most ad platforms can't do this natively. Shopify can, which is where most of the diagnostic work actually happens.

#Problem 2: Platform Attribution Overcredits Meta

This is the iOS 14.5 problem and it hasn't gone away, it's just gotten more sophisticated.

Meta reports ROAS using a 7-day click and 1-day view attribution window, blended with modelled conversions that fill in gaps where tracking is missing. The modelled conversions are optimistic. Meta's own reported ROAS is often 30% to 50% higher than what post-purchase surveys (the "how did you hear about us" question) or geo-holdout tests show.

That gap matters. If Meta says your ROAS is 2.8x and post-purchase attribution says it's 1.9x, you're not breaking even, you're underwater. But you're running budget decisions against 2.8x because that's the number on the dashboard.

You don't need a full incrementality testing setup to correct for this. A simple post-purchase survey on your checkout (one question, multiple choice, "where did you first hear about us") is one of the highest-leverage data sources in DTC and most brands don't run one. Compare what customers say to what the platform claims, and the delta is your attribution correction factor.

#Problem 3: ROAS Ignores Contribution Margin

ROAS treats a dollar of revenue as a dollar. A dollar of revenue is not a dollar. It's a dollar with COGS, shipping, fulfilment, payment processing, returns, and overhead pulled out of it, and what's left is contribution margin. That's the number that actually covers your CAC.

Take a product with a $60 AOV, 55% gross margin, $7 shipping cost, $2 payment processing, and a 12% return rate. The contribution margin per order isn't $33 (revenue times gross margin). It's closer to $22 after the variable costs. A 2x ROAS on this product means you spent $30 to get $60 in revenue, and you netted negative $8 on contribution. You lost money, at a 2x ROAS that looks "healthy" on the dashboard.

For most DTC brands, the actual break-even ROAS on new customer acquisition (before considering lifetime value) is between 2.5x and 3.5x, not 1x and not 2x. If you're targeting 2x as your goal and you don't know your contribution margin, you're setting a target that guarantees losses on every new customer.

This is the single most common profit leak in DTC advertising. Founders who cannot recite their contribution margin to within 5% are making spend decisions on ratios that don't map to their unit economics.

#Problem 4: ROAS Can Look Healthy While Growth Stalls

This is the slowest failure and the most expensive one. A brand's blended ROAS stays at 3.5x for six months. Revenue is flat or growing slowly. Nothing looks broken. Then the founder notices that new customer count has been declining for 14 weeks. The ROAS was healthy because repeat customers were carrying it. The acquisition engine quietly stopped working months ago and nobody spotted it because the headline number never flinched.

ROAS is backward-looking and efficiency-weighted. It tells you whether the dollars you spent produced revenue. It doesn't tell you whether the business is growing. Those are different questions. A brand can have high ROAS and a shrinking customer base, which is a slow-motion failure mode that looks fine on a dashboard until the repeat base starts aging out and there's no new customer cohort behind it.

The leading indicator here isn't ROAS. It's new customer count, new customer revenue, and new customer CAC. If those three numbers are healthy, ROAS is usually fine and you don't need to look at it. If those three are soft, no amount of blended ROAS healthiness will save you.

#What You Should Actually Track

Four metrics. Together, they give you the signal ROAS pretends to give you but doesn't.

#Metric 1: New Customer CAC

New customer CAC is total paid acquisition spend divided by new customers acquired from that spend. You can calculate a rough version in Shopify by pulling total Meta spend for the period and dividing by new customer count with Meta as the first-touch source.

A more precise version uses post-purchase survey data to attribute new customers to channels. The calculation is the same, but the input is cleaner. A rough version is still better than no version.

Why this matters: new customer CAC is the number you manage budget against. If it creeps up 20% over three weeks, you have a problem even if ROAS looks fine. If it stays stable while spend grows, the acquisition engine is scaling. Track it weekly. Set an alert if it moves more than 15% from the trailing four-week average.

#Metric 2: Contribution Margin Per New Customer

This is AOV on first orders multiplied by contribution margin, minus CAC. It's the number that tells you whether you make money on each new customer acquisition at the point of first purchase.

For most DTC brands, this number is negative on first order. You're buying the customer at a loss and counting on repeat purchases to make the economics work. That's fine, but only if you know how negative and how fast you recover it.

Example: $58 first-order AOV, 38% contribution margin after variable costs, $35 CAC. Contribution margin per new customer at first purchase is $58 × 0.38 - $35 = -$12.96. You lose $13 on every new customer on day one. The question becomes: how fast does the repeat behaviour make up that $13 and then earn you meaningful profit on top?

Which leads to the next metric.

#Metric 3: Payback Period

Payback period is how long it takes for the contribution margin from a new customer to cover the CAC. It's expressed in days or months.

The calculation: pull a cohort of new customers from a specific month, track their cumulative contribution margin over time, and find the point where the cumulative contribution margin equals the CAC you paid to acquire them.

Different businesses have different acceptable payback periods. A subscription business might target 90-day payback because the predictable recurring revenue makes longer paybacks safer. A one-off purchase business with irregular repeat might need 60-day or first-order payback because you can't count on the second purchase. A high-AOV consideration purchase (furniture, appliances) might tolerate 180+ days.

The specific target matters less than knowing yours and tracking it. If your payback period extends from 75 days to 110 days over three months and nothing else about the business has obviously changed, your acquisition economics are deteriorating in a way blended ROAS will not show you.

#Metric 4: Incremental Lift (If You Can Measure It)

This is the hardest of the four and the most important if you can pull it off. Incremental lift measures how much of the revenue Meta is claiming would have happened anyway versus how much is truly caused by the ads.

There are three ways to measure it, from simplest to most rigorous:

• Post-purchase survey attribution compared to platform attribution. The delta is your incrementality correction factor.
• Geo holdout testing. Turn Meta off in one geography (or a set of zip codes) while keeping it on in a comparable region. Compare new customer acquisition between the two.
• Full conversion lift studies through Meta's own platform, which requires enough spend to hit the threshold.

Most brands can't do rigorous incrementality testing without meaningful budget and analytical chops. That's fine. A post-purchase survey gets you 70% of the value at 5% of the complexity. The question you're answering is: when Meta says it drove a conversion, would that conversion have happened without Meta? The more "direct" or "organic" or "Google" answers you see in the survey, the more Meta is claiming credit it doesn't deserve.

If you do nothing else on this list, add a post-purchase survey. It's the highest-leverage data source most DTC brands don't have.

#How to Set Targets on the Right Metrics

Once you're tracking the four metrics, the next question is how to set targets that make sense for your business.

Here's the flow, in order:

1. Calculate your true contribution margin per order. Revenue per order minus COGS, shipping, payment processing, packaging, fulfilment, returns reserve. This is the ceiling on what you can spend to acquire a customer before you start losing money on day one.

2. Decide your acceptable first-order payback. Break-even on first order is the safest. If you have strong repeat behaviour and enough runway to finance the float, you can spend more than contribution margin and accept a payback period of 30, 60, or 90 days.

3. Back into max CAC. Your max CAC is contribution margin per order if you require break-even, or contribution margin plus acceptable loss per order if you're willing to pay for LTV.

4. Set weekly targets on new customer CAC and payback period. These become your primary budget gates. If either breaches the target, investigate before you scale.

5. Use ROAS as a secondary check, not a primary one. ROAS is fine as a daily sanity check. It's not the number you make budget decisions on.

Here's what that looks like in practice.

The daily ROAS glance is fine. Making weekly or monthly budget decisions based on it is not.

#Common Objections and How to Handle Them

These come up every time I run this framework with a founder.

"But my agency reports ROAS."

Ask them to report new-customer ROAS alongside blended ROAS, plus new customer CAC, plus a rough payback estimate. If they can't, that's useful information about whether they're thinking about your business or just managing your ad account. Many agencies will do this immediately when asked. Some won't, and the reason is usually that their internal optimisation is built around the wrong metric.

"I don't have time to build out contribution margin tracking."

You don't need to build anything. You need to sit down once with your accountant or finance person, calculate contribution margin per order across your top 5 SKUs, and write the number down. It doesn't change week to week. It changes when your COGS, shipping, or return rate shifts meaningfully, which is maybe quarterly. One afternoon of work produces a number you'll use weekly for the next year.

"Post-purchase surveys will hurt my conversion rate."

They show after purchase, not before. They don't touch conversion. On mobile, a single well-designed question adds about 5 seconds to the post-checkout experience. Customers fill it in because they just bought something and they're in a good mood. Response rates of 40% to 70% are normal.

"Incrementality testing is too complex for my budget."

Then don't do it. Use post-purchase survey attribution instead. It's not as rigorous but it's directionally correct and accessible. Incrementality testing is for brands spending enough on Meta that the decision risk justifies the analytical complexity. If you're spending $15K/month on Meta, survey attribution is enough.

"I need a fast metric to check daily."

ROAS is fine for daily directional checks. A 3x day is probably better than a 1x day. Just don't make structural budget decisions from the daily number. Daily ROAS is for noticing that something broke. Weekly CAC and monthly payback are for deciding what to do about it.

"My ads agency says this is overthinking it."

It depends on the agency. The ones who push back on this framework are often the ones whose internal systems optimise for ROAS because that's what their dashboards are built around. The ones who agree with it (there are plenty) run their client strategy on new customer economics, not platform-reported ratios.

#Why Meta Still Matters

This post is not anti-Meta. Meta Ads remains the most effective paid acquisition channel for most DTC brands at $1M to $10M. The argument is not that you should stop running Meta or stop trusting paid acquisition. The argument is that Meta's own ROAS dashboard is not the right place to measure whether your paid acquisition is working.

The distinction matters because founders who hear "ROAS is broken" sometimes overcorrect into anti-advertising positions. That's the wrong takeaway. Meta is still where you'll find new customers efficiently. The platform attribution is just an imperfect mirror of what's happening, and blended ROAS is the wrong single metric to manage against.

A healthy paid acquisition operation in 2026 looks something like this:

• Meta is the primary acquisition channel
• New customer CAC is tracked weekly against a contribution-margin-derived ceiling
• Payback period is tracked monthly against a business-specific target
• Post-purchase survey corrects platform attribution by a known factor
• Blended ROAS is a secondary check, not a primary decision input

The brands that grow profitably in DTC aren't the ones avoiding paid. They're the ones measuring paid correctly.

#A Quick Worked Example

Let's run through a scenario so this isn't abstract.

You're doing $250K per month. Meta is your primary paid channel at $40K/month spend. Meta reports a blended 3.1x ROAS. Your agency says things are healthy.

You run the framework.

• Contribution margin per order: $22 on a $62 AOV (35% after everything variable)
• Post-purchase survey attribution: 52% of Meta-attributed new customers say "Instagram" or "Facebook" as the first touch. The rest cite Google, direct, or organic. Your real incremental Meta contribution is around 52% of what Meta claims.
• True new customer CAC after the attribution correction: $41 (not $27 as Meta suggests)
• Contribution margin per new customer on first order: $22 - $41 = -$19
• Payback period: 68 days (you recover the $19 loss over roughly two months of repeat purchases)

Now the picture is clearer. You're not making money on day one of a new customer acquisition. You're losing $19 per new customer upfront and recovering it in about two months through repeat behaviour. If your cash position can finance that float, it's a viable model. If not, the acquisition machine is eating cash faster than the business is making it, even though blended ROAS looks fine.

The decision question isn't "is Meta working?" It's "is a 68-day payback acceptable for our cash position, and is that payback stable or extending?" Those are different questions than "what's our ROAS" and they lead to different answers.

#Building This Into Your Weekly Rhythm

You don't need a BI tool or a dedicated analyst to run this framework. You need:

• Contribution margin per order, calculated once and updated quarterly
• New customer count and new customer revenue, pulled weekly from Shopify
• Meta spend, pulled weekly from Ads Manager
• Post-purchase survey data, ongoing
• A simple note or spreadsheet where the four metrics live each week

That's a 15-minute weekly update once the initial setup is done. Compared to the amount of time you already spend staring at ROAS numbers that don't mean what you think they mean, it's a lower total time investment with dramatically better signal.

This is part of the gap Arlo was built to close. We pull your Shopify data weekly and surface new customer count, new customer revenue, repeat rate, and cohort behaviour alongside contribution-margin-aware context, so the numbers that actually determine your acquisition economics are the ones in front of you. The weekly report doesn't replace Meta Ads Manager (you still need that for creative, audience, and campaign management) but it does replace the broken habit of using blended ROAS as the primary signal for whether paid is working.

You still make the budget calls. We just make sure the calls are based on the right numbers.

#Frequently Asked Questions

Is new customer CAC the same as CPA in Meta Ads Manager?

Not quite. Meta's CPA counts any purchase attributed to Meta by Meta's attribution model, which includes repeat customers who clicked on a Meta ad. True new customer CAC counts only first-time purchasers attributed to Meta. The second number is always higher, sometimes significantly.

What's a healthy new customer CAC for a DTC brand?

There's no universal number. It depends entirely on your contribution margin and payback tolerance. A brand with $80 AOV and 40% contribution margin can afford a higher CAC than a brand with $35 AOV and 30% contribution margin. The right question isn't "is my CAC low" but "is my CAC below my contribution-margin-derived ceiling."

How often should I recalculate contribution margin?

Quarterly is plenty for most brands. The number moves when COGS, shipping costs, payment processing, or return rates change meaningfully. If any of those shift, recalculate.

What if my repeat rate is strong enough that I don't need to worry about first-order economics?

Even strong repeat brands need to know first-order economics, because repeat rate can change without warning and a brand financing aggressive acquisition on predicted repeat behaviour can get caught out when cohorts start behaving differently. Know the number even if you're comfortable running a loss on first order.

Can I use Shopify's native reports for all of this?

Mostly yes, with some caveats. Shopify reports new customer count and new customer revenue cleanly. It doesn't pull Meta spend automatically, so you'll need to combine Shopify data with Meta data manually or through a tool. Contribution margin isn't native to Shopify either; you'll need to calculate it separately.

How do I handle seasonality in CAC?

Compare CAC week-over-week against the same period last year, not against the prior week. Q4 CAC for a gifting brand will always be different from Q2 CAC. Year-over-year comparisons correct for seasonality better than week-over-week does.

Should I kill Meta if the numbers don't work?

Probably not. The more likely answer is that you need to adjust targeting, creative, budget allocation, or offer structure to get the numbers to work. Meta can be made to work for most DTC brands with the right setup. The question is whether your current Meta operation is tuned to the right metrics. Usually it isn't, and the fix is tuning, not cutting.

What's the single most important change I can make this week?

Add a post-purchase survey. One question, three to five answer options. "Where did you first hear about us?" The gap between what customers say and what platforms report is the single most revealing number in DTC marketing, and most brands don't have it.

If you want your Shopify data summarised weekly with new customer economics front and centre instead of blended vanity metrics, that's what Arlo does. The weekly report surfaces what's actually growing and what's being carried by repeat behaviour, so you can make Meta budget decisions against the right signal. You still run the ads. We just make sure the numbers in front of you reflect the business, not the dashboard.