Menu
  1. Services
  2. Case studies
  3. Method
  4. Pricing
  5. About
  6. Contact

More pages

Get my free audit

Blog · Guide

ROAS, CPA or CPL: which metric to watch for your business

By Published 5 min read

Short answer

If you sell online, watch ROAS. If you collect leads, watch CPL and how many of those leads actually buy. None of it means much without your margin: use it to get the minimum ROAS you need to stay profitable, then check it against your real sales, not just what the platform says.

What does each metric actually measure?

All three answer the same question from different angles: how much your ad spend costs you, or how much it brings back. If any of these acronyms are new to you, they’re all in the glossary.

Metric How it’s calculated What it tells you
ROAS attributed revenue ÷ ad spend Revenue for every unit you spend
CPA ad spend ÷ sales or customers won What you pay for each sale
CPL ad spend ÷ leads (forms, calls, messages) What you pay for each contact
MER total revenue ÷ total ad spend across platforms How efficient the whole business is
LTV what a customer is worth over the whole relationship How much you can pay for the first order

Higher ROAS is better: a ROAS of 5 means 5 in revenue for every 1 you spent. CPA and CPL work the other way around. Lower is better.

There’s a catch hiding in the word “revenue.” ROAS tracks sales, not profit. A high ROAS on a thin margin can still lose you money, which is why we’ll work out your minimum further down.

Which one should you watch: online store, leads or messages?

Online stores. Your main metric is ROAS, because the platform sees the value of every purchase. I always pair it with CPA and average order value. A ROAS that climbs because you sold a few pricey items behaves very differently from one that climbs because you sold more.

Service businesses and B2B companies. Here the sale almost never happens on your site. The platform sees the form fill or the call, and your team closes days or weeks later. CPL is fine for daily decisions, but it can fool you: a cheap lead that never buys is an expensive lead. What really matters is cost per qualified lead and, if you connect your CRM data, cost per closed customer.

Businesses that sell through messaging. When the ad opens WhatsApp, the direct metric is cost per conversation. Same as with leads, the next question is how many of those chats turn into a sale.

  1. 1Ad spend
  2. 2Click
  3. 3Lead
  4. 4Qualified lead
  5. 5Customer

I use the same split in the accounts I manage. For e-commerce I report revenue and ROAS; for leads and messaging, cost per lead or per conversation. You can see it client by client in my case studies.

46.29X

Highest ROAS across my case studies (hotel chain, Argentina)

Case 02 · jeanmorillo.com

Why do Meta and Google say you sold more than what hit your bank account?

Because each platform counts by its own rules, and neither one sees what the other saw. There are four reasons, and it’s worth knowing them before you argue with a number.

1. Attribution windows. With the standard setting, Meta counts conversions that happen up to 1 or 7 days after someone clicks your ad and up to 1 day after they view it, per Meta’s attribution docs. Google Ads uses a 30-day click window by default, which you can change, per its help page on conversion windows.

2. Both claim the same sale. If someone clicks a Meta ad, then a Google ad a few days later, and then buys, each platform can count that purchase as its own. Your store counts it once. Add up what each platform reports and you’ll get more than what actually came in.

3. The date of the sale. Google Ads logs the conversion on the date the ad was shown, while other tools log it on the day of the purchase, as Google explains in its guide to data discrepancies. That’s why last week’s ROAS can keep moving for a few days.

4. Estimated conversions. When it can’t observe a conversion, say because the user declined cookies or uses a browser that limits them, Google reports modeled conversions. Meta does something similar and covers it on its page about modeled conversions. They’re reasonable estimates, but they’re estimates.

Here’s how I reconcile them: I use each platform’s ROAS to make decisions inside that platform (which ad to scale, which ad set to kill) and MER, built on your store’s real revenue, to know whether the business as a whole is making money.

How do I figure out my break-even ROAS?

Break-even ROAS is the point where the sales you get from ads cover exactly the cost of the product and the ads. Below it you lose money; above it you make money. The formula is short:

Break-even ROAS = 1 ÷ margin (margin as a decimal, after product cost, shipping and fees).

And its cousin for CPA: max CPA = average order value × margin.

Here’s an example with made-up numbers, just to show the math:

Illustrative example Value
Average order value 100
Product cost, shipping and fees 60
Margin (100 − 60) ÷ 100 0.40
Break-even ROAS: 1 ÷ 0.40 2.5
Max CPA: 100 × 0.40 40

With those numbers, any sale that costs you more than 40 in ads loses money on that first order. Notice the “first order” part. If your customers come back and buy again, LTV lets you accept a lower ROAS up front, as long as you have real repeat-purchase data and not a hunch.

Leads work the same way, with one extra step. Your max CPL is what a customer is worth to you times the share of leads you close. If you keep 400 per customer and close 1 out of every 10 leads, each lead can’t cost you more than 40.

What should I do with this starting Monday?

  1. Pick one main metric. ROAS if you sell on your site; cost per qualified lead or per closed customer if you sell services; cost per conversation if you sell through messaging.
  2. Work out your break-even. Use your real margin, not the one you wish you had.
  3. Compare every week. What each platform reports versus what actually landed in your store or CRM. If the gap suddenly widens, something changed in your tracking.
  4. Check tracking before you touch the campaign. A pixel that double-counts purchases or a form that doesn’t fire its event throws off every metric above.

Jean's tip

Before you ask for a higher ROAS, ask what minimum your business actually needs. A lot of the time the account was already profitable and nobody knew it. Or the opposite: everyone was celebrating a number that was losing money.

If you want me to look at your numbers this way, you can request a free audit of your account. And if you’re curious how I handle tracking from day one, it’s all in my method.

FAQ

What's a good ROAS?

Any ROAS above your break-even ROAS. You get that number from your margin: 1 divided by your margin as a decimal. A store with fat margins can make money at a ROAS that would sink a store with thin ones, so there's no single good ROAS for everyone.

Why doesn't my Meta ROAS match my store's sales?

Meta counts sales inside its attribution window, can estimate conversions it can't observe, and has no idea whether Google also claimed the same sale. Your store counts each sale once. That's why it helps to also track MER: total revenue divided by total ad spend.

Should I track CPA or CPL?

It depends on what you sell. If customers buy on your site, use CPA and ROAS. If they leave their info and your team closes the deal later, CPL works for day-to-day calls, but the real decision should be based on cost per closed customer.

What is MER and when should I use it?

MER divides all your revenue by everything you spend on ads, across every platform. It tells you whether the whole business is profitable without fighting over which channel gets credit for each sale. Use it alongside each platform's ROAS, not instead of it.

Sources

  1. About attribution models and attribution settings, Meta Business Help Center
  2. About Meta's modeled conversions, Meta Business Help Center
  3. About conversion windows, Google Ads Help
  4. About modeled online conversions, Google Ads Help
  5. Data discrepancies: Factors and troubleshooting, Google Ads Help

Want this working in your account?

I start with a free audit: I review your account and your tracking, and tell you what I'd change first.

Get my free auditSee: The free audit

Keep reading