agentcocoon

Guides

Blended vs new-customer ROAS: which should an eCommerce brand track?

By Drew Amato · Published

Track both, but make spending decisions on new-customer ROAS and payback. Blended ROAS divides all sales by all ad spend, so repeat buyers and subscription renewals make it look strong even when ads bring in nobody new. New-customer ROAS counts only first orders. For our brand, July to October, blended ROAS was 19.1 and new-customer ROAS was about 1.25 on the same spend.

For July through October 2026, our golf apparel brand spent $23,320 on ads and booked $445,392 in sales. That is a blended ROAS of 19.1. Sales from first-time buyers were $29,081, a new-customer ROAS of about 1.25. Same months, same spend, a fifteen-fold gap.

This guide covers what each number measures, what four months of our books showed, and how our ad agent reports them now.

What is the difference between blended and new-customer ROAS?

Blended ROAS is all sales divided by all ad spend. New-customer ROAS is sales from first orders divided by the same ad spend. Both ignore platform attribution, which is the point. They come from your books, not from Meta or Google.

Metric Formula What it tells you
Blended ROAS All sales / all ad spend Whether the business covers its ad bill
New-customer ROAS First-order sales / all ad spend Whether ads are buying new customers at a sane price
Cost per new customer All ad spend / new customers What one new customer costs you

The platforms recognize the difference too. Google Ads can report new and returning customers separately, and counts someone as new if they have not purchased in the last 540 days when it detects this on its own (Google Ads Help).

What did our numbers show month by month?

They showed blended ROAS moving opposite to new customers. These are from our bookkeeping system, read live this month:

Month Ad spend All sales Blended ROAS New-customer sales New-customer ROAS New customers Cost per new customer
July $5,399 $123,208 22.8 $11,162 2.07 46 $117
August $5,050 $97,218 19.3 $5,854 1.16 27 $187
September $2,284 $111,225 48.7 $2,413 1.06 12 $190
October $10,586 $113,741 10.7 $9,601 0.91 59 $179

Look at September and October. Spend went up more than four times, total sales rose about 2%, and blended ROAS fell from 48.7 to 10.7. On a blended dashboard, October looks like the worst month. It brought in 59 new customers against 12 the month before.

September looks like the best month for the opposite reason. Spend was low, and sales held up because subscribers renewed regardless. Blended ROAS rewarded the month when ads did the least.

Across the four months, 145 of 1,973 orders came from new customers, about 7%. The other 93% were orders from people who had bought before, and every one of them counts toward blended ROAS.

Why does blended ROAS mislead subscription and repeat brands?

Because most of the numerator was never up for grabs. Renewals, replenishment orders and loyal regulars count as sales against ad spend even when no ad touched them. The higher your repeat share, the more blended ROAS measures loyalty instead of advertising.

The platforms lean the same way. Meta's own reporting credited our subscription renewals to ads, which we covered in Can an AI agent manage Meta ads for a small eCommerce brand?. Meta used to offer a cap on spend aimed at existing customers in its automated sales campaigns. That setting is no longer available, and Meta now tells advertisers to rebuild it with a custom audience exclusion or a separate ad set with a spending limit (Meta Business Help Center). If you do nothing, past buyers stay in the pool.

Is a new-customer ROAS below 1 always bad?

No. It means the first order did not pay for the ad. Whether that is a problem depends on what the customer does next. Our books follow each month's new customers as a cohort:

  1. July cohort. 46 new customers at $117 each. Gross profit per customer, after product cost and fees, was $96 in the first month and $176 by month three. That is about 1.5 times what they cost.
  2. August and September cohorts. Cost per new customer near $190. Gross profit per customer was $84 by month two for August and $89 by month one for September. Not paid back yet.
  3. October cohort. 59 new customers at $110 each, $120 of gross profit in the first month. Already ahead.

Profit is the stricter test. New-customer profit before ads came to $0.65 per ad dollar across the four months, so first orders alone lost money. The July cohort shows how a subscription brand earns it back when renewals arrive. The August and September cohorts show it does not always happen on schedule.

How should you set this up?

Report both every month, and make decisions on the new-customer side. This is the check that produced the tables above:

  1. Pull spend from the books, not the platforms. One total across Meta, Google and anything else.
  2. Split sales by first order versus repeat. Most stores can tag this at the order level. Google's conversion tag also accepts a new-customer flag if you want its reports to match (Google Ads Help).
  3. Show blended ROAS, new-customer ROAS and cost per new customer side by side. Never one alone.
  4. Track each monthly cohort's payback. Gross profit per customer at months one, two and three, against what it cost.
  5. Flag big swings for a person. The agent drafts and explains. A person decides budget.

One caveat from our own setup. Our ad accounts were not yet connected to the books, so spend in the table comes from payments and is dated by when we paid, not by when ads ran. That can shift dollars between adjacent months. Connect the ad accounts if you can; our bookkeeping guide covers why.

Which number should go on the dashboard?

Put new-customer ROAS and cost per new customer at the top, with blended ROAS underneath as context. Blended still answers a real question: can the business carry its ad bill? It just cannot tell you whether the ads are working. For a brand where most revenue repeats, that is the question that decides next month's budget.

Common questions

What is a good new-customer ROAS?

It depends on margin and repeat rate, not on a universal number. A first order that loses money can still be a good buy if customers come back. Work it out from gross profit per new customer over three to six months versus what you paid to acquire them, then set the target from that.

Is blended ROAS the same as MER?

Close. Both divide total sales by total spend instead of trusting platform attribution. MER is often calculated on all marketing spend, while blended ROAS usually uses ad spend only. Either way, neither one separates new customers from repeat ones, which is the gap this guide is about.

Why is the ROAS in Meta different from both?

Meta reports purchases its ads touched inside its attribution window, which can include customers who were going to buy anyway. Your books count actual orders. Neither is wrong, but only your books can tell you whether the buyer was new, so use them to check the platform.

How often should you check new-customer ROAS?

Monthly for decisions, weekly as a warning light. Small brands get few new customers per week, so a weekly number swings a lot. We read it by calendar month, next to cost per new customer and the payback of each monthly cohort.

Sources

  1. Google Ads Help: Measure your lifecycle goals campaigns
  2. Google Ads Help: Set up new customer acquisition parameter in your conversion tracking tag
  3. Meta Business Help Center: How to replicate existing customer budget cap

About the author. Drew Amato is the founder of Mully, a golf apparel brand in Detroit, and of AgentCocoon. He builds and runs the AI agents behind Mully's own operations. Follow on X.