Why ROAS is a useful but dangerous metric

ROAS measures how efficient a campaign is in the short term. But when a client asks me to cut the budget "because we're above break-even," I show them why the ROAS threshold is a trap.

LTV/CAC is the right framework: a customer who buys three times has a real value far above their first ROAS. Optimizing for immediate break-even means burning your best customers.

When cutting at 2x ROAS is the most expensive move you can make

A brand in the beauty sector, annual revenue around €800k, scales Meta Ads up to €15,000/month in ad spend. Blended ROAS 2.1x. The CFO decides it's too low and cuts the budget by 60%.

What they hadn't accounted for: that campaign was acquiring customers with an average repurchase frequency of 2.8 times in 12 months. Average order value (AOV) €67. Average 12-month lifetime value (LTV): €188. Customer acquisition cost (CAC): €31. Payback period: less than 5 weeks.

A ROAS of 2.1x looked insufficient. An LTV/CAC of 6x was extraordinary.

After the cut, new customer acquisition stopped. Revenue from existing customers stayed flat for 90 days, then started shrinking because the pool was running dry. That cohort — acquired at "low" ROAS — was the most profitable of the entire year.

A fixed-threshold ROAS doesn't tell a high-LTV cohort from a low-LTV one. It cuts indiscriminately. And the best cohorts often have the lowest initial ROAS, because premium customers compare, think about it, come in from longer paths.

The LTV/CAC framework: how to build your real ROAS target

You don't invent the ROAS target. You calculate it backwards from your cohort data.

Base formula: target ROAS = Average price / (LTV × gross margin × (1 / acceptable payback period))

In practice, this is how I work with every client before touching the campaigns:

Pull the monthly cohorts from the last 18-24 months. Calculate 6-month LTV and 12-month LTV per cohort.
Identify the average gross margin per product category (not the net margin, don't mix them up).
Define the acceptable payback period — usually 60-90 days for e-commerce brands with good liquidity, up to 180 for premium brands.
Calculate the maximum sustainable CAC: max CAC = LTV × margin × (target payback / 365).
Translate it into ROAS: minimum ROAS = AOV / max CAC.

A brand with a 12-month LTV of €210, 58% margin, acceptable payback 90 days: max CAC = 210 × 0.58 × (90/365) = €30.06. With an AOV of €85, minimum ROAS = 85/30 = 2.83x.

Not 4x. Not 5x. 2.83x is the correct number for this business. Everything above that threshold is excess efficiency — and you usually get it by sacrificing volume on high-LTV cohorts.

In the premium food case study the final blended ROAS was 6.2x — but that number was the result of a strategy built on LTV, not the starting point to optimize.

The mechanism that burns your best customers

Meta's algorithm optimizes for conversions. If you optimize for high immediate ROAS, you're telling the algorithm to go find the people who buy fast and with minimal discounts.

The problem: the people who buy fast are often the ones who respond to promotions, who are less loyal to the brand, who always compare prices. They're low-LTV customers.

High-LTV customers — the ones who'll come back three times, who become ambassadors, who bring referrals — usually take longer before the first purchase. They need more touchpoints. They cost more to acquire. They generate a lower initial ROAS.

When you optimize for a high ROAS threshold, you're literally training Meta to ignore your best customers.

The signal that this is happening: ROAS going up, revenue growing, but the percentage of repeat customers going down. You're buying low-LTV volume. Before long the business starts depending entirely on continuous acquisition — take the budget away and revenue stops in 30-60 days.

The 3 metrics that matter more than first-order ROAS

After years of campaigns, the metrics I keep an eye on every week are three:

1. LTV/CAC per acquisition cohort. Not the blended ratio — the one per monthly cohort. I want to see whether the quality of the customers we're acquiring is getting better or worse over time. An LTV/CAC that drops month over month is a warning sign regardless of ROAS.

2. Median payback period. How long does a cohort take to pay back the CAC? If it goes from 45 to 90 days, something has changed — product, offer, seasonality, or we're acquiring customers with lower LTV.

3. Percentage of revenue from repeat customers (30/60/90 days). A healthy e-commerce brand gets at least 25-35% of monthly revenue from customers it already acquired. If that percentage drops, you're burning the base.

First-order ROAS stays useful as an indicator of a campaign's operational efficiency. Not as a decision metric for budgets.

How to present LTV/CAC to a client who only wants to hear "ROAS"

The client who only asks about ROAS isn't stupid. They don't have the data to think about it any other way. Your job is to build the context.

The presentation I use: I take the last 4-6 monthly cohorts, show the initial ROAS of each one and then the 6-month LTV. Usually the cohorts with the lowest initial ROAS have the highest LTV. That alone changes the conversation.

Then I show the calculation of the real target ROAS for their business — the one derived from LTV, not a generic industry benchmark. When the client sees that their correct target ROAS is 2.6x and not 4x, they get that they were optimizing in the wrong direction.

The right question to ask is: "Would you rather have a 4x ROAS that brings in customers who never come back, or a 2.8x ROAS that brings in customers who buy three times a year?" I've never had a client answer that question badly when it's put to them with the numbers in hand.