Your ROAS can improve while the business gets worse
A ROAS chart can rise while the business gets weaker.
That sentence sounds dramatic until you have watched it happen. The dashboard is green. The finance conversation is not. Everyone is looking at accurate numbers that point at different truths.
ROAS asks what revenue the platform associates with ad spend. The business asks what cash and contribution remained after discounts, costs, returns, and customers who were coming anyway.
Those are not the same question.
When the ratio improves for the wrong reasons
There are several ordinary ways for ROAS to look healthier while the operation does not.
Discounts. A promotion can lift conversion rate and attributed revenue enough to improve ROAS while contribution margin falls. You bought efficiency on revenue and paid for it with profit.
Repeat customers in the attribution window. If the platform credits ads for purchases that loyalty, habit, or email would have produced, ROAS rises without incremental growth. The account is harvesting demand it did not create.
Product mix. Ads can over-index on lower-margin SKUs that convert easily. Revenue efficiency improves. The P&L quietly softens.
Blended comfort. Prospecting can deteriorate while branded and remarketing hold the account-level ROAS together. The average looks stable. The growth engine is not.
None of these require incompetence. They require a team that stopped at the platform ratio.
“Is it making you money?” is still the right question
I keep returning to a plain question: is it making you money?
Not “is the ROAS above target?” Not “is the CPA inside the benchmark?” Is the business better off because this spend happened?
Answering that requires incrementality thinking, even when you cannot run a perfect holdout every month.
Incrementality asks what would have occurred without the ads. Platform attribution asks what events occurred after exposure or a click within a chosen model. Better tracking improves the second. It does not automatically answer the first.
That distinction is where financial credibility starts.
Dashboard logic versus business logic
Dashboard logic says: spend $20,000, attribute $80,000 in revenue, celebrate 4x.
Business logic asks whether the business kept more money.
Take two consecutive months with the same $20,000 media spend. Keep three ideas apart: total contribution is what the business kept on attributed revenue after costs and ads; attributed contribution is the slice of that story the platform or a segmentation rule assigns to a group such as new customers; experimentally measured incrementality is what a designed test shows would not have happened without the ads. The new-customer math below is an attributed contribution proxy. It does not measure incrementality.
Month 1:
- Attributed revenue: $80,000
- Reported ROAS: 4.0x
- Discounts: $4,000
- Returns: $2,000
- COGS and variable fulfillment: $40,000
- Contribution before ads: $80,000 − $4,000 − $2,000 − $40,000 = $34,000
- Contribution after ads (total contribution on attributed revenue): $34,000 − $20,000 = $14,000
- Existing-customer share of attributed revenue: 30%
- New-customer share of pre-ad contribution (attributed contribution proxy): 70% × $34,000 = $23,800
- New-customer-attributed contribution proxy after ads: $23,800 − $20,000 = $3,800
Month 2, after a broader sale and heavier remarketing:
- Attributed revenue: $100,000
- Reported ROAS: 5.0x
- Discounts: $15,000
- Returns: $5,000
- COGS and variable fulfillment: $48,000
- Contribution before ads: $100,000 − $15,000 − $5,000 − $48,000 = $32,000
- Contribution after ads (total contribution on attributed revenue): $32,000 − $20,000 = $12,000
- Existing-customer share of attributed revenue: 55%
- New-customer share of pre-ad contribution (attributed contribution proxy): 45% × $32,000 = $14,400
- New-customer-attributed contribution proxy after ads: $14,400 − $20,000 = −$5,600
ROAS improved from 4.0x to 5.0x. Total contribution after ads fell from $14,000 to $12,000. The new-customer-attributed contribution proxy after ads flipped from +$3,800 to −$5,600 because more of the attributed revenue came from existing customers, discounts widened, and returns rose. That proxy still does not measure incrementality. The dashboard got greener while cash and contribution got weaker.
That is the gap in one place. Business logic still has to ask:
- How much of that revenue was incremental?
- What contribution margin remained after discounts and variable costs?
- How much cash was tied up before payback?
- Did we acquire new customers, or mainly accelerate existing ones?
- Are we scaling the motion that created growth, or the motion that reports well?
I have seen teams optimize themselves into a beautiful remarketing ROAS while new-customer acquisition quietly collapsed. The account looked efficient. The brand was eating its seed corn.
A short diagnostic you can run without new software
Take last month’s strongest ROAS campaign and ask:
- What share of its conversions came from existing customers?
- What was the average discount rate in those orders?
- What was contribution margin on the top attributed products?
- What happened to new-customer count while ROAS improved?
- If we cut this campaign by half, what do we believe we would truly lose?
If the team cannot answer, ROAS is being asked to carry more meaning than it contains.
This is also why I care about unified data and first-party measurement. Platform self-attribution is one evidence source. It should not be the only one a serious business trusts. When CRM outcomes, order margins, and media delivery live in separate conversations, the ROAS meeting and the cash meeting will keep disagreeing. The point is not a prettier dashboard. The point is a shared set of facts closer to the business.
What to optimize instead of worshipping the ratio
I am not anti-ROAS. Used carefully, it is a useful efficiency signal inside a channel.
I am against letting it become the definition of success.
Better paired metrics usually include new-customer CAC, contribution after discounts, payback, repeat rate by acquisition cohort, and some read on incrementality where practical. Those metrics are harder to game with a weekend sale.
They also force a healthier conversation between marketing and finance. Marketing stops defending a ratio. Finance stops treating all media as undifferentiated cost. Both sides look at whether growth is creating value.
The memory I want this essay to leave is simple.
A rising ROAS is not the same thing as a stronger business. Before you scale the chart, ask what the business kept.