The migration of return on advertising spend from a useful measurement to a key strategic KPI has occurred largely without critical examination. High ROAS doesn’t always equal good – a reality that’s becoming clearer as digital advertising costs rise and competition intensifies.
This addresses the industry-wide problem of low incrementality, defined as the “diminishing returns” that occur when scaling marketing budgets. ROAS can be gamed by targeting existing customers, brand-loyal segments, or those who would have purchased anyway, creating the illusion of efficiency while delivering minimal business growth.
Visualising the problem
One of the most shocking exercises for any marketing team is plotting monthly spend against new customer acquisition. Many organisations boasting “high” ROAS discover weak correlations between spend increases and customer growth.
A real example: a £100m+ brand showed virtually no correlation (0.22) between marketing spend and new customer orders over two years. ROAS was up year-on-year, but new customer orders and overall revenue were down. The deity of ROAS was satisfied, but the business wasn’t growing.
A correlation of 0.22 means only about 5% of variation in new customer acquisition is explained by marketing spend changes – suggesting most customer acquisition happens through other mechanisms.
Why ROAS misleads
The fundamental issue lies in who these campaigns reach. Research by John Dawes and Byron Sharp found that extremely light buyers contribute about 40% of a brand’s total sales. Yet focusing spend on heavy repeat buyers may boost short-term ROAS while limiting ability to attract the new or lapsed buyers who fuel long-term success.
A campaign laser-focused on ROAS can win the battle and lose the war – yielding impressive returns now while undermining future growth.
The overreliance problem
Although it is surely the future, one of the more damaging trends in 2024 and into 2025 has been the overreliance on Meta’s Advantage+ campaigns. The promise of automation and scale is attractive, with Meta reporting 17% lower CPA and 30% higher ROAS compared to standard campaigns.
But the machine learning is laser-focused on hitting the highest immediate ROAS by serving ads to people easiest to convert. This produces fantastic in-platform ROAS figures, but without sufficient thought about what’s actually happening, overuse can be dangerous for incrementality.
Advantage+ is a vital tool for mid and bottom-funnel activity. My theory is that AI tools deliver exceptional results for early adopters, but as mass data saturates the system, performance inevitably declines, and an arbitrage opportunity opens up with manual campaigns.
A better approach
Success means hitting the profit number for the year, not the ROAS number that week. The relationship between weekly ROAS and annual profit growth has broken down as CPMs have risen due to increased competition.
The solution isn’t the blanket industry advice of “spend 60% on brand whatever the cost”. We need nuance. Currently, I recommend setting aside 10-15% of Meta budget for top-of-funnel campaigns. The aim is reach: first-time impressions, low audience overlap, and no expectation of immediate return.
Measuring what matters
If you’re not running holdout tests, you’re not measuring performance. Make incrementality part of the roadmap, not an afterthought. The goal isn’t to make metrics look good – it’s to grow the business.
Track the correlation between overall spend and new customer revenue over time. This metric deserves the same attention as CAC, ROAS, or contribution margin.
The diversification solution
The real antidote to diminishing returns is diversification across funnel, channels, and creative:
- Funnel diversification: Balance upper-funnel awareness, mid-funnel consideration, and lower-funnel conversion activation.
- Channel diversification: Reduces pressure on individual channels and makes scaling easier.
- Creative diversification: Gives algorithms liquidity and helps expand reach.
But beware spreading budgets too thinly, especially when moving to advertising channels like TV and OOH. These require sufficient frequency to achieve mental availability – reach without frequency is unlikely to influence consumers and drive revenue.
The incrementality mindset
Driving incrementality isn’t a tactic to deploy occasionally – it’s a principle to operationalise across planning, measurement, and execution. Without rigour around correlation tracking, proper holdout testing, and strategic diversification, it’s easy to drift into the ROAS illusion. The pain won’t come all at once but will chip away slowly.
Jake Higgins is a partner at MNC
