Introduction

Marketing budgets are finite, and how they’re allocated can make or break a campaign. Yet many decisions rely on ‘linear thinking’—assuming past ROAS guarantees future success—leading to inefficient spend allocation.

In this blog post, I will walk through a simplified scenario analysis to show how decision-makers can evaluate alternatives and identify the most optimal allocation of marketing spend. To set the stage, I will also revisit the concept of saturation curves and highlight how they provide invaluable insights into the efficiency of incremental spend in driving additional revenue.

ROAS Measures Historical Efficiency, Not Future Growth Potential

ROAS—the amount of revenue earned for every dollar spent on a campaign—is widely used to evaluate the performance of marketing channels. It is a measure of historical performance efficiency, calculated as:

While a high ROAS reflects strong past performance, it is important to recognize that ROAS is an average (i.e. mean). Averages are inherently simplistic—they smooth out variability in the data and often fail to account for important dynamics (sometimes called the “flaw of averages”), such as diminishing returns. In fact, averages almost always perform poorly as predictive models because they fail to reflect the complexities of real-world scenarios, leading to high error rates.

Importantly, ROAS as a static metric provides us no information about how additional, new spend will perform. A channel with high ROAS may already be saturated, meaning its incremental returns—the additional revenue generated for every additional dollar spent—are minimal. Conversely, a channel with lower ROAS may still be in its growth phase, offering greater incremental returns and making it a better candidate for further investment.

This distinction is critical: decisions based solely on historical ROAS performance often lead to overinvestment in saturated channels and underinvestment in high-potential channels. Without considering incremental returns—sometimes called incremental ROAS—decision-makers risk allocating budgets inefficiently and missing out on significant growth opportunities.

The Problem with Linear Thinking

Linear thinking — as illustrated in the plot below—assumes that performance remains constant as spend increases, indefinitely. This approach often leads decision-makers to prioritize channels with the highest historical/average ROAS, assuming they will always deliver the best returns.

However, linear thinking is flawed for several reasons:

1. Infinite Returns Fallacy:

• If channels behaved linearly, one channel would always dominate (i.e. would have the steepest slope and, therefore, highest returns), and budget could scale infinitely as long as ROAS exceeded 1.
• This would imply unlimited returns, ignoring real-world constraints like finite audiences and demand.

2. Performance is not static:

• Channels exhibit non-linear growth, where returns diminish as spend increases.
• Factors like audience saturation, ad fatigue, and competition limit scalability.

Understanding these dynamics ensures budgets are allocated based on a channel’s growth projection rather than its historical ROAS.

Illustration: Understanding Incremental ROAS (Rise/Run)

The chart below visually demonstrates the incremental revenue generated as additional marketing spend is applied. By breaking down incremental spend and revenue gains at different points along the curve, it reveals how efficiency declines as the channel approaches saturation, emphasizing the importance of optimizing incremental rather than average performance.

The Δ (Greek letter delta) symbol means “change in.” So:

• Δ Return = change in revenue or return (e.g., the difference in revenue when spend increases).
• Δ Spend = change in marketing spend (e.g., the additional dollars spent).

For example, for the first “chunk” of spend, an increase of $2,000 generates $58,000 in incremental revenue, yielding an impressive incremental ROAS of 29x. This means for every dollar spent, $29 in revenue is gained—an extremely efficient investment.

As spending increases by another $3,000, incremental revenue drops to $26,000, resulting in a smaller but still strong incremental ROAS of 8.7x. This indicates profitability, but the efficiency is starting to decline.

Finally, when the spend increases by another $3,000, the incremental revenue is just $6,000, leading to an incremental ROAS of 2x. While the total revenue still increases, this small return highlights the diminishing returns as the channel approaches saturation.

This illustration underscores the importance of looking at incremental ROAS rather than static averages to ensure optimal budget allocation and avoid overspending into inefficient zones.

Common Saturation Curves

To move beyond linear thinking, it’s essential to understand how saturation curves impact marketing performance. Marketing channels usually follow non-linear growth patterns like convex curves (i.e. exponential) or S-curves.

• Convex Curves: these curves exhibit steep growth initially but plateau quickly as returns diminish. Convex curves are ideal for short-term efficiency or small budgets, as they provide strong early returns but reach saturation quickly.
• S-Curves: these curves start with slow growth, accelerate around an inflection point, and eventually saturate at higher spend levels. S-curves are best suited for long-term growth and larger budgets, as they offer significant potential during the mid-spend phase.

Scenario Analysis: Why ROAS Alone Is Misleading

To illustrate the limitations of relying on a historical ROAS metric, let’s consider a simple two-channel scenario:

• Channel 1: a convex curve that saturates at approximately 100 units of return, with diminishing returns starting around 30 spend units.
• Channel 2: an S-curve that saturates at approximately 120 units of return, with growth accelerating after 10 spend units but flattening closer to 25+ spend units.

The chart above shows the underlying data that was used to fit the saturation curve (note: I give a high level summary of the fitting procedure here). For our purposes, we will remove the underlying data and focus on the fitted curves for each channel, like so:

SCENARIO 1: at an initial spend of 15 for both channels:

• Channel 1: produces a return of 77.69 with an Average ROAS of 5.18.
• Channel 2: produces a return of 9.10 with an Average ROAS of 0.61.
• Total Revenue: 86.79

Here is a quick breakdown: Channel 1 outperforms Channel 2 primarily due to its higher cumulative return and greater efficiency at the same spend level. This discrepancy arises from the fundamental difference in their saturation behaviours:

• Channel 1: exhibits a steep, convex curve, delivering strong returns early and reaching diminishing returns more gradually. This makes it highly efficient at lower and moderate spend levels.
• Channel 2: shows an S-curve with delayed saturation. While its potential for growth accelerates at higher spend levels, it lags significantly at lower levels, where incremental returns are minimal.

The visualization above highlights how Channel 1’s efficiency dominates at moderate spending levels, while Channel 2 requires additional investment (i.e. has slower “activation” period) to unlock its growth potential.

At first glance, however—if we did not fit a saturation curve to our channel spend data and run this analysis—it would seem logical to anyone that the best course of action is to allocate more budget to Channel 1 due to its higher average ROAS. However, when incremental returns are analyzed, a different picture emerges.

Results from Alternate Spend Allocations

Average ROAS (Scenario 1) looks backward at overall efficiency for the total spend. Incremental ROAS (Scenarios 2 & 3) looks forward to predict how each additional dollar spent (or shifted between channels) will perform, providing a more dynamic view for decision-making.

SCENARIO 2: increase spend on Channel 1 by 3 units (to 18) and decrease spend on Channel 2 by 3 units (to 12).

• Channel 1: produces a return of 83.47 with an Average ROAS of 4.64
• Channel 2: produces a return of 2.16 with an Average ROAS of 0.18
• Total Revenue: 85.63
• Incremental Revenue: (85.63 – 86.79) = -1.16 (a loss compared to Scenario 1)

Incremental ROAS: to calculate the incremental ROAS from Scenario 1 to Scenario 2 for both channels, we apply the formula:

Channel 1:

• Scenario 1: Spend = 15, Return = 77.69
• Scenario 2: Spend = 18, Return = 83.47
• Change in Spend (Δ Spend): 18 – 15 = 3
• Change in Return (ΔReturn): 83.47 – 77.69 = 5.78

Interpretation: for every additional dollar spent on Channel 1 between 15 and 18 spend units, you receive $1.93 in additional return.

Channel 2:

• Scenario 1: Spend = 15, Return = 9.1
• Scenario 2: Spend = 12, Return = 2.16
• Change in Spend (Δ Spend): 12 – 15 = -3 (a decrease)
• Change in Return (Δ Return): 2.16 – 9.10 = -6.94

Interpretation: for every dollar reduced from Channel 2 between 15 and 12 spend units, you sacrifice $2.31 in return.

Analysis: these results illustrate that shifting budget away from Channel 2 is costly, as it was in its growth phase, while Channel 1 delivers moderate incremental efficiency gains as it nears saturation. Although Channel 1 generates positive incremental ROAS, its additional returns are not enough to offset the lost returns from Channel 2 (with its higher incremental ROAS). In this case, Scenario 2’s total strategy (spending more on Channel 1 and less on Channel 2) results in less total revenue, even though Channel 1’s incremental returns are positive. This reveals that Channel 2’s growth potential (higher incremental ROAS) was undervalued in this reallocation.

SCENARIO 3: decrease spend on Channel 1 by 3 units (to 12) and increase spend on Channel 2 by 3 units (to 18).

• Channel 1: produces a return of 69.88 with an Average ROAS of 5.82
• Channel 2: produces a return of 32.27 with an Average ROAS of 1.79
• Total Revenue: 102.15
• Incremental Revenue: (102.15 – 86.79) = 15.36 (a gain compared to Scenario 1)

Incremental ROAS: to calculate the incremental ROAS from Scenario 1 to Scenario 3 for both channels, we again apply the formula:

Channel 1:

• Scenario 1: Spend = 15, Return = 77.69
• Scenario 3: Spend = 12, Return = 69.88
• Change in Spend (ΔSpend):  12 – 15 = -3  (a decrease)
• Change in Return (ΔReturn):  69.88 – 77.69 = -7.81 

Interpretation: reducing spend in Channel 1 by $1 decreases returns by $2.60.

Channel 2:

• Scenario 1: Spend = 15, Return = 9.10
• Scenario 3: Spend = 18, Return = 32.27
• Change in Spend (ΔSpend):  18 – 15 = 3 
• Change in Return (ΔReturn):  32.27 – 9.10 = 23.17 

Interpretation: increasing spend in Channel 2 by $1 generates $7.72 in additional returns.

Analysis: these results illustrate that reallocating budget from Channel 1 to Channel 2 unlocks greater total returns, as Channel 2 is still in its steep growth phase. While Channel 1 delivers positive incremental ROAS at reduced spending levels, its incremental returns are smaller due to nearing saturation. Conversely, Channel 2 benefits from the additional investment, as it continues to operate in the high-growth portion of its S-curve, delivering higher incremental ROAS. In this case, Scenario 3’s strategy (spending less on Channel 1 and more on Channel 2) results in significantly higher total revenue, demonstrating the importance of recognizing and prioritizing channels with untapped growth potential over those closer to their efficiency limits. This highlights the value of dynamic budget allocation based on incremental returns rather than relying on static averages.

This shift mirrors real-world campaigns where budget reallocation to underutilized channels unlocks growth, even if they initially appear less efficient.

The Incremental Gains Curve (Rate of Change)

The revenue/saturation curves above represent the accumulation of total returns as marketing spend increases. The charts below (dashed grey line) also include the corresponding incremental gains curve, which measures how much additional revenue is generated for every additional unit of spend. This is the rate of change of returns (i.e. first order derivative) at each spend level, and it provides critical insights into the efficiency of additional investments.

Channel 1: Rapid Decline in Incremental Gains

For Channel 1, the incremental gains curve begins steep but declines quickly as the channel saturates. For example:

• At Spend = 5: the incremental gain is 6.06, meaning every additional dollar spent generates $6.06 in revenue. This highlights how efficient the channel is at low spend levels.
• At Spend = 15: the incremental gain drops to 2.22, indicating diminishing returns. At this point, over-investing in Channel 1 becomes inefficient, as each additional dollar yields minimal revenue.

This steep decline illustrates the characteristic behavior of an exponential (convex) curve: rapid early growth followed by quick saturation. Decision-makers must recognize when a channel is nearing its efficiency limit to avoid overspending.

Channel 2: Unlocking Growth Potential

Channel 2 tells a different story. Its incremental gain curve reflects the S-curve behavior, characterized by a slower start, a steep growth phase, and eventual saturation. For example:

• At Spend = 10: the incremental gain is 0.40, indicating the channel is still in its activation phase, with modest returns per additional dollar spent.
• At Spend = 20: the incremental gain peaks at 15.00, marking the “sweet spot” of spend, where returns are maximized.
• At Spend = 25: the incremental gain declines to 4.96, indicating that the channel is entering saturation and incremental efficiency is declining.

This contrast between Channel 1 and Channel 2 demonstrates the value of finding the right balance. While Channel 2 initially lags, its growth potential surpasses Channel 1’s once the steep growth phase is reached.

Why it matters: Incremental gains reveal where the next dollar of marketing spend will have the greatest impact. Instead of relying on historical averages like ROAS, which may mask inefficiencies, incremental gains help identify the most profitable opportunities for additional investment.

Maintain Profitability: Incremental ROAS provides a clear signal of whether the next dollar spent is not only generating revenue but also maintaining or improving profitability. Decision-makers need to assess whether incremental ROAS exceeds a specific threshold, such as:

• Break-even ROAS: ensuring each additional dollar spent generates enough revenue to cover costs (e.g., ROAS > 1.0 if costs are 100% of revenue).
• Profit Margin Targets: Incremental ROAS must exceed the margin requirements to ensure profitability (e.g., ROAS > 2.0 for a 50% margin goal (i.e. revenue retained after costs). For a business with a 50% gross margin, $0.50 of every $1 in revenue is retained as profit).
  

Key Lessons for Decision-Makers

The scenario analysis above highlights several critical insights for optimizing marketing spend. First, it underscores the importance of understanding saturation dynamics. Channels that appear efficient early on, like Channel 1 in this example, often reach a point where additional spend yields minimal incremental returns. Investing further in these channels may seem logical based on historical performance, but it results in diminishing returns and inefficient budget allocation. Conversely, channels like Channel 2, which are still in their growth phase, offer untapped potential for higher incremental returns, even if their current ROAS appears lower.

Second, the analysis reveals that historical ROAS can be misleading as a decision-making metric. ROAS reflects an average return on investment, but averages fail to capture the non-linear growth patterns that define real-world marketing dynamics. A channel with high average ROAS may already be saturated, offering limited opportunity for growth, while a channel with lower ROAS may have significant room to grow. This makes incremental returns—the additional revenue generated by each additional dollar spent—a far more reliable guide for budget allocation.

Finally, the results demonstrate the value of incremental thinking. By reallocating budget based on where the highest incremental returns can be achieved, businesses can optimize their total returns. In our example, reallocating budget from the saturated Channel 1 to the growing Channel 2 resulted in a significantly higher total return, illustrating the power of dynamic, data-driven decision-making.

Wrap-Up: The Value of Incremental Thinking

Moving away from averages like ROAS and adopting incremental thinking allows businesses to:

• Identify channels with untapped growth potential.
• Avoid over-investing in saturated channels.
• Maximize total returns by reallocating budget dynamically.

Linear thinking oversimplifies marketing dynamics and often leads to suboptimal decisions. By understanding saturation curves and focusing on incremental returns, decision-makers can optimize marketing spend to drive both efficiency and growth.

The next time you allocate marketing budgets, ask: Is my decision based on averages like ROAS? Or am I considering the true incremental returns of each channel? Embracing incremental thinking could be the key to unlocking untapped growth.