.png)
Average ROAS involves certain risks when splitting your budget across marketing channels. Here, discover how to address these issues with three practical examples.
Calculating the average ROAS has its limitations and risks when allocating your budget across different marketing channels.
This article explores these limitations and offers practical solutions. By the end, you’ll understand:
In recent years, advertisers have embraced attribution modelling to make investment decisions more data-driven and accountable. With unified measurement techniques, we can establish a data-driven foundation for strategic and tactical budget allocation and media scenario planning.
ROAS has become a key tool for data-driven budget optimisation. In fact, Return on Ad Spend, or ROAS, is one of the most widely used marketing metrics, especially for decisions about marketing budgets. It’s often prioritised over ROI (Return on Investment) when focusing on ad costs and returns.
While ROI remains relevant, it doesn’t help optimise marketing activities at a campaign level—it provides a broader view, comparing revenue to overall marketing spend.
For campaign budget decisions, you need a precise metric that links business value directly to specific ad campaigns, providing the right data for allocating budgets across channels.
Attribution results can be an essential step towards data-driven media planning. However, relying solely on historical media impact and effectiveness is insufficient.
The following realistic example illustrates how this approach works in practice. Suppose we’re making a budget decision for an upcoming campaign. For this campaign, we need to allocate budget between two channels: TV and (Online) Video.
To make this allocation in a data-driven way, we examine historical attribution results from similar campaigns.
The historical campaign results indicate that TV and Video had similar investment levels on average. However, Video generated higher average revenue (€6,550) compared to TV (€2,400), resulting in a higher ROAS for Video (€2.63) than for TV (€0.96).
So, Video has a higher average ROAS than TV. Does this mean Video always performs better than TV in this scenario? Not quite.
Looking at this example, it might seem reasonable to direct more investment towards Video, given its higher ROAS. This is a classic case of the “flaw of averages.”
Sam Savage, writing in the Harvard Business Review, explains that “plans based on assumptions about average conditions usually go wrong.” In a humorous example, Savage tells the story of a statistician who drowned while crossing a river he calculated to be three feet deep on average.
This story illustrates the limitations of relying on averages in decision-making, showing how average-based decisions can lead to serious consequences. ROAS is no exception.
In our example, it might seem sensible to invest more in Video because of its higher average ROAS. But this isn’t the best approach. Instead, we need to examine historical investments and assess their incremental impact on the KPI of interest.
When you base your ad budget on average ROAS, you assume that each additional investment will yield similar ROAS, implying a linear relationship between investment and revenue. This approach suggests that each extra euro spent will continue to produce the same return—an unrealistic expectation.
In practice, we often see non-linear relationships between investments and revenue. It’s crucial to assess historical investment levels and compare them to the observed incremental revenue. To achieve this, we create investment sensitivity curves that reflect expected revenue at varying investment levels.
Investment sensitivity curves help identify where additional investment has the greatest impact and should form the foundation for scenario analysis. Incremental ROAS calculations can also reveal the optimal budget allocation for future investments. These sensitivity curves are also called ‘shape effects’ or ‘investment elasticity curves.’
Returning to our example, we’ve developed the following investment sensitivity curves:
Investment sensitivity curves illustrate the expected revenue at different investment levels. In our example, the Video curve shows a diminishing trend: the initial investments are relatively effective, but as budgets rise, we see diminishing returns.
The TV curve, however, presents an S-shaped pattern. It requires a minimum investment to begin generating a revenue impact. After reaching this initial threshold, TV becomes more efficient until about €3,000, beyond which it also starts to show diminishing returns.
These patterns align with typical trends in channels like TV and Video. For more traditional channels such as TV and Radio, a minimum investment level is often needed to observe an impact, and these channels generally don’t support smaller budgets.
To understand this more thoroughly, let’s delve into investment sensitivity with three possible scenarios. Suppose you need to allocate budget between TV and Video for an upcoming advertising campaign. How would you use ROAS to guide your channel investments?
In previous campaigns, we observed a budget split of €2,500 for TV and €2,500 for Video.
This scenario is represented by the purple markers on the graphs, showing the expected revenue for each channel.
Here, we anticipate revenue returns of €2,400 from TV and €6,550 from Video, totaling €8,950. This scenario would yield a ROAS of €1.79, which is considered strong.
In the second scenario, we adjust the budget split by shifting €500 from TV to Video, following the “flawed” assumption that funds should go to the channel with the highest average ROAS. This adjustment results in a new budget split of €2,000 for TV and €3,000 for Video.
This scenario is shown by the purple markers on the graphs, with arrows highlighting the budget adjustments from the original split.
Here, Video’s investment is increased to €3,000 (+€500), leading to an expected revenue of €6,850, while TV’s investment is reduced to €2,000, yielding an expected revenue of €800.
The total revenue in this scenario would be €7,650, with a ROAS of €1.53, which is lower than the ROAS in the initial investment.
In the third scenario, we adjust the budget by moving €500 from Video to TV, based on insights from the investment sensitivity curves. This results in a budget split of €3,000 for TV and €2,000 for Video.
This scenario is represented by the purple markers on the graphs, with arrows indicating the budget changes from the historical split.
In this case, the TV investment is increased to €3,000 (+€500), yielding an expected revenue of €4,900.
Meanwhile, the Video investment is reduced to €2,000, resulting in anticipated revenue of €6,250. This scenario generates a total revenue of €11,150 and a ROAS of €2.23, the highest among all three scenarios.
Comparing the three scenarios, each involves the same total investment. With the historical budget split, we achieve a revenue of €8,950.
At first glance, increasing investment in Video (the channel with the higher ROAS) seems logical. However, the investment sensitivity analysis suggests otherwise. Specifically, Video appears to be saturated at this spend level, resulting in a 15% decrease in revenue compared to the historical budget split.
Conversely, TV is not yet saturated at these investment levels, leaving room for additional spending. Thus, shifting more budget to TV leads to a potential 25% increase in revenue for this example.
From our experience with leading advertisers, we know that attribution results are a valuable starting point for data-driven media planning. Yet, relying solely on historical media impact and effectiveness is insufficient. To optimise your advertising budget, you need to utilise marketing metrics effectively.
This example illustrates how average ROAS can be a “flawed” metric, potentially resulting in poor budget decisions for future campaigns. Instead, investment sensitivity offers a clearer view of the incremental impact of each future investment.
In this case, analysing investment sensitivity curves led to a possible 25% increase in revenue, whereas ignoring this analysis could cause a 15% decline.
Of course, ROAS isn’t the only metric for optimising your marketing budget. Other factors also play a significant role.
Your industry, for example, is crucial. Optimising budgets for insurance brands involves unique steps, while e-commerce brands require a different approach.
Interested in learning more about optimising your marketing spend? Explore how predictive marketing can help you forecast campaign performance and revenue.
