Return on Advertising Spend (ROAS) is a cornerstone metric for evaluating the effectiveness of marketing campaigns, offering a seemingly straightforward way to connect advertising investment with revenue generation. However, its true value is entirely dependent on the accuracy of the attribution model used to track sales, according to industry experts. Without a robust and precise attribution system, ROAS can present a distorted picture, leading to potentially flawed strategic decisions regarding budget allocation and campaign optimization.
The fundamental calculation of ROAS is indeed simple:
ROAS = (Sales Attributed to Ads) / (Cost of Ads)
This ratio provides marketers with a quick snapshot of how much revenue is generated for every dollar spent on advertising. A ROAS of 5:1, for instance, indicates that for every dollar invested in advertising, five dollars in sales are generated. This metric is invaluable for comparing the performance of different campaigns, channels, and even individual ad creatives, enabling marketers to identify what is working and where resources should be concentrated. For example, recent industry surveys indicate that over 70% of digital marketers consider ROAS a primary KPI, underscoring its widespread adoption and perceived importance.
The Achilles’ Heel: Attribution Accuracy
Despite its widespread use and apparent simplicity, ROAS harbors a significant vulnerability: its reliance on accurate attribution. Mike Murphy, Vice President of Marketing at Incremental, a firm specializing in attribution solutions, argues that ROAS can become misleading when the attribution process overvalues or undervalues certain advertising touchpoints, or worse, completely overlooks crucial aspects of the customer journey and subsequent sales.
Consider a hypothetical e-commerce company that invests $10,000 in retail media advertising. Using a basic "last-touch" attribution model, the company attributes $50,000 in sales to this campaign, yielding a ROAS of 5:1. On the surface, this success would strongly suggest further investment in retail media. However, as Murphy points out, this calculation often fails to capture the nuances of consumer behavior.
"ROAS credits the last ad touchpoint before a sale, whether or not it caused anything," Murphy explained via email correspondence. "This leaves a very big gap. ROAS can take credit for organic sales that would have happened anyway, or for a sale an earlier touchpoint drove. It also only sees what can be tracked directly."
This issue is particularly pronounced within the retail media environment. Platforms like Amazon Advertising or Walmart Connect benefit from an inherent advantage: high purchase intent from shoppers already browsing for products. When a shopper sees an ad for a product they were already intending to purchase on that very platform, the ad may receive undue credit for a sale that was largely pre-determined.
The Crucial Distinction: Attribution vs. Incrementality
Murphy’s critique highlights a critical distinction between simple attribution and incremental sales, often referred to as "incrementality." Attribution, in its basic form, assigns credit for a sale to a specific touchpoint based on a predefined model. Incrementality, on the other hand, measures the additional sales generated by an advertising effort that would not have occurred otherwise.
Returning to the retail media example, if a shopper sees an ad for a product on a retailer’s site and was already planning to buy that exact item, the last-touch attribution model would credit the ad with the full sale. However, the incremental value of that ad might be significantly lower, or even zero, if the sale would have occurred organically through search results or direct browsing.
To illustrate this, let’s revisit the e-commerce company’s $10,000 retail media investment. Further analysis, perhaps through controlled testing or more advanced attribution methodologies, reveals that in approximately half of the instances where the retail media ad appeared, the product was not prominently visible in the organic search results. In these scenarios, the ad truly drove the sale and warrants full credit. However, in the other half of cases, the ad appeared alongside the product’s existing organic listing. Here, the ad’s contribution is more nuanced; it may have influenced the decision or merely reinforced an existing intent.
If the company’s advanced analysis attributes the full $50,000 in sales based on last-touch, the ROAS remains 5:1. But if a more sophisticated model assigns credit only for the incremental revenue generated when the ad appeared alongside organic results, the ROAS could be significantly lower. For instance, if the incremental sales are only $30,000, the ROAS drops to 3:1. This revised figure more accurately reflects the campaign’s true impact, indicating that while the advertising was effective, its contribution might not be as overwhelmingly dominant as the initial ROAS suggested.
The Reverse Problem: Undercounting Sales
The attribution challenge doesn’t solely manifest as overcounting. It can also lead to the undercounting of sales, a phenomenon particularly prevalent in today’s multi-device, privacy-conscious digital landscape.
Murphy notes that an ad might indeed be the catalyst for a sale, but the attribution system may fail to detect it. Retail media provides a prime example again. A consumer might see an advertisement for a product on a large marketplace like Amazon. Intrigued, they might then navigate to the merchant’s own direct-to-consumer (DTC) website to make the purchase, perhaps to avoid marketplace fees or to take advantage of a loyalty program. In this scenario, the initial retail media ad was the driving force, but if the attribution system is confined to tracking within the marketplace, this sale might go uncredited to the ad campaign.
Similarly, the modern consumer often interacts with brands across multiple devices. A shopper might research a product on their tablet during the day, see a retargeting ad on their mobile phone in the evening, and finally complete the purchase on their desktop computer later that week. If the attribution system cannot seamlessly track user journeys across these devices, the sale might be attributed to a different touchpoint, or not attributed at all, leading to an artificially low ROAS for the ad that initially sparked interest.
Major advertising platforms are acutely aware of this issue. Google Ads, for instance, employs sophisticated "conversion modeling" to estimate conversions that may not be directly observable due to privacy restrictions and cross-device behavior. Google itself states that without such modeling, reported conversions would represent only a fraction of actual campaign performance. This highlights the inherent limitations of purely observational tracking and the necessity of advanced statistical methods to bridge the "attribution gap." When an ad genuinely influences a sale but the attribution system fails to register it, the reported ROAS is artificially depressed, potentially leading marketers to underestimate the true value of their advertising efforts.
Testing for True Impact: The Role of Incrementality Studies
To combat these attribution inaccuracies and gain a clearer understanding of true campaign performance, marketers can implement various testing methodologies. These tests are designed to move beyond simple tracking and assess whether attributed sales are genuinely incremental.
For businesses with smaller advertising budgets, Mike Murphy suggests a straightforward "holdout test." This involves selecting a group of products or customer segments and temporarily halting advertising efforts for a defined period (e.g., several weeks) while continuing normal advertising for a control group. By comparing the sales performance of the two groups, marketers can derive directional insights into the impact of their advertising. While not yielding precise figures, these tests can reveal the broad influence of ad spend.
Companies with larger budgets and more established relationships with retail media networks may have access to more sophisticated testing protocols, such as randomized controlled trials (RCTs) or geographic tests. RCTs involve randomly assigning users to see or not see an ad, providing a statistically robust measure of causal impact. Geographic tests, on the other hand, might involve running an ad campaign in specific regions while withholding it from comparable control regions. Regardless of the specific methodology, the overarching goal is to isolate the impact of advertising and assess the accuracy of the reported ROAS. These tests are crucial for validating whether the revenue attributed to campaigns is truly incremental or simply a reflection of existing demand or other marketing efforts.
Establishing a "Source of Truth" Beyond ROAS
While ROAS is a valuable metric, its limitations necessitate a broader perspective. The fundamental expectation of advertising investment is that increased spending should correlate with increased sales and, ultimately, increased profit. Conversely, a reduction in ad spend should ideally lead to a proportional decrease in sales. If these correlations are not observed, it is a strong indicator that the attribution model, and by extension the ROAS calculation, may be flawed.
Beyond ROAS, marketers should consider a suite of complementary metrics to paint a comprehensive picture of campaign performance. These can include:
- Customer Acquisition Cost (CAC): The total cost of sales and marketing efforts needed to acquire a new customer. This metric helps understand the efficiency of bringing new buyers into the fold.
- Customer Lifetime Value (CLTV): The total revenue a business can expect from a single customer account throughout their relationship. A high CLTV can justify a higher CAC, as long as the overall profitability remains strong.
- Market Share: The percentage of total sales in an industry generated by a particular company. Advertising plays a crucial role in capturing and expanding market share.
- Brand Awareness and Sentiment: While harder to quantify directly in dollars, shifts in brand recall, recognition, and public perception are critical long-term indicators of advertising effectiveness, often measured through surveys and social listening tools.
However, as Mike Murphy emphasizes, the ultimate arbiter of advertising success remains the bottom line. "Your P&L should be your first source of truth—it doesn’t lie," he asserts. The Profit and Loss statement provides a clear, unvarnished view of a company’s financial health, directly reflecting whether advertising investments are contributing to sustainable profitability. Metrics like Gross Profit Margin and Net Profit Margin offer a clearer perspective on the efficiency and effectiveness of all business operations, including marketing. If advertising spend is rising, but profit margins are eroding, it signals a potential disconnect between marketing activities and business objectives, irrespective of a seemingly healthy ROAS.
In conclusion, while ROAS remains a vital performance indicator, its utility is intrinsically tied to the sophistication and accuracy of the underlying attribution model. Marketers must move beyond simplistic calculations and embrace methodologies that account for the complexities of the modern consumer journey, the nuances of different advertising environments, and the critical concept of incrementality. By integrating robust testing and a broader set of financial and marketing metrics, businesses can develop a more accurate understanding of their advertising’s true impact, ensuring that their investments drive not just revenue, but sustainable and profitable growth.
