Muslim
A recent analysis of digital advertising accounts has uncovered a pervasive issue that is systematically inflating performance metrics, leading advertisers to believe they are achieving significantly better results than they actually are. This phenomenon, primarily driven by misconfigured conversion tracking settings within platforms like Google Ads, is creating an "illusion of success" that can have profound financial and strategic implications for businesses. The core problem lies in how various conversion actions are counted and valued, often leading to inflated Return on Ad Spend (ROAS) figures that do not reflect genuine sales or revenue.
The investigation, stemming from observations by industry experts during live training sessions and account audits, highlights a critical flaw: a significant portion of reported "conversions" are not derived from actual sales or tangible business outcomes, but rather from artificially assigned values or duplicated tracking. This misrepresentation is not always malicious but often a result of oversight, outdated settings, or a misunderstanding of how conversion tracking mechanisms function.
The Root of the Problem: Inflated Conversion Values and Duplicated Counts
At the heart of the issue is the way conversion values are calculated and aggregated within advertising platforms. A common scenario involves static values assigned to non-purchase conversion actions, such as leads. While these values are intended to provide an estimated worth for lead generation efforts, they can become problematic when they are changed without a corresponding adjustment in historical reporting periods.
Consider an example where an account generates $100,000 in revenue from 1,000 orders and an additional 2,000 leads, with a quarterly ad spend of $20,000. If each lead is assigned a static value of $10, the reported conversion value would be $120,000, resulting in a ROAS of 6.0 ($120,000 / $20,000). Now, imagine that before the next quarter begins, the static value for leads is increased to $20. If this change only applies to new conversions, and the account experiences the exact same number of orders and leads in the subsequent quarter, the reported conversion value would jump to $140,000 ($100,000 from orders + $40,000 from 2,000 leads at $20 each), yielding a ROAS of 7.0 ($140,000 / $20,000). In this instance, the ROAS has increased by one-sixth, not due to any actual improvement in sales or lead quality, but solely because a human manually altered a single input field. This creates a misleading narrative of improved campaign performance.
The danger here is that when comparing performance year-over-year or quarter-over-quarter, these figures are often presented as directly comparable, despite being measured in different "units" of value. The charts and reports, drawn as a continuous line, fail to acknowledge that the underlying valuation methodology has changed, leading to an inaccurate perception of growth and efficiency.
The "Every" Setting: A Miscounting Conundrum for Leads
Another significant source of inflated conversion numbers stems from the "Every" setting for conversion actions, particularly for lead generation. While the "Every" setting is appropriate for tracking purchases (as a single individual can indeed make multiple purchases), it often leads to overcounting for leads. Google’s own best practices suggest that typically only one unique lead per ad click adds value. However, when the "Every" setting is applied to lead forms, even minor user errors can trigger multiple "conversions." For instance, a user submitting a form, realizing a typo in their phone number, and resubmitting it would be counted as two separate leads. Similarly, a thank-you page that loads twice due to a technical glitch can also result in duplicate lead counts.
Data from account audits reveals that approximately four out of every five accounts analyzed that feed leads into bidding systems have at least one lead action configured with the "Every" setting. Across all lead-related actions within the conversions column, this issue affects roughly half of them. This means that a substantial volume of reported leads are not unique, valuable interactions but rather inflated counts from single user journeys.
Duplicate Purchase Tracking: A Stealthy Inflator
Beyond lead miscounts, a significant number of accounts also suffer from duplicate purchase tracking. More than one in four accounts with purchase tracking have two or more purchase actions listed within the conversions column. While in some cases these might track genuinely different sales channels or products, it is often the case that they are counting the same order multiple times. This duplication can occur through various means, such as tracking a sale via the Google tag on a website and simultaneously importing the same order data through Google Analytics 4 (GA4) or a Customer Relationship Management (CRM) upload.
The most insidious form of this duplication is not the easily identifiable broken counter that is typically removed, but rather a secondary tracking mechanism that closely mirrors the primary one. This "dangerous duplicate" looks perfectly healthy because it accurately reflects back-end sales. However, when this secondary action is added to a campaign’s conversion goals, the reported conversion value can effectively double overnight without any system alerts, as nothing technically "broke." This can lead to campaigns aggressively optimizing towards a falsely inflated revenue figure.
Unused Primary Actions: Ghosts in the Bidding Machine
A further complication arises from conversion actions that are marked as "primary" but are not actively used by any campaign goals. Approximately two out of every three analyzed accounts exhibit this issue. These actions, despite being designated as primary, are not part of the immediate bidding strategy and often go unnoticed because they do not appear in the main conversions column. The reasons for their initial setup are frequently forgotten over time.
However, a simple edit to a campaign goal or the creation of a custom goal can inadvertently activate these dormant primary actions. When a custom goal is configured to include a secondary action, that action will be used for bidding once the goal is implemented in a campaign. The problem is exacerbated by the fact that audits rarely involve a thorough review of custom goals, leaving these potentially irrelevant or misconfigured actions to influence bidding strategies without scrutiny. This can lead Smart Bidding algorithms to chase events that were set up years ago for reasons long forgotten, diverting budget away from more impactful activities.
Cheap Placements: The Bait for Misguided Bidding
The impact of these misconfigurations is amplified when combined with Smart Bidding strategies, particularly Target ROAS. When inexpensive actions like page views, file downloads, outbound clicks, or directions requests are incorrectly included in the conversions column, Smart Bidding algorithms can be easily misled. In approximately six out of ten accounts, these "soft events" are present alongside actual sales-driving actions.
These soft events do not fire evenly. Cheap placements, which are often less competitive and easier to achieve, trigger these soft events far more readily than a search ad typically generates a genuine order. Consequently, campaign budgets begin to shift, incrementally week by week, towards the placements that are most efficient at "filling the counter" of these soft events. The reporting then interprets this as improved efficiency and performance, masking the fact that the budget is being allocated to activities that do not directly contribute to core business objectives.
While there can be a strategic exception for using soft events as proxies in very small lead generation accounts where actual leads are scarce, this must be a deliberate decision, with the proxy valued appropriately based on its conversion rate to actual sales. In most cases, however, these soft events are included in the conversions column through oversight rather than strategic intent.
This practice is particularly detrimental to Target ROAS bidding. When the underlying conversion counting is broken, a Target ROAS goal does not protect profit margins; instead, it provides the algorithm with precise instructions based on flawed data, leading to potentially disastrous bidding decisions.
The Imperative for Pre-Testing: A Free Diagnostic Before Expensive Trials
In light of these findings, industry experts advocate for a rigorous internal audit of conversion tracking before engaging in more complex and costly performance testing. While incrementality testing, which measures the true lift in sales generated by paid search efforts, is a valuable tool, its insights are only as reliable as the data it relies upon. If the conversion column feeding into these tests is already inflated or inaccurate, the results of the incrementality test will be similarly compromised.
A thorough review of conversion settings, including static value assignments, counting methods ("Every" vs. "One"), duplicate tracking mechanisms, and the active usage of primary actions, can be completed in a matter of hours for most accounts. This "free test" – an afternoon’s diagnostic work – can prevent significant financial waste and strategic missteps that might arise from costly incrementality tests run on faulty data.
The analysis suggests a phased approach to performance verification:
- Initial Audit: Scrutinize all conversion actions within the "conversions" column, paying close attention to static values, counting methods, and potential duplicates.
- Value Reconciliation: Separate reported conversion value into actual order revenue and any estimated or typed-in values. This allows for a clear understanding of what is being measured and what is being estimated.
- Performance Benchmarking: Establish a baseline of true, measured performance based on accurately tracked sales and revenue.
What survives this initial three-step process is growth that can be reliably compared year-over-year. This verified number should then be the focus of budget discussions and the subject of more sophisticated testing, such as incrementality studies.
The Bottom Line: Measurement Must Be Truthful
Ultimately, a ROAS figure is only a meaningful measurement if every component contributing to it has been accurately measured. A typed-in value represents an estimation from a specific point in time, not a real-time reflection of revenue. A second counter that duplicates an existing order adds no new information, only inflated numbers. When these elements are included in the critical conversion column that dictates advertising spend, they become expensive liabilities.
The student who reported a ROAS exceeding 500% with no actual sales serves as a stark, albeit extreme, example of this phenomenon. Her account’s error was so glaring that it was impossible to miss. However, the more common scenario involves subtler inaccuracies that create a number advertisers want to believe.
Moving forward, a crucial step for any advertiser is to ensure that reported order revenue and any estimated or typed-in lead values are presented as distinct lines in their reports. This transparency will empower stakeholders to understand precisely which portion of their reported ROAS is based on tangible sales and which is derived from estimations, fostering a more accurate and actionable understanding of campaign performance. The integrity of advertising data is paramount, and neglecting the fundamentals of conversion tracking can lead even the most sophisticated campaigns astray.
Nana Wu
Asro
