Google and Meta’s Paid Media Incentive Problem
▼ Summary
– Digital advertising platforms like Google and Meta have superior data and optimization capabilities, leading some to question if advertisers could simply provide a budget and URL for fully automated campaigns.
– Platforms have a history of misaligned incentives, such as Google reps activating a declined product feature without authorization, resulting in wasted ad spend for which the advertiser bore the cost.
– Common platform pitches, like using gross margin to justify unlimited spend or claiming higher CPCs buy better traffic, often misrepresent economic realities by ignoring diminishing marginal returns and non-incremental conversions.
– Reporting practices like blending funnel performance, counting view-through conversions, and using competitor benchmarks can obscure true campaign inefficiency and pressure advertisers to increase spend.
– Default account settings and explanations like extended “learning phases” or tracking gaps often prioritize platform revenue over advertiser success, creating pitfalls especially for new or trusting users.
A recent discussion in my executive MBA program raised a provocative question: if platforms like Google and Meta possess superior data and processing power, why shouldn’t advertisers simply hand over a budget and let the algorithms run autonomously? The logic seems sound. These platforms have access to unparalleled datasets and their optimization engines are incredibly sophisticated. However, this approach requires absolute trust that the platforms will prioritize an advertiser’s business outcomes over their own revenue growth. A review of common industry practices reveals why such trust is often misplaced.
Consider the experience of having a new Google product activated without consent. After explicitly declining a pitch for broader targeting, representatives enabled the feature anyway. The result was a significant budget overspend with no improvement in conversion rates. Attempts to recoup the wasted funds were met with a defense that the platform was authorized to spend up to our budget cap. This framing treats a budget as an invitation, not a strategic ceiling. The incentive structure clearly pushed reps for feature adoption, yet offered no accountability when their unilateral decision failed.
Another classic scenario involves a profit maximization pitch. Google representatives once calculated that, given a client’s gross margin, any media spend generating revenue above cost was justified. This simplistic math assumes all reported conversions are incremental and ignores the reality of diminishing marginal returns. In practice, a substantial portion of conversions, especially in brand campaigns, would have occurred without paid ads. Furthermore, the cost per conversion escalates as spend increases, making the last dollars spent the least efficient.
The suggestion to raise cost-per-click (CPC) bids to access “better quality” traffic is another frequent recommendation. While higher bids can improve ad position and frequency, this argument conveniently overlooks the other side of the equation. The value of additional impressions declines rapidly, and you pay more for each click. Often, this strategy simply yields the same results at a higher cost, eroding return on ad spend (ROAS).
Platforms also frequently invoke the learning phase of their algorithms to explain poor performance. While machine learning models genuinely require data to optimize, this concept has become a catch-all excuse that delays accountability. Without a clear definition of success or a definitive endpoint, “it needs to learn” can function as a blank check to continue spending despite unsatisfactory results.
When direct conversions are lacking, a common pivot is to brand lift metrics like recall and sentiment. While these are legitimate brand health indicators, the shift typically occurs only after harder conversion metrics fail. Introduced reactively, they act as a consolation prize without a clear framework for evaluating cost-effectiveness or connecting to revenue.
Reporting practices can further obscure performance. Blending upper-funnel and lower-funnel campaigns into a single average cost-per-acquisition (CPA) can mask wildly inefficient segments of spend. Similarly, the default inclusion of view-through conversions,particularly for retargeting audiences,significantly inflates reported performance. These users were already likely to convert, making the ad’s incremental impact questionable.
Competitive benchmark data is a powerful lever. Reps often present figures showing rivals outspending you, implying a need to close the gap. This tactic leverages competitive anxiety, bypassing crucial questions about whether that competitor’s spend is effective or if the comparison is even relevant to your business model.
For new advertisers, default settings present a major trap. Guided setup processes often opt users into broad match keywords, expanded networks, and non-optimal targeting that maximizes platform revenue from day one. These configurations are revenue-optimized for the channel, not performance-optimized for the advertiser.
Finally, legitimate tracking gaps from privacy changes are sometimes used to defer accountability. Arguments that “conversions are happening but we can’t see them” or that modeled data needs more time can justify sustained spend in the absence of any measurable result. The uncertainty inherent in measurement tends to be resolved in the platform’s favor.
This landscape doesn’t mean automation lacks value or that agencies are irreplaceable. Automated bidding strategies like tCPA are now standard, and in-house teams are more viable than ever. However, the case for fully autonomous, channel-run advertising hinges on the assumption that the platform will optimize for your profit, not its own. Even with potential future profit-sharing models, this carries inherent risk.
The essential role for advertisers, whether working with an agency or in-house, is to persistently ask critical questions. What is the true marginal return at this spend level? How many conversions are view-through? Are we measuring incrementality or just correlation? While full automation may be the future, the entity building the system should not be the sole authority on when it is truly ready for your business.
(Source: Search Engine Land)
