Microsoft’s AI Ad Play

Microsoft is exploring the use of its AI capabilities to enhance its advertising business. The company has filed a patent for a machine learning model that incorporates “incrementality estimation” to predict the performance of sponsored content.

Microsoft defines incrementality as the impact of ad budget on performance. The model is trained to forecast how content will perform based on the allocated budget, enabling the assessment of budget effectiveness in achieving desired outcomes, such as clicks, impressions, or job applications.

In its patent filing, Microsoft criticizes common machine learning models for inaccurately equating higher budgets with better performance.

These models often rely on biased historical data, leading to inaccurate recommendations. To address this issue, Microsoft’s model employs an approach called “asymmetric budget split.” It trains on two distinct sets of data—one with a high-budget focus and another with a low-budget focus. By avoiding the pooling of data, the model ensures unbiased estimations and avoids judgments of low-budget ads based on the performance of high-budget ones, and vice versa.

While advertising is not Microsoft’s core business, the company generated $3 billion in ad and search revenue in the most recent quarter.

However, this figure pales in comparison to competitors like Meta and Google, which earned $28 billion and $61 billion, respectively, from ads during the same period.

Despite this, Microsoft’s subsidiary LinkedIn has exhibited faster growth, reporting $3.6 billion in revenues, an 8% increase from the previous year.

Therefore, it is understandable that Microsoft’s patent filing emphasizes the improvement of job posting performance alongside ads.

Implementing AI into its advertising strategy makes sense for Microsoft, given its leadership in the AI domain.

The company is leveraging its strengths to address its weaknesses, even though competition in AI-generated ads exists with players like Google, Meta, and Amazon.

Microsoft’s patent has the potential to render advertising A/B testing unnecessary, as AI could accurately predict the effectiveness of ads prior to publication.

This technology could significantly enhance advertising outcomes, as pointed out by Dan Ratner, CEO of Australian ad agency uberbrand.

The patent filing not only provides insights into Microsoft’s advertising approach but also highlights the broader impact of AI on the digital advertising industry.

Arun Kumar, EVP of data and insight at Hero Digital, believes AI will revolutionize content creation, advertising delivery, and effectiveness measurement.

While concerns about job displacement persist, Kumar argues that AI will enable better efficiency by reallocating human resources to more value-added projects.

Overall, Microsoft’s exploration of AI in advertising showcases the potential for technology to drive improvements and efficiency in the industry.