How your human brain helps improve the Google Ads AI

As one of the world’s leading pioneers in artificial intelligence, Google often claims that the AI it uses to power its PPC advertising platform is smarter than ever. However, while it is true that Google is constantly testing and refining its algorithms and introducing new features and functions, it still cannot replace human intuition. Because of this, advertisers should not rely on it entirely if they want to maximize their return on investment.

Bikes vs. bicycles and Google’s misunderstanding

One of our international clients is a large e-commerce store that specialises in a wide variety of consumer goods. To help them maximise their reach, we launched specific campaigns for each product category and then created customised ad groups for them. However, we noticed that one of the ad groups was performing significantly below expectations. In July 2019, we introduced an ad group to promote bicycles for the store. Naturally, we made sure that all our keywords included the word ‘bicycle’, along with various related terms pertaining to the products and services offered by our client. We came up with more than 1,000 keywords, which we added to the ad group, and each ad mentioned the word ‘bicycle’ at least twice. As such, the ad group was perfectly optimised for the topic at hand. From then on, the performance of the campaign continued to improve until February 2020. In March, due to the disruptions caused by the pandemic situation, we decided to take a closer look at the search terms associated with our campaigns with a view to addressing brand safety concerns. However, we were surprised to discover that, for the ad group promoting bicycles, we ended up with a disappointing Quality Score of only 3.8 out of 10. This was despite having a keyword-optimised landing page and ads that clearly mentioned bicycles. The reason for this turned out to be even more surprising. While we had only used keywords that mentioned bicycles, Google decided to display our ads for users looking for bikes. Though that made sense, what surprised us was how Google gave us a low quality score, which turned out to be because the ads did not mention ‘bikes’. As such, the click-through rate (CTR) was lower, and the reduced quality score meant inferior ad positions and higher costs per click.