Google’s AI Overviews Coincide With Drop In Mobile Searches
Study reveals Google’s AI overviews coincided with decline in mobile search volume, likely causing Google to pull back on the feature .Google’s AI Overviews Coincide With Drop In Mobile Searches
AI overviews rollout coincided with a significant drop in mobile searches.
Google quickly dialed back AI overviews after user feedback and errors.
Study also reveals nearly 60% of Google searches end without any clicks.
You can read more SEO articles
A new study by search industry expert Rand Fishkin has revealed that Google’s rollout of AI overviews in May led to a noticeable decrease in search volume, particularly on mobile devices.
The study, which analyzed millions of Google searches in the United States and European Union, sheds light on the unexpected consequences of AI integration.
AI Overviews Rollout & Reversal
In May 2024, Google rolled out AI overviews in the United States, which generate summaries for many search queries.
However, the feature was met with mixed reactions and was quickly dialed back by the end of the month.
In a blog post published on May 30, Google admitted to inaccurate or unhelpful AI overviews, particularly for unusual queries.
Google says it implemented over a dozen technical improvements to its systems in response.
A subsequent study by SE Ranking found the frequency of these summaries decreased, with only 8% of searches now triggering an AI Overview. However, when shown, these overviews are now longer and more detailed, averaging 25% more content.
SE Ranking also noted that after expansion, AI overviews typically link to fewer sources, usually around four.
Decline In Mobile Searches
Fishkin’s analysis reveals that the introduction of AI Overviews coincided with a marked decline in mobile searches in May.
While desktop searches saw a slight increase, the drop in mobile searches was significant, considering that mobile accounts for nearly two-thirds of all Google queries.
This finding suggests that users may have been less inclined to search on their mobile devices when confronted with AI-generated summaries.
Fishkin commented:
“The most visible changes in May were shared by both the EU and US, notably… Mobile searches fell a considerable amount (if anything spooked Google into rolling back this feature, I’d put my money on this being it).”
He adds:
“If I were running Google, that dip in mobile searches (remember, mobile accounts for almost 2/3rds of all Google queries) would scare the stock-price-worshiping-crap outta me.”
Impact On Overall Search Behavior
Despite the dip in mobile searches, the study found that search behavior remained relatively stable during the AI overviews rollout.
The number of clicks per search on mobile devices increased slightly, while desktop clicks per search remained flat.
This indicates that while some users may have been deterred from initiating searches, those who did engage with the AI Overviews still clicked on results at a similar or slightly higher rate than the previous months.
Implications For Google & the Search Industry
The study highlights the challenges Google faces in integrating AI-generated content into its search results.
Additionally, the research found other concerning trends in Google search behavior:
Low Click-through Rates: Only 360 out of every 1,000 Google searches in the US result in clicks to non-Google websites. The EU fares slightly better with 374 clicks per 1,000 searches.
Zero-click Searches Dominate: Nearly 60% of searches in both regions end without any clicks, classified as “zero-click searches.”
Google’s Self-referral Traffic: About 30% of clicks from US searches go to Google-owned properties, with a somewhat lower percentage in the EU.
Why SEJ Cares
This study underscores the need for adaptable SEO strategies.
As an industry, we may need to shift focus towards optimizing for zero-click searches and diversifying traffic sources beyond Google.
The findings also raise questions about the future of AI in search.
While major tech companies continue to invest in AI technologies, this study suggests that implementation may not always yield the expected results.
Follow HiTrend on X