AI Content Performance in Google Search: 16-Month Study

Generating and publishing content with artificial intelligence is remarkably fast, but achieving lasting visibility in Google search is a different challenge. A recent 16-month study tracked the performance of purely AI-generated articles on new websites with no established authority. The findings reveal a clear pattern: initial promise followed by a steep decline, highlighting the limitations of relying on automation without a strategic foundation.

The research involved launching 20 new domains, each targeting a distinct niche like Health, Finance, or Travel. Every site published 100 articles created by AI without any human editing, for a total of 2,000 pieces of content. The goal was to observe how this content would perform organically without backlinks, promotional efforts, or advanced SEO work.

In the first month, the results appeared strong. Over 70% of the pages were indexed by Google, and the sites collectively garnered over 122,000 impressions. Early visibility was significant, with 80% of the sites ranking for at least 100 keywords. This initial phase demonstrated that Google is willing to crawl and test AI-generated material, especially in broad, evergreen topics.

Growth continued into the second and third months, with cumulative impressions climbing past 526,000. However, a pivotal shift occurred around the three-month mark. Rankings collapsed. Only 3% of pages remained in the top 100 search results, down from 28% initially. The content stayed indexed but virtually disappeared from searchers’ view. This ranking drop underscores that early relevance is not enough to sustain positions without stronger quality signals.

The long-term outlook, monitored for over a year, showed mostly stagnant performance. While a temporary spike occurred for some sites coinciding with a Google spam update in August 2025, it did not lead to a sustained recovery. By the end of the 16-month period, cumulative clicks across all 2,000 articles totaled just 1,381, with the majority of activity stemming from the early growth phase before the decline.

The core issue is that the AI content lacked the elements Google uses to assess quality and trust. The sites had no backlinks to build domain authority, no named authors or credentials to establish E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness), and offered little unique insight compared to existing web content. Without these critical signals, Google’s algorithms ultimately deprioritized the pages.

A follow-up test provided an interesting secondary insight. Adding new AI-generated content to some of the stagnant sites months later resulted in a noticeable traffic boost. Intriguingly, the increase primarily benefited older pages on the site, not the newly published ones. This suggests that publishing fresh content can signal to Google that a site is active, potentially giving it a temporary lift in visibility. However, this effect appears to be short-term and does not equate to building lasting rankings.

The key lesson is clear. AI content creation is a tool for scale and speed, not a substitute for a holistic SEO strategy. It can produce a foundation of content quickly and may even provide short-term visibility gains. For enduring success, however, that content must be enhanced with human oversight, strategic optimization, and the building of authority signals that search engines reward. Automation handles the quantity, but quality, expertise, and trust still require a human touch.