Shedrick Williams

← Back to blog

June 22, 2026

Does AI ruin SEO for new websites?

A study to see how the current level of AI integration in search engines affects the growth of new websites starting from scratch.

TL;DR: I predict AI search integrations will cause a lot of changes for the internet, so I want to conduct a study to see how it affects new websites. I'll be gathering metrics on this domain over 1, 3, and 6 months to see if AI has a positive, negative, or neutral impact on growth.

AI SEO experiment search results preview

The integration of Large Language Models (LLMs) into search engines is fundamentally altering the internet. The sheer volume of information, and the medium through which it is served, will change drastically as AI overviews continue to replace traditional search results. To understand exactly what this means for the modern web, I am conducting a live study to see how the current level of AI integration affects the growth of a brand-new website.

This study will track and analyze the growth metrics of this domain over specific time scales: 1 month, 3 months, and 6 months post-launch.

Methodology

While I do not have a pre-existing, high-traffic domain of my own to use as a control group, I will benchmark this site's performance against historical SEO growth data available across the web. At each milestone, I will gather the analytics generated from this domain and draw comparisons to standard natural growth rates.

Even without a direct 1:1 control, the data logged from this domain’s growth will be sufficient to identify any unnatural bottlenecks or acceleration at these time scales. At the arrival of each milestone, a follow-up post will be published breaking down the analytics, analyzing the traffic sources, and comparing the results to the broader web.

Expected Results

The resulting data should allow us to deduce whether AI integration into search engines has a positive, negative, or neutral impact on domains starting from scratch today. Further analysis will hopefully reveal whether AI favors specific types of written content. Particularly whether simply formatted, question-and-answer structures are more susceptible to AI usage and citation.

Follow the Experiment

If you want to track the results of this study and see the data at the 1, 3, and 6-month marks, subscribe to the email list below. You will be notified the moment the next batch of metrics is published.

Automate good info

Want the data when it's out?

Enter the email you want the results delivered to.