The AI Content Study is planned research into how AI-assisted content performs in search compared with human-written content, and how much editing and review change the outcome. It hasn't been run, so there are no results yet.
It's the question behind a lot of nervous meetings: “If we use AI for content, will it hurt us?” Opinions are everywhere. Careful tests are rarer.
Status
Planned. No data has been collected or published for this study yet. Anything you see quoted elsewhere as coming from this study isn't ours.
The question
How do pages written with AI assistance perform in search compared with human-written pages on similar topics, and does the amount of human editing make a difference?
Why it matters
Teams are deciding how much to rely on AI for content right now. Knowing whether heavy editing pays for itself would make that decision much easier.
How we plan to run it
- Pages on comparable topics, split into groups: human-written, AI-drafted with heavy editing, and AI-drafted with light editing.
- Tracking indexing, rankings, and clicks over several months.
- Honest notes on the limits, since no test can control every factor in search.
- Sharing the method and, where possible, the page-level data.
What the finished study will include
- The full methodology, including its limits.
- The sample size, and who or what was included.
- Findings, including the unexciting ones.
- Charts.
- Downloadable data, where we can share it.
- Citations for anything we build on.
- Publication and update dates.
Until then
Until there's data, a sensible approach is:
- Use AI inside a quality process with research, rules, and human review.
- Track your AI-assisted pages separately so you can compare them with the rest.
- Judge each page by whether a subject expert would put their name on it.
Where to go next
- AI content and Google: our current practical stance
- Keyword to article workflow
- All research