The Test Setup
We selected 12 client sites across different industries and ran a controlled test over 24 weeks. Each site published a mix of: (1) pure AI-generated content with minimal editing, (2) AI-drafted content heavily edited and enriched by a human expert, and (3) fully human-written content from subject matter experts. We tracked rankings, organic traffic, time on page, and bounce rate for each content type.
What AI Is Genuinely Good At
Structure and outlines. AI is excellent at producing well-organized content frameworks. Give it a topic and target keyword, and it reliably produces a logical structure with appropriate headings. We now use AI for every content brief — it saves hours of planning time.
First drafts for simple topics. For straightforward informational content on well-documented topics — "how often should you get a dental checkup," "what is a 301 redirect" — AI drafts are a solid starting point. They don't require the depth of expertise that makes human writing valuable for these.
Meta descriptions, titles, and short-form copy. AI consistently produces better meta descriptions than most humans, because it naturally hits keyword targets and stays within character limits. Same for ad copy and social posts — it's faster and often better.
Where It Falls Short
Pure AI content ranks poorly for competitive terms. This was our clearest finding. Pure AI content — unchanged from the model output — consistently underperformed in rankings by an average of 8–12 positions compared to human-edited equivalent content targeting the same keywords. The content was technically accurate and well-structured, but it lacked the specificity, nuance, and distinctive perspective that Google's quality raters and algorithms reward.
Engagement metrics are worse. AI-generated content had 34% higher bounce rates and 40% lower average time on page compared to expert-written content. Readers can tell the difference, even if they can't articulate it — the content feels generic, lacks memorable examples, and doesn't reflect real experience.
E-E-A-T is missing. AI can describe experience but can't have it. For local service businesses, the most valuable content comes from actual knowledge: the unusual cases you've seen, the local factors that affect your service, the mistakes clients make before coming to you. None of that can be replicated by a language model, no matter how capable.
Our Current Approach
We use AI tools extensively, but always with human expertise in the loop. Our workflow: AI generates structure and first draft → subject matter expert (the business owner or our SEO specialists) enriches with real experience, local references, and specific insight → editor refines for readability and on-page optimization → publish.
This hybrid approach produces content that ranks 20–30% better than pure AI output, with significantly better engagement metrics. It also takes about 60% less time than purely human writing — the best of both worlds.
The Bottom Line
AI is a powerful writing assistant. It's not a replacement for expertise, experience, or local knowledge. If you're publishing pure AI output and expecting it to compete with expert-written content, you'll be disappointed. But if you're using it to multiply the productivity of genuine expertise, it's one of the best tools we've added to our workflow.
Curious how we'd apply this to your content strategy? Get in touch — we're happy to walk through our approach with any new client.