AI Content v AI Quality

Lily Ray has captured the industry's mounting frustrations with AI content on Substack. The perception of AI as a "golden egg-laying goose" led to a focus on low-hanging fruit and manufactured content, ultimately causing us to lose sight of the inherent value of quality content.

While Google has commoditised certain types of information, the question remains: who is producing non-commoditised content?

For the last decade, Google has pushed E-A-T, recently adding a second "E" for Experience. This framework functions until it does not, by which point the engagement cycles or packages for such content have typically concluded.

I am playing devil's advocate because the way Google's systems and LLMs function, specifically how they select citations, is complex and multifaceted. Despite strong correlations and evidence both for and against specific influences, the notion of mass-producing content has never been a sound strategy.

Many businesses have built themselves upon robust programmatic programs. There is a fundamental misunderstanding between the strategic nature of programmatic SEO and the simple act of having AI churn out thousands of pages.

Examples such as Zapier and other connector SaaS products naturally lend themselves to strong programmatic campaigns because they address a linear search path: "Does X work with Y?" or "Can Y work with Z?"

Mass AI content generation has never truly been about SEO rather, it has always been about the relationship between the brand and the client. While this may have been a viable play for those seeking passive income or short-term gains, Google's systems are becoming increasingly preventative and intelligent, preventing low-quality content from surfacing.

This means these tactics work for the start or duration of a relationship, and may buy some time when performance begins to decline, but the client is left in a no better position by the end.

While various LLMs have different success rates in combating this, we can see a shift in user behaviour. When searching for reviews, ChatGPT is increasingly appending websites like Reddit within prompts to find authentic human experiences.

I understand the current desire to produce less content, as this approach challenges the legacy SEO tactics that effectively built the industry.

In the past, we utilised keyword mapping and dedicated pages for specific phrases, exploring keyword universes to create content clusters targeting precise search terms. This is what clients were educated SEO was, and what we in turn sold.

We then evolved to focus on search intent, the primary purpose of the page, and the specific value we were providing to the user.

We are now better informed to sell, rather than just relying on arbitrary internal linking, keyword mentions, optimised headers, and other controllable variables.

This transition was logical when we were driven by search volume. We used third-party data to identify what people were searching for and in what quantity, creating pyramid structures of content. But we are now in an environment focused on the "infinite tail" where the monthly search volume (MSV) is often just one.

If you take this new data approach and the sheer breadth that these low-volume, personalised "infinite tail" queries bring, and attempt to apply the old way of producing content, whether commercial or informational, the strategy fails.

The AI content tsunami has introduced many marketers to the nuances of how Google crawls and indexes websites, exposing them to technical elements they were previously unaware of.

When Google Search Console reports "Crawled - currently not indexed" or "Discovered - currently not indexed" it is often perceived as a technical fault. But it is frequently a quality issue.

You can damage a domain's reputation by producing too much content, regardless of quality, as Google restricts resources. I’ve seen this a number of times, from both poor programmatic content, AI slop, and from Google mass indexing over a billion URLs it shouldn’t. This isn't just because the content was written by AI, it's because a large number of low value URLs were introduced into the ecosystem in such a way Google pulled back resources.

What we need to do

We must recognise that LLMs and Google’s core ranking systems are rapidly evolving to circumvent content manipulation tactics.

The algorithmic focus has pivoted toward surfacing genuine utility and brand-client trust, meaning that short-term, low-quality strategies are becoming increasingly higher risk and lower reward.

In this "infinite tail" environment, attempting to scale via AI-generated slop is a losing game, long-term viability now demands a departure from legacy volume-based content plays and now catering for an MSV of one, built on strong information retrieval foundations and adding value.