AI-generated Content

This is an update targeting low-quality content generated by AI. As someone who runs an agency that generates a large volume of AI-generated content for about 20 to 30 websites, I just want to talk about the impact this has had on our clients.

Google releases a few of these spam updates every year, and they can target a couple of different types of spam. Sometimes they target link spam, sometimes they target reputation abuse spam, and sometimes they target content spam.

Among most spam updates, this one rolled out very quickly. It was completed in only 19 hours, and this made it very easy for Google to evaluate the effect. From essentially all the reporting online, this was not targeting link spam or reputation abuse. The sites that got hit the hardest, and some sites really got hit hard, were sites that were generating a large volume of AI-generated content.

So this morning, 30 days after the update, I went through the Google Search Console of every one of my clients’ websites to see what impact the spam update had on their traffic. Content creation is the primary strategy we use for all of our clients. Essentially all of our content is 100% written by AI, and for most of our clients, we’re publishing content every single week.

So why did I wait 30 days after the spam update to check if there was a traffic loss? It’s because I understand how Google actually evaluates spam. And sure enough, we didn’t see a single client’s website get hit by this update. A handful of them saw an increase in traffic, likely caused by competitors who were hit, and for the rest, there was no disruption in their traffic pattern.

So how did we avoid being hit, and when do you actually need to worry about getting hit by one of these spam updates? First, you need to understand that Google doesn’t care if you’re using AI to generate your content. All they care about is the quality of the content.

The thing is, it’s really hard for their algorithms to detect quality. Even attempting to evaluate quality would dramatically increase Google’s costs. Google has no idea if your content is accurate or insightful. Instead, Google’s algorithms are trying to evaluate whether your content was low effort and whether it satisfied the intent of the searcher.

Even though the content we’re generating for our clients is entirely written by AI, it’s far from low effort. We have about 12 steps in our content creation process, from keyword research to in-depth research on the topic, collecting information from our client, drafting the article according to our writing guidelines. That process is connected to our client’s brand ambassador, which has a bunch of information about the client, and then a series of quality checks and revisions. But at the end of that process, it’s rare that a single word is written by a human.

We still have a lot of human involvement in our process, but as the tools get more and more capable, more and more of the process can be handled by AI. In fact, right now we’re in the middle of reworking that process to see which steps could be handled better by AI than by a human.

And that’s the key distinction. We’re not giving responsibilities to the AI that humans could handle better just to cut costs. The tasks we’re giving to the AI are the ones where they’re actually better than humans.

Look, I worked with a lot of talented content writers before ChatGPT came out, but none of those writers could remember every single aspect of a client’s business, every detail of every product and service and team member. None of them could remember a hundred SEO guidelines and a hundred more specific to this client. None of them could ever capture every opinion the brand shared on social media and incorporate that into the content. When you’re smart about how you use AI to write content, Google doesn’t see this as spam.