-
-
August 2026 Insights and Action Items
-
- 1. AI content is now roughly half the internet, and Claude just started watermarking it.
What's happening: A Graphite/GPTZero study found the web is now about a 50/50 split between human- and AI-generated content, and as of August 2nd Claude began embedding machine-readable watermarks to comply with the EU AI Act, though the mark may not survive selective copy-paste and third-party checkers appeared within hours.
2. Google's position hasn't changed: it's not whether you used AI, it's the quality.
What's happening: Claude's watermark news has folks spooked , but Google still judges the content and not the tool, so the takeaway isn't "stop using AI," it's that everything making AI content trustworthy happens before and after the tool, through unique data, SMEs, governance, and review.
3. The programmatic-content easy button is a real trap for established brands.
What's happening: Lily Ray's research on 200+ sites found these AI and programmatic plays spike fast then crash, with more than half of sites losing 30% or more of their peak organic traffic, nearly 40% losing half or more, and one in five losing 75% or more, so the more authority you've built, the more you have to lose.
4. "Surviving" listicles share four traits.
What's happening: Nick's research found that while listicles are declining as AI citations overall, the ones still winning are fresh, genuinely comprehensive (10 to 20 options with real pros and cons, not thin five-to-eight self-serving lists), backed by a transparent data methodology, and cleanly structured for crawlers.
- August 2026 Action Items
-
- 1. List the unique inputs you currently include in your content creation workflow. What else can you add?
- 2. Review your 10 most recently published pages. How many of them match Lily Ray’s “high-risk templates”?
AMA with Alisa and Nick
As always, we ended with open Q+A. Have a question you want us to cover next month? Submit it here.
Nick's POV: That's really scrappy, and it's something I want to go into our own Search Console and look at immediately. It could be evidence. I'd line it up with the model releases: when 5.5 became the default at the beginning of May, is that when the site searches start increasing? We know site search was used about 55% more in 5.6 than 5.5, so it could definitely be a signal.
Alisa's POV: A big spike, like the June 11th one the question asker shared, also tells you that you have an audience using the paid version of ChatGPT. Which raises the harder question of whether your audience is on paid or free, because that one shift can significantly change the brands appearing for a prompt. What's old is new again.
Alisa and Nick's Latest Thoughts
Our Early View of ChatGPT Ads: 5 Initial Observations - Updated July 28, 2026
When we first published this on July 15, we were working from three days of ChatGPT ads data. Revisiting the data this week, our main takeaways haven't changed,…
The Proof: ChatGPT 5.5's fanout patterns reveal the importance of brand
If you've been looking at the whole "do I have to do listicles" thing and throwing up in your mouth a little, this post is for you.
Stop Chasing AI Rankings Before You Fix How LLMs See Your Brand
The counterintuitive truth about AI visibility: Most brands are trying to win elections before voters know their name.
Study of 800K AI Responses: How Review Profiles Shape Brand Presence in AI Search
Does having a brand presence on third-party review sites impact your visibility in AI search?