Oren Greenberg

Oren GreenbergAug 24, 2026

Reddit mentions in chatGPT dropped off a cliff. It's not a bad thing as heavily prone to manipulation. Now you may be asking: 'so....what...?' ChatGPT is now indexing more homepages + including curated domains. So if before you were thinking 'ahh, I was just enjoying the imminent demise of battling wordpress / webflow. Yes finally freedom from our damn website!' Unfortunately you're now back to: 'we must make our website number 1 priority like I've been saying all along Chuck... Yes! AND destroy all competition we must! (and mosquitoes while we're at it)' It's like a merry-go-round you can never get off.

Oren GreenbergSep 8, 2026

You already know that posting more frequently on LinkedIn increases follower count. But what's the distribution look like? Analysis last 12 months of my posting on LinkedIn. 60% of my posts netted zero new followers (but I reckon there's still value for existing network who see / engage). What you can see in the 2nd image is that higher frequency results in more followers. So, how is it possible that most posts don't get new followers, but posting more frequently nets more new followers on average? Because of the outliers who do well. But since it's really hard (at least for me) to know how a post is going to perform in terms of engagement / followers a higher-frequency posting volume strategy works well. Conclusion: Stop sweat the performance of an individual post. Keep posting. Some will take off, and those are the ones you build your influence based on. Note worthy: 'followers gained' on indiivdual posts in LinkedIn analytics shows you only about 10% of all new followers. So it isn't reliable.

Oren GreenbergSep 10, 2026

AI models have an inbuilt bias that changes how you should think about your brand ranking. Say you're looking for a ride-hailing app. You ask Claude or ChatGPT to run a search and find the best options for you. What you expect is that the AI model will search online for something like 'ride-hailing app for a busy body like me'. But the reality is that the AI model picks brands like Uber, Lyft, Waymo from memory. Then searches "Uber vs Lyft", "Waymo reliability vs Uber" etc. Essentially variants of the same top names over and over again. So the search isn't really a search. It's self-homework marking. The model already picked its shortlist before it opened a single webpage. The web "search" just confirms what it already knows. But before I get into that this is exactly the kind of thing we'll be digging into at Spotlight by Semrush on the 13th of October in London. Join over a 1,000 marketers with 80+ speakers from the likes of OpenAI, Visa, Uber, Wix and Spotify all unpacking AI & search. Experts walking through what they've built, tested, and shipped. I spoke there last year and it was ace (find your promo code for 40% off:  bit.ly/4z9sEU3). Now let's get back to how AI and humans are similar in their bias. If you think about how humans argue down the pub. You don't Google a brand you've never heard of to settle a bet - you Google the ones already in your head (and then ignore anything that disagrees with you if you’re anything like my friend Jo). To put some numbers to this the brands the AI model has in memory are 3 times more likely to be included in the query it runs. Roughly 55% versus 17%. But big brand names (like the top 5 above) are 66% more likely to be included in the query the AI model runs. At this point you may be asking: 'so what?' So now back to what It means: Smaller / less known brands are at a disadvantage; AI models haven't really democratised information due to this bias. You're better off focusing on building your brand through online PR, being mentioned & cited through creating original research. The searching happens after the shortlist, not before. Dave in marketing is currently three tools deep trying to "optimise for AI search." Bless him. You may want to share this post with him.