Seer Interactive’s Post

AI vs Search is all the rage but holistic search marketers know that Groups in places like LinkedIn, Whatsapp, Facebook, Discord and others are places people go to get answers often in groups that are curated & reside in trusted networks. How do you write the content that gets human rewards not just algorithmic rewards? Content that is shared in people's private groups, newsletters, emails, etc. Here's a little ad hoc "case study" on what happened when we prioritized getting traffic from places like this over Google for one of our own posts. It wasn't going to be easy, we knew that what was ranking on Google likely wasn't good enough. Who would actually share those URLs in a trusted network? So Wil Reynolds put his expertise into 1 post that took over 8 hours to flesh out. Is winning for humans in your content success equation? https://lnkd.in/gmmiNqb7

If your prompts on Claude 3 or Gemini don't seem to get you great results, consider taking these official docs from Anthropic and Google Gemini on how to best prompt their systems to get better answers. - h/t Alisa Scharf [Claude 3] - https://lnkd.in/dtz_4Vv5 [Gemini PDF] - https://lnkd.in/eKz8t4Dm

A consideration for those of you making the decision to block ChatGPT from crawling and training on your content. Keep in mind blocking ChatGPT seems to also prevent your posts from showing up in answers when people ask questions on that platform.

See what happens when we were willing to risk the high ranking of a page on our site to pivot and write something on the topic that stood out and offered unique value over every other top ranking page for this keyword. The willingness to sacrifice "ranking high" lead to a lot of learning, lessons, and thoughts on the future of content marketing, SEO, and more.

We're seeing more domains than ever in Supernova, but we're seeing less "domain diversity" in the top 10. Ever feel like you see the same site listed multiple times for a query even when the match isn't that great (esp. on brand or navigational searches). But it isn't just brand. Meghan Evans & Tracy McDonald collaborated to whip up something to tell us if our eyes were deceiving us, and they weren't. Looking at preliminary data from Supernova across multiple industries & domains, we're seeing that over time, the % of unique domains in the top 10 organically seems to be declining, if you want us to dive deeper let us know / put your questions / thoughts below and we'll get a more deeper dive in a blog post.

We are about to delete over 50% of our email recipients. When Wil Reynolds took an honest look at our data, he saw a slow negative slide, death by 1,000 unsubscribes one at a time & he said let's do better. Starting with...be less annoying. We believe that in a world with more content being produced than ever, that what people really want is less content and more curation of all the stuff that is out there. We also believe that schedules create discipline, but they also create complacency, it can lead to "I gotta send the weekly newsletter" when it should have always been, "Let's email people when we think we can share something that will help them do their jobs better". We are going to start off by sending emails whenever we think we wrote or saw something that changes the dialogue, makes you pause and think. It could be weekly or it could be once a month, it might go out on a Sunday if we're feeling it, oh the horror. We believe that if you create value you don't have to play the "clickbaity" & "best practices" game, we try to challenge the status quo in our service offering, why wouldn't email be any different? As Wil would say - "We weren't put on this planet to be the best practices agency." See the last 40-49% of you all on the other side, looking forward to re-building.

We don't always get it right, but we try hard to get back to everyone to close the loop. We know it's hard right now for people looking for jobs and sometimes it is helpful just to let them know how competitive this job was. Also Dave, thank you for putting a little good in the world today too. cc Emily Allen, PHR

Going "all in" on an LLM? Jason Stinnett dropped this gem internally and his answer is don't lock in, the world is changing a bit too much. As artificial intelligence continues to evolve, you have to keep in mind the trajectory of AI models, the pace of change, and how quickly they improve. In the coming years, we can expect to see all kinds of AI models, each with their own unique capabilities and resource requirements. AI models will continue to be developed in different sizes – small, medium, large, and extra large, catering to the specific needs of users and balancing capabilities with efficiency. Finding the right price point for these capabilities is crucial. Comparing models like Claude Opus and Gemini Pro can show a real difference in cost, highlighting the challenge of balancing performance and affordability. The concept of "plugins" adds complexity to the AI landscape. Certain functionalities may be available in consumer-facing user interfaces but not yet accessible through APIs, creating a gap between cutting-edge features and their scalability for businesses. It's essential to remain adaptable and open-minded. So, all that said, what do you do? IMO, punt the decision for as long as you can, and continue to explore as many different models as possible. Given the variety of offerings and the rate of innovation, now is not the time to lock yourself into one model, just ask Bloomberg :)