Christopher is an authority in digital marketing in marketing technology, data science, and machine learning.
Christopher is a recognized and best-selling author and keynote speaker for Google Analytics, a fan-driven marketing team for email marketing.Â
He is a four-time IBM Champion and IBM analytics brand.
Top 24 over 100 digital marketers and top AI marketing influencers.
Work with custom brands such as Twitter, T-Mobile, Go Daddy, AAA, Mcdonald’,s and many others.
Knowledge Bombs
18:26 Ranking In Google Christoper said that you need to create helpful content that’s unique and authoritative, and timely.Â
20:43 Process of Learning SEO
You could hook into Wikipedias or RSS feedback and make an exact copy of Wikipedia that was dynamically generated so that you could pull a lot of traffic. The strategy back then could have been more effective, and everything is experimentation. Nowadays, Google needs to learn how its search algorithm works because the neural network models underneath the search interface are vast and have no explainability. You have no empirical way of proving that unless you’ve done a very large scale like propensity score matching on tons of data. You don’t know how Google works because Google doesn’t know how Google works.
27:31 Transcribing Into A Language Model
If you are using chat gpt, you have to copy and paste it. If you have some technical skill, you can use the GP3 model with Open AI, take large quantities of text, and know funnel in and parse. Christopher said you could use AHREFS as an SEO tool with keywords, match the trend data from Google Trends, and get keyword volumes and trends. And then, you can use a predictive analytics forecast for 52 weeks and find out when this next keyword will trend based on the past five years of data, and then I take that keyword and the combination of keywords that trend with it.Â
29:30 Other pieces of data to collect to have more effective prompts
It’s not data that you’re collecting for effective prompts. There’s a paperback in 1957 called the synopsis of linguistic theory by John Rupert Firth that prompted that one sentence that paper that has formed the basis of all-natural language processing theory for 70 years, you shall know a word by the company it keeps. You shall know a word by the company it owns. But this concept means you will get irrelevant or low-quality results f you create props with insufficient numbers of relevant words.Â
You need to use words and phrases that funnel the attention of these models towards particular subtopics, and these large language models can understand all the text they’ve ingested. He also shared a brain hack prompt engineering is a lot like SEO. Think of it as a prompt optimization. What keywords would you use if you’re doing regular SEO to get a search engine to recognize a page use that same thought process for writing a prompt what keywords would go in a prompt that would tell a search engine about this and you will get much better results out of these large language models.Â
31:36 Envisioning prompt engineering and essay evolving over the next few years
It starts to go away, according to Christopher. The reason for this is that prompt engineering and the skill that you need to use. We are all working essentially on massive models like chat gpt3. With all of these models, if you have sufficient knowledge, technical skill, or deep enough pockets, you can fine-tune where you take your content and retrain the main part of the model. In this way, you are pre-training these models because you have already pre-trained them and tuned them on some content. You will require much less effort to generate good results from your model, which the Fortune 50 companies already do.
Your unbranded search traffic will be around for a while, and your unbranded search traffic will drop precipitously in 6-12 months. For people who’ve used chat, gpt doesn’t think about your behavior of the app in the application and how many times you’ve said it. Accordingly, this is going to devalue the traffic. As a defense for marketers, we have to be good at the brand. 37:39 Distinction between experience and expertise
When you have an experience, your experience is valid. The expertise says that it was still well done even though you didn’t like it. Expertise inherently involves more profound technical knowledge of a subject than someone with experience alone wouldn’t have and the easiest way to. That is, it’s what versus why a person with experience can with what happened but someone with that deep technical expertise. Expertise can tell you why that thing happened. You can have both and probably will have both in some ways; you can be an expert and also have an experience because you should have the expertise to become the expert. There are also who have experience but don’t necessarily have expertise.Â
39:45 Important factors to consider when evaluating someone’s expertise
It depends on your level of risk. Suppose you are curious about chat gpt. With chat gpt, they gave you some suggestions, and maybe it worked or didn’t. If you are about to roll out a big project and you are going to use the gpt3 model underneath it, you probably want that to go well. So you will want someone with expertise.
42: 24 How to become an expert if we don’t have that background
You will learn it. You may not necessarily get a professional certification, and nothing stops you from reading all the papers and understanding how this goes. You can make good decisions and why you are doing that because you spent time learning the technical parts of the subject.
Google News
Let’s start with the latest from Google News.
Google says you don’t need to host the favicon in the same domain to be eligible for a favicon in Google Search results.
Google clarified that Discover uses many of the same signals as Search, in both the Discover documentation and the helpful content system page.
https://developers.google.com/search/updates
Does the depth of the category negatively affect rankings?
Ted Kubaitis shared his thoughts on SEO Fight Club this past week.
