Gary Taylor retweeted
If Apple integrates OpenAI at the OS level, then Apple devices will be banned at my companies. That is an unacceptable security violation.
Gary Taylor retweeted
Google's @JohnMu on toxic links again "The concept of toxic links is made up by SEO tools so that you pay them regularly" seroundtable.com/google-agai… (note, seems like John edited his response later - story updated)
Gary Taylor retweeted
Bing gained less than 1% market share after launching Bing Chat / Copilot GPTs about a year ago seroundtable.com/bing-market…
Gary Taylor retweeted
Today I wanted to share about the belief that there is some type of “perfect page” formula that must be used to rank highly in Google Search. There isn't, and no one should feel they must work to some type of mythical formula. It’s a belief dating back to even before Google was popular, as I wrote about when I was a journalist in 2000, in the article below. As was the case then, so it remains true now. There’s no perfect formula to follow….
Gary Taylor retweeted
BREAKING NEWS!🔥
Shots Fired at Nakatomi Plaza, untold number of hostages. #NakatomiPlaza #YippeeKiYay
Gary Taylor retweeted
Oldie but goodie - Google's @methode warned years ago not to build your business around Google Discover traffic. He said, "I just hope people realize that this may go away overnight and they don't build their whole business model around it." seroundtable.com/google-warn…
Gary Taylor retweeted
Major changes coming to Google search rankings seroundtable.com/google-sear…
Gary Taylor retweeted
Google's 3 pillars of search ranking seroundtable.com/google-3-pi…
Gary Taylor retweeted
On this date in 1985 Duck Hunt was released in North America for the Nintendo Entertainment System. #80s #1980s #videogames
Gary Taylor retweeted
Google September 2023 helpful content is now done rolling out searchengineland.com/google-…
Gary Taylor retweeted
Google says links are no longer in the top three search ranking signals seroundtable.com/google-link…
Gary Taylor retweeted
Google August 2023 Core Update rollout is now complete after 16 days searchengineland.com/google-…
Gary Taylor retweeted
Google published a set of rules (checklist) for #GA4 UTM tracking, but most people continue to ignore this checklist and tag URLs however they please. Then, they complain about the unassigned traffic appearing in their reports.
.
.
Unassigned traffic in GA4 refers to the website traffic that is not attributed to any one of the following default channel groups:
Affiliates
Audio
Cross-network
Direct
Display
Email
Mobile Push Notifications
Organic Search
Organic Shopping
Organic Social
Organic Video
Paid Other
Paid Search
Paid Shopping
Paid Social
Paid Video
Referral
SMS
.
.
Google uses certain rules (called ‘channel rules‘) to group similar website traffic sources (usually that all belong to the same traffic medium).
For example,
Consider the following traffic sources:
google (as in google / organic)
yahoo (as in yahoo / organic)
bing (as in bing / organic)
aol (as in aol / organic), etc.
Since they all have the same traffic medium, ‘organic’, they will be grouped together under the channel group ‘organic search’.
Google does not have any pre-defined channel rules to group traffic sources that have either user-defined sources and mediums or they are (not set).
So what it will do is group these traffic sources under the ‘unassigned’ channel group.
.
.
Your UTM parameters should strictly follow the channel rules for manual traffic (check the screenshot below) wherever you can so that your website traffic is attributed to one of Google‘s default channel groups.
For example, consider the following URL of a Facebook ad:
https :// www. abc .com/book-maths-and-stats/?utm_source=facebook&utm_medium=paid-fb-ad&utm_campaign=pdf-book-campaign&utm_content=ad1
Here, ‘paid-fb-ad’ is not a system-defined traffic medium.
So Google will label all the website traffic from this URL under the ‘unassigned’ channel group.
But this is not what you want. You want this traffic to be grouped under the ‘paidsocial‘ channel.
To do that, rewrite the UTM parameters like the one below:
https :// www .abc .com/book-maths-and-stats/?utm_source=facebook&utm_medium=paid&utm_campaign=pdf-book-campaign&utm_content=ad1
Here, paid is a system-defined traffic medium.
.
.
Use system-defined traffic sources and system-defined traffic mediums wherever possible.
A system-defined source is one that is already defined/recognized by GA4 as a traffic source.
Similarly, a system-defined medium is one that is already defined/recognized by GA4 as a traffic medium.
The more user-defined traffic sources and/or medium you will use, the higher will be the unassigned website traffic in GA4.
Memorize channel rules and definitions for accurate tagging.
It is very important to know how Google groups similar traffic sources and how it defines a particular marketing channel.
This knowledge will help you tremendously in tagging URLs that follow GA4 default channel rules.
Share this information with all those who tag URLs in your organization. Everybody needs to be on the same page.
Gary Taylor retweeted
GPT-4 is getting worse over time, not better.
Many people have reported noticing a significant degradation in the quality of the model responses, but so far, it was all anecdotal.
But now we know.
At least one study shows how the June version of GPT-4 is objectively worse than the version released in March on a few tasks.
The team evaluated the models using a dataset of 500 problems where the models had to figure out whether a given integer was prime. In March, GPT-4 answered correctly 488 of these questions. In June, it only got 12 correct answers.
From 97.6% success rate down to 2.4%!
But it gets worse!
The team used Chain-of-Thought to help the model reason:
"Is 17077 a prime number? Think step by step."
Chain-of-Thought is a popular technique that significantly improves answers. Unfortunately, the latest version of GPT-4 did not generate intermediate steps and instead answered incorrectly with a simple "No."
Code generation has also gotten worse.
The team built a dataset with 50 easy problems from LeetCode and measured how many GPT-4 answers ran without any changes.
The March version succeeded in 52% of the problems, but this dropped to a pale 10% using the model from June.
Why is this happening?
We assume that OpenAI pushes changes continuously, but we don't know how the process works and how they evaluate whether the models are improving or regressing.
Rumors suggest they are using several smaller and specialized GPT-4 models that act similarly to a large model but are less expensive to run. When a user asks a question, the system decides which model to send the query to.
Cheaper and faster, but could this new approach be the problem behind the degradation in quality?
In my opinion, this is a red flag for anyone building applications that rely on GPT-4. Having the behavior of an LLM change over time is not acceptable.
Have you noticed any issues when using GPT-4 and ChatGPT lately? Do you think these problems are overblown?





























