How Artificial Intelligence has changed organic search, and how businesses can adapt to the new market conditions

Rise to the Top with ChatGPT

First, let’s define what ChatGPT is.

ChatGPT is an intelligent language model developed by OpenAI. It uses deep learning technology to understand and generate human language, allowing it to answer questions, engage in conversations, and provide information on various topics. Here are the key aspects of ChatGPT:

1. Language model: ChatGPT is based on the GPT (Generative Pre-trained Transformer) architecture, which is specifically trained on large volumes of text data. This enables it to understand context and generate human-like text.
2. Artificial intelligence: The model uses artificial intelligence algorithms to process and analyze language data. It can answer questions, hold conversations, write text on a given topic, generate code, assist with translations, and more.

Few people know that ChatGPT existed before 2022 when it was released to the public, but it was used by a select few, performing specific tasks without much hype around it. However, everything changed on November 30, 2022.

OpenAI announced the release of ChatGPT, which is a language model based on GPT-3.5, for public access through a web interface.

How it happened:

1. Launch and announcement: On November 30, 2022, OpenAI announced the release of ChatGPT through its blog and official social media channels. The model became available for free on the OpenAI website, where anyone could create an account and start interacting with ChatGPT.
2. Availability and integration: ChatGPT was presented as a web application accessible through a browser, making it easy for users to interact with the model without complicated setups or technical knowledge. Additionally, OpenAI launched an API, allowing developers to integrate ChatGPT into their own applications and services.
3. Feedback and improvement: After the launch, OpenAI actively gathered user feedback to improve the model. Considering comments and suggestions, the OpenAI team made numerous enhancements to the model’s performance, ensuring more accurate and useful responses.

Early versions of ChatGPT were more akin to entertainment and didn’t hold much value. Users often noted that the chatbot couldn’t navigate the links provided for analysis; instead, it inferred information from the URL’s title and the context of the question, often inventing facts or product prices. For instance, in 2022, users asked the bot for the price of the iPhone 15, which wasn’t yet released in Ukraine, and it would respond with random numbers, proving its inability for analytical conclusions.

Over time, developers acknowledged that the chatbot only utilized information available to it as of its release date, November 30, 2022, and had not received new knowledge. But everything changed in 2023.

Improving the ChatGPT Model

ChatGPT gained the ability to browse links with the launch of web-browsing tools in June 2023. This was part of advanced capabilities that OpenAI introduced to make the model more functional and helpful for users. Adding web browsing enabled ChatGPT to:

1. Access up-to-date information: The model can now find the latest data and events, essential for questions requiring recent news or changes.
2. Verify information: With web browsing, ChatGPT can confirm facts or find sources for additional information, improving the accuracy of responses.
3. Interact with web applications: This extends the model’s capabilities, allowing it to interact with other websites to perform various tasks, such as booking tickets, finding products, checking schedules, and more.

These functionality expansions made ChatGPT an even more valuable tool for various tasks, enhancing its ability to provide relevant and accurate answers to complex questions.

At this point, users realized they had a ready-made tool to fill their websites with content, and, most importantly, to make money. For some, this model replaced traditional Google search.

The boom of generated content

Even the first public version of ChatGPT-3.5 was capable of generating content that publishers began to use widely to fill their sites. Copywriters quickly started losing jobs. It seemed we were witnessing a revolution as a whole industry collapsed, leaving hundreds of thousands of professionals out of work.

The Google index grew rapidly, with hundreds of thousands of new generated articles appearing within hours, all of which needed processing and indexing. This number of pages only continued to grow. Google engineers realized that if nothing changed, they would soon be overwhelmed by the data volume and would either have to expand indefinitely to handle millions of new pages or take drastic action.

HCU Update

To address the millions of new documents on the internet every day, and as a result, to relieve Google’s crawlers from scanning all these pages, engineers released a new core search engine update to help clear out unnecessary, generated content and de-index millions of documents that held no real value.

HCU (Helpful Content Update) is a Google algorithm update aimed at improving the quality of search results by identifying and lowering the rankings of content that doesn’t benefit users. This update was introduced by Google in August 2022. Copywriters regained their jobs as publishers hesitated to use generated content for their websites, but it was already too late.

The goal of HCU:

1. Improving content quality: The HCU update was designed to elevate the visibility of high-quality, informative content that truly helps users. It reduces the ranking of websites with content created solely to manipulate search engine algorithms.
2. Fighting low-quality content: HCU helps diminish the impact of content created without considering users’ interests and needs. This applies to content that does not offer new or useful information or is written solely to attract traffic without genuine value.
3. Focus on human value: The update rewards content created for people rather than for search engines. Google aims for websites to provide helpful, informative, and original information that meets users’ queries.

Key aspects of HCU:

• User-focused content: Content should cater to the interests and needs of users, providing them with real value.
• Quality over quantity: Preference is given to in-depth and well-researched content over mass production of low-quality articles.
• Originality and expertise: Content should be original, offering new information or unique insights based on the author’s knowledge and experience.

The Google HCU update was introduced to improve the quality of search results, focusing on helpful and informative content that brings real value to users. This update helps reduce the visibility of websites with low-quality content.

“Losers” and “Winners” of the Update

Losers

But things didn’t go as planned. Hundreds of reputable large publishers, with extensive teams of copywriters and specialists, began losing traffic.

In many cases, the picture of organic traffic looked like this: a site lost 40% of its traffic in a single day without generating content using AI.

Following the first update, a U.S.-focused wedding-themed website lost 30% of its traffic in a week.

In the image, you can see a site I worked on, where traffic reached nearly one and a half million unique users per month, generating significant revenue for the client. The site was maintained by a whole team of copywriters and experts, with no generated articles. Now the site is completely de-indexed from Google search.

Cases like this have flooded Network X (Twitter). Publishers began sharing their stories of traffic loss, the entire industry suffered significant losses, and people lost millions of dollars.

An X user shares his observation on how his site lost traffic after the core update.

The article he wrote, containing useful guides, photos taken during a city visit, and conclusions based on personal experience—everything Google’s engineers claimed the search engine wanted to see—simply disappeared from the top. Instead, a five-year-old Reddit thread that didn’t provide any answers and only posed questions took its place, making it unhelpful to users.

Traffic to the article:

The image shows the organic traffic for an article that previously attracted nearly 6,500 users per month but dropped to zero after the update.

Overall traffic to the site:

The image shows the site’s organic traffic, which fell from one million users per month to almost zero following Google’s “helpful content update.”

Winners

However, if some sites lost traffic—or rather, their rankings—others took their place. And, of course, platforms like forums became these replacements. Notably, the platform Reddit became particularly successful.

As shown in the image, among social networks, Reddit saw the largest growth over the past year.

Traffic to Reddit was 130 million users per month before the update; it has now reached 600 million users per month.

Not as significant a beneficiary as Reddit but still notable is the platform Quora.

Traffic to Quora grew from 47 million users per month to 170 million.

What caused such changes and the clear bias in favor of forums? The issue is that after the success of the ChatGPT model, Google’s engineers feared they might lose their search dominance, as more and more users began turning to such models for minor queries rather than Googling.

Gemini – AI-based language model from Google

Gemini is a new generation of artificial intelligence from Google, developed by the company DeepMind. This family of models integrates advanced technologies, including large language models (LLMs) and generative AI models, similar to ChatGPT. Gemini was introduced as a successor to large-scale language models, such as GPT-4 from OpenAI.

Key features of Gemini:

1. Integration of understanding and generation: Gemini combines natural language understanding with generative capabilities, allowing it to create text, images, and other forms of content based on user prompts.
2. Multimodality: Gemini can work with various data types, including text, images, audio, and video, making it a versatile tool for a wide range of applications.
3. Deep learning: The Gemini models use advanced deep learning and neural network technologies, enabling them to handle complex tasks and deliver high accuracy in results.
4. Contextual understanding: With enhanced natural language processing algorithms, Gemini can better understand the context of prompts, allowing it to provide more accurate and relevant responses.
5. Scalability: Gemini models are designed with scalability in mind, allowing them to be easily adapted to different user and business needs.
6. Efficiency: Through optimized algorithms and infrastructure, Gemini delivers high performance and quick response times.

To enhance its model, Google signed a $60 million contract with Reddit, allowing Google to use forum responses for model training.

Users link this to the recent popularity of forums, though both Google and Reddit deny this connection.

This year, Google plans to introduce AI in search results, with the expectation that this function could replace the traditional search experience.

AI Overview

Google announced new AI-based search features at Google I/O. Here’s a closer look at these updates:

1. AI summary customization: Users will be able to “simplify” or “elaborate” information to match their comprehension level.
2. Multistep reasoning: The search will handle complex queries by breaking them down into multiple steps for detailed responses.
3. Planning: AI will help create plans for various tasks, such as organizing vacations or planning menus.
4. Organized search results: Results will be grouped under AI-generated headings for easier viewing.
5. Lens video search: Users can search for information using video and ask questions during recordings.

Search trends, Washington Post analysis

Search Trends:

• 🟩 AI auto-generated summaries remain relatively stable (declined from 17% to 13%) 📉
• 🟨 Prompted AI summaries (appearing after clicking “Generate in SERPs”) dropped from 45% to 16%.
• 🟩 + 🟨 Over the past weeks, AI summaries have steadily declined (62%-29%) 📉
• 🟥 The number of queries without AI summaries rose from 38% to 70% 📈

In 2023, Google rapidly increased the coverage of queries for AI summary testing and gradually reduced it before the official launch.

Google’s Profits

• Google’s parent company, Alphabet Inc., after a successful quarter, will pay dividends for the first time in its history.
• In the three months ending March 31, Alphabet’s revenue reached $67.6 billion, surpassing analysts’ forecast of $66.1 billion. This growth was driven by Google Cloud’s growth, whose revenue increased to $9.6 billion from $7.5 billion.
• Google’s ad revenue rose by 13% to $61.66 billion. Search ad revenue grew by 14% to $46.16 billion, and YouTube’s by 21% to $8.1 billion. Net profit amounted to $1.89 per share, exceeding Wall Street’s estimate of $1.53.
• The company will also pay a dividend of 20 cents per share and repurchase shares worth $70 billion. Google co-founder Sergey Brin will receive $146 million in dividends, and Larry Page $78 million.
• Following the financial report release, Alphabet’s stock rose by 13%. Over the past 12 months, the company’s capitalization increased by 45% to $1.95 trillion.

Conclusions

However, it seems that users are still hesitant to trust AI-generated results, continuing to rely more on traditional search. This indicates that it’s premature for businesses to shift from organic search to solely paid traffic sources. Although Google is leaning towards this by increasing ad blocks in search results up to the first four results.

It’s still early to declare the end of organic traffic. Users will continue to find businesses in organic search, so optimizing websites, obtaining backlinks, and improving technical aspects remain essential for maintaining top search positions, generating organic traffic, and leads.

The changes that occurred in the organic search market following the launch of ChatGPT and the subsequent HCU (Helpful Content Update) from Google have been significant and have had a substantial impact on businesses and internet marketing. Let’s examine the main conclusions and their implications for business:

1. Impact of ChatGPT on organic search: Since ChatGPT’s release and its advanced capabilities in 2023, there has been a surge in using AI-generated content on websites. This has led to an overwhelming amount of information online, complicating search engines’ task of processing and ranking content.
2. Google’s response with HCU: In response to the increase in low-quality content, Google implemented HCU, reducing the visibility of sites with templated or AI-generated content and improving the rankings of sites with high-quality, informative content. Many publishers, even those with original content, have experienced significant traffic losses.
3. Winners and losers: Platforms focused on community and forums, such as Reddit and Quora, emerged as major winners from HCU. Their traffic has grown significantly as users seek more in-depth, authentic discussions. This shift highlights users’ interest in content that offers real value and useful information.
4. Adapting to new conditions: Businesses should adjust their content strategy to meet these new standards. This includes creating original, helpful, and well-researched content that meets user needs and Google’s quality standards. Investment in quality content, SEO, and audience engagement is crucial for maintaining high search engine rankings.
5. Development prospects: In a competitive and evolving market, businesses need to keep an eye on new developments, like Google’s Gemini AI model. Integrating new technologies can provide a competitive edge and help adapt to changes in search algorithms.