At the same time that OpenAI announced the new GPT-4o model, free accounts can now also use custom GPTs created by other users, though with usage limits. This change indicates that more people will start using custom GPTs. They can access them through the GPTs Store, Google searches, or social media channels shared by companies or individuals.

I am confident that custom GPTs will become increasingly important and popular.

In theory, this speculation should hold true. But what about in reality? The data speaks for itself. With the help of Semrush and ChatGPT, I analyzed the overall data of the custom GPTs crawled by Semrush. The results are detailed below. Enjoy!

Research background:

• I analyzed the overall trend of custom GPT pages crawled by Semrush.
• I also examined the top 1,000 custom GPTs in the US by SEO traffic
• All data in this article is based on Semrush, except where noted otherwise.

Here are some important data at a glance:

• Google has indexed over 470,000 custom GPTs pages.
• According to Semrush, SEO traffic accounts for 18% of all custom GPTs crawled.
• There are over 37,000 keywords for custom GPTs ranked in the top 10 on Google.
• The top three countries contributing traffic to custom GPTs are the United States, Japan, and China.
• Among the top 1,000 GPTs in the US, 95.4% use generic keywords as their names.
• The top 1,000 GPTs in the US generate over 290,000 SEO traffic per month, accounting for 88.24% of all custom GPTs in the US.
• Among the top 1,000 GPTs in the US, 51% are used for productivity.

I also discovered some very interesting uses of custom GPTs:

• Ranking for future GPT model names in advance, such as GPT-5, GPT-6, GPT-7, etc.
• Detecting whether content is generated by AI.
• Assisting in the creation of custom GPTs.
• Serving as virtual partners (e.g., virtual girlfriends).
• Searching for other GPTs in GPT stores.
• Educational purposes (e.g., language learning, technology, various subjects).
• Converting files to different formats (e.g., converting WebP to PNG).
• Now, feel free to dive in and enjoy the following detailed data breakdown!

I. Overall Data of Custom GPTs

1. Indexing

• Google has indexed over 470,000 custom GPTs (Google search data, site:https://chat.openai.com/g/).
• Semrush has crawled over 200,000 custom GPTs pages.

2. Traffic Sources

• The largest traffic source for custom GPTs is direct traffic, accounting for 75%, followed by SEO channels, which account for 18%.
• Referral traffic is also significant, reaching 6%.

3. Keyword Ranking Distribution

• Over 37,000 keywords rank in the top 10, accounting for 13.03% of all keywords.
• 42% of the keywords are ranked beyond the 5th page, representing the largest proportion.

4. Top 10 Traffic Source Countries

• The top three countries for traffic are the United States (21.63%), Japan (5.63%), and China (4.69%).
• The top 10 traffic source countries collectively account for 56.9% of the total traffic.

II. Top 1000 Custom GPTs(US)

1. Custom GPT Names

• 95.4% of the top 1000 custom GPTs use generic keywords as their names
• Only 2.6% of custom GPTs use Brand name as their names

Note: Learn more about how to optimize custom GPTs for SEO performance and brand awareness here.

2. SEO Keywords

• The top 1000 custom GPTs collectively carry 29,435 organic keywords.
• One of the most powerful custom GPTs carries up to 1,335 organic keywords.
• 464 custom GPTs have fewer than 10 organic keywords (46.4%).
• 121 custom GPTs have more than 50 organic keywords (12.1%).
• 42 custom GPTs have more than 100 organic keywords (4.2%).

3. SEO Traffic

• The top 1000 custom GPTs generate a total of 293,812 SEO traffic per month, accounting for 88.24% of the overall GPT SEO traffic in the US.
• The top 50 GPTs generate 191,193 SEO traffic per month, accounting for 60.61% of the overall GPT SEO traffic in the US.
• One of the most powrful custom GPTs generates the maximum SEO traffic of 31,756 per month, accounting for 10.05% of the overall GPT SEO traffic in the US.
• 345 custom GPTs have a monthly SEO traffic of more than 100.
• 127 custom GPTs have a monthly SEO traffic of more than 300.

4. Different Purposes of Custom GPTs

• Among the top 1000 GPTs, the main purpose is to improve productivity (51%)
• Followed by solving problems in daily lifestyle (14%)
• Writing (11%) and images creation (10%) are also important scenarios of custom GPTs

5. Prompt Summaries (Based On Top 50 Custom GPTs)

• The longest prompt used in the top 50 custom GPTs is surprisingly over 1700 words.
• It is highly recommended to define the specific functions of each feature (such as data analysis, DALL-E, web browsing) when creating custom GPTs.
• It is necessary to define both the function and the scope of restrictions for custom GPTs to ensure the output meets your needs.
• Instructions should be detailed and clearly structured.
• If available, third-party APIs should be used in the Actions part to enhance custom GPT performance with additional support.

III. How to Write Instructions for Custom GPTs

Finally, based on the analysis of the instructions of the Top 50 custom GPTs, here is what I learned and summarized the best practices about how to write instructions when creating your own Custom GPTs.

Share it with your friends if you think it’s valuable and helpful!

Best Practices for Custom GPTs Instructions

#Role definition
You are a specific version of ChatGPT designed for [specific purpose or field].
#Function
You can [list core functions].
#Background and context
When providing responses, you should consider [relevant background or background information].
#Constraints
You should avoid [specific limitations].
#User interactions
##User engagement methods
Interact with users through [specific engagement methods].
##Response tone
Maintain [specific tone] in your responses.
#Additional features
You integrate [specific tools or data sources].
#Example scenarios
You can help users in the following example scenarios: [list scenarios].

An Example for Creating Custom GPTs for Financial Planning and Advice

#Role definition
You are a special version of ChatGPT designed for financial planning and advice.
#Functionality
You are able to provide detailed financial advice, create budget plans, analyze investment opportunities, and explain complex financial concepts in simple terms.
#Background and context
When providing advice, you should consider the user's financial goals, risk tolerance, and current market conditions.
#Constraints
You should avoid providing legal or tax advice, predicting future market conditions, or recommending specific financial products.
#User interactions
##User engagement methods
Interact with users by asking clarifying questions to better understand their needs and providing step-by-step instructions.
##Response Tone
Maintain a professional, empathetic, and supportive tone in your responses.
#Additional Features
You integrate financial calculators, market data feeds, and investment analysis tools.
#Example Scenarios
You can help users with example scenarios for creating a retirement plan, evaluating a portfolio, and explaining the impact of inflation on savings.