Understanding which products are making money and their revenue scale is crucial for entrepreneurs and investors alike. This article will introduce you to a method of estimating AI product revenues using publicly available data, providing valuable insights for market trends and startup decisions.
The Shortcut to Finding Profitable Products
The method is simple: look at the charts.
This method was figured out a little bit by many entrepreneurs, hats off to them.
For web-based products, you can check the revenue rankings on Toolify and IndieHackers. These rankings compile revenue data from numerous AI tools, offering a quick window into the market.
Note that here, although at first glance it seems to be in accordance with the flow of sorting, as if it has nothing to do with the income, but in fact, just because generally speaking, the greater the flow of the higher income, so some of the products in front of just the flow of rankings and income rankings overlap. Just look at #5 and #6 to see that this list is not a traffic chart.
So where does this list come from in terms of revenue?
It is based on the data from the collection platforms marked with a blue box in the screenshot above.
Diving into Data: Methods for Estimating Revenue
Let’s talk about the principle, when a website is connected to Stripe payments, and a user wants to pay, he or she clicks on the Stripe cashier page from the website to make the payment.
Case Study: ChatGPT
Let’s take for example the ChatGPT website to upgrade to the Plus package.
When I click on the “Upgrade to Plus” button, the following page will open：
As you can see, although the domain name of the page that opens is “pay.openai.com”, you can tell from the “Powered by Stripe” in the bottom left corner that it’s actually Stripe’s page.
The reason it shows up as http://pay.openai.com/is because OpenAI bundles custom domains in the Stripe backend.
So based on the number of visits to the domain http://pay.openai.com/, you can calculate the number of orders for ChatGPT.
At Similarweb you can find out that 6.1 Million of the visits to this domain in June came from external links.
And 99.9% of the traffic in the external link comes from chat.openai.com, so we can assume that the cashier page was opened a total of 6.1 Million times in the month of June, and let’s assume that one time it was opened it was an order, which means that there were 6.1 Million orders.
Let me explain more here, generally when you open Stripe’s cashier page, you have created an order. When the user opens this order, there is a possibility that he/she will not enter any information and just close it, or he/she will enter his/her credit card information and confirm the payment. Here there are two results, successful payment or failed payment, and the failure can be due to various reasons, such as being restricted, insufficient card balance, wrong card information, and so on.
In other words, the number of orders cannot be equated to the number of successful payments. There is a conversion ratio here, and for my own AI tools products, the percentage of successful payments ranges from 10% to 50%.
Let’s assume ChatGPT order payment success rate is 20%, that is, there are 1.22M orders, each average payment of $20, then the total revenue is $24.4M. ChatGPT’s actual payment success rate may be more than 20%.
With this method in hand, we can at least roughly determine what a product’s monthly revenue is.
General Method
Some products don’t have custom domains setup in Stripe, so we don’t have a way to directly look at the traffic to the cashier domain to determine revenue.
At this point there is actually a way to observe that the Stripe register address checkout.stripe.com that opens up after clicking pay, then we can go to checkout.stripe.com and see the traffic.
You can see that this domain has 20 to 30 million visits per month, which means that all the websites that have access to Stripe Payments but don’t have a customized domain set up, in total, have almost 20 million orders per month.
How do you break out the number of orders for each site?
Just now I explain, in the access to the Stripe payment site click will open the Stripe cashier page, this time, “access to the Stripe payment site” is the payment page of the external links site, so you can see the data of each site in Similarweb’s hot door external links site module.
Real-World Examples
Clicking on the link in the bottom right corner opens a new page and we get the order rankings.
1. Roblox
For example, the above is the ranking of the number of orders in June, ranked first is roblox.com, is a Web3 game related website, there are a lot of various kinds of small games on it, the site visits in June is 755.4M, of which the number of orders is 2.8M.
2. Midjourney
Looking at the second product Midjourney is the quintessential AI product, with 515.5k orders in June.
Midjourney offers several packages, let’s assume the average order is $25 and the payment success rate is 40%, then the monthly revenue is $515.5k.
In fact, Midjourney is a subscription-based company, and if users don’t cancel their subscriptions, they will be automatically charged next month, so the $6.7 million can be interpreted as new revenue, and there are still subscriptions from old users every month, so the total revenue is definitely more than what we have calculated.
Check out the tech press report that says Midjourney’s 2023 revenue is $200 million, which works out to an average of $16.66 million per month.
As we can see here, it’s not really that easy for us to estimate the real monthly revenue for such a superb product.
However, in fact, many times we do not need to know the real income, as long as we know the number of orders, we can determine that a certain AI product in the end there is no income, and then go to judge the product corresponding to the demand is probably how big, whether we are worth to enter the market.
So is it completely impossible to know exactly how much an AI product earns per month?
Not really.
Other Ways to Obtain Revenue Information
The first way
There are some developers who follow the “Build in Public” principle and publish the monthly revenue of their products every month as a way to publicize their products.
By following these developers on Twitter, you can find out the real income.
You can even go to Similarweb to see the traffic of his website, so that you can deduce the payment success rate and other information.
The sencond way
In addition, IndieHackers website also has a revenue ranking, the data are real, from the developers authorized IndieHackers website to read the real income of Stripe.
Here is the link.
Conclusion
The methods introduced in this article allow you to roughly understand the revenue scale of an AI product and judge the size of its market demand. While this estimation method may not be precise, it provides valuable information for market research and startup decisions.
Remember, this method isn’t just applicable to AI products; it can be used to analyze other types of online products as well. We hope this information helps you make more informed decisions on your AI entrepreneurship journey.
If you want to dive deeper, check out these links:
Wishing you all the best in your startup journey!
English 
French 
Chinese 
German 
Spanish 
Japanese