Generative AI will revolutionize online search. The new search engine Bing gives a glimpse of the possibilities offered to Internet users. However, this potential represents a huge danger for some categories of websites that will become obsolete. If your business model is based on inbound marketing, read this article carefully to understand if your business model is in danger. Don’t wait until you lose all your traffic to react. Solutions exist.
- Information and entertainment websites (infotainment)
- Recipe websites
- Technical forums
- Translation websites
- Product testing websites
- Search engines
- Weather websites
- Outsourcing platforms
- Image websites
According to Microsoft, based on generative AI, 50% of online searches could be done via a conversational agent (ChatBOT). ChatGPT is the perfect example. The consequences would be terrible for all websites that use SEO as their main lead-generation source. It would mean that outbound traffic would be in free fall and that Internet users would no longer have to leave the search engine. However, we have to qualify Microsoft’s radical statements by reminding that the context of a search also determines the added value of a classic search engine. I have explained here why I don’t believe in a massive conversion of Internet users to conversational agents such as ChatGPT and Bing.
This being said, we still have to recognize the threat and know how to anticipate it.
Information and entertainment websites (infotainment)
In general, free news websites are on the verge of collapse. Their business model is based on selling advertising space in exchange for the promise of high traffic for advertisers. Buzzfeed was the first infotainment website to announce its intention to use ChatGPT for content creation (much more serious publishers like Axel Springer followed later). Buzzfeed carried out very early that the little added value of its content made its business model extremely vulnerable to generative AI.
Currently, Buzzfeed is still ahead of the game because its journalists stick to the news, and the dataset used by ChatGPT ends in December 2021. But the new Bing, and its Prometheus technology, incorporates the most recent data.
Recipe websites have had their golden age. ChatGPT is an element of disruption.
What’s the point of going to a recipe website when ChatGPT can provide you with a much more personalized service? Tell it what you have in the fridge, and it tells you what to do with it. When I think that a few years ago, applications were developed to do exactly the same thing.
For technical topics, self-help forums are an inexhaustible source of information. All the big technology companies have them, and some have even put some members in charge of moderation.
For example, we have seen the capabilities of GPT-3 to produce ready-to-use code. Now we have to ask ourselves. What is the use of these forums when it is possible to obtain a detailed answer to all problems, even the most complex ones?
What is the future of translation websites? Deepl was already the winner when it was released. The user can become more efficient with ChatGPT, Bing, and others. Imagine having to translate into multiple languages and format text. Now you can do it in one step. Here’s how I now translate my newsletter editorial in one go by asking ChatGPT to use deepl.com and structure the results for me in a tableau.
Product testing websites
Very popular websites have been built around product benchmarking. These websites are necessarily very topical. Their core business is to test and publish their benchmark results as quickly as possible. Websites like pcmag.com or tomsguide.com have, for example, more than 20 million monthly visitors.
Their business model is a mix of advertising and affiliation with e-commerce websites. Their content is free, and this is their Achilles heel. Therefore, their test results are shared for free, allowing the new Bing to provide its benchmark (see the screenshot above). It’s amazing, even if there are still some factual errors. So, it would be best if you checked the information provided.
Many Internet users will likely seek simplification as technology improves by turning to generative AI. The new Bing is thus capable of integrating ultra-fresh information into its results. The comparison below was proposed on Twitter only 30 minutes after publicizing the VR headset “Bigscreen Beyond” data.
Compare interest rates, energy prices, credits, and airfares, … thousands of websites have made it their business. Here are some examples:
- com (insurance): 4 million monthly visitors
- com (airline tickets): 20 million monthly visitors
- com (interest rates): 17 million monthly visitors
Fed directly by the most up-to-date information sources, generative AI can propose comparisons in the blink of an eye. This is what it looks like for Indian bank interest rates (source).
Here is a test I carried out to compare the interest rates of real estate loans in France.
This is a huge business. The keyword “weather” generates nearly 100,000,000 queries annually in the United States. And one can wonder what will happen to all the websites that sell it (through targeted advertising). In addition to banner ads, did you know that these websites sell your data as a core business? The applications, in particular, sell your geolocation data, making tracking your movements possible thanks to advertising cookies.
But tomorrow, the future of the weather could look like this.
Outsourcing platforms (Upwork, …) will suffer because algorithms will replace many tasks performed today by freelancers. Customers will therefore do without humans in low-cost countries to automate these tasks.
Content creation is the first victim with content (often very similar) created by thousands effortlessly by ChatGPT. But other jobs, especially in visual creation, may disappear with generative AI like the one commercialized by MidJourney.
The combination of GPT3 with Midjourney yields surprising results, as Ethan Mollick showed on Twitter (see screenshot below).
I tried branding using Bing and Midjourney. I asked Bing to propose collaborations between Louis Vuitton and artists and describe the resulting bags. I then used this description to generate the design in MidJourney. The resulting images look quite convincing to me. The result is simply amazing.
As an extension of the above example, image websites could also decrease their market share. Today you must be very careful about the images you put on your website to avoid troll copyrights like PicRights. Buying images from websites like Shutterstock or iStock will protect you. But why pay for images when artificial intelligence can generate them for you?
Look at the images I had Midjourney generate. Aren’t they as qualitative and aesthetic as some of the ones you can buy on these websites?
Posted in Marketing.