Post by sinthiya007 on Nov 9, 2024 23:09:19 GMT -5
Common phrases, lots of fluff, a sheet of text, no emojis or buttons. In addition, ChatGPT couldn't decide how to address the reader - "you" or "you"
The neural network text is long and looks more like an email newsletter. And texts for WhatsApp should be short: according to our data, messages up to 100 characters have the highest Response Rate — 83% .
In WhatsApp mailings, it is advisable to refer to the experience of past interactions with the user so that the person understands who is writing to him, remembers the company and does not block the mailing.
The main information in a messenger newsletter should be revealed in the first two sentences, and not spread across a sheet of text.
The CTA should be shorter and clearer. For example: “Order products on sale right now” + a button that you can click to go to the Yves Rocher online store.
We tried different types of requests: we told on page seo service the neural network about the context and goals of the newsletter, about the Yves Rocher company, asked to write more specifically and briefly, but still got nonsense. Finally, we threw out all the other tips and asked to add emoji. And suddenly everything became much better:
We conducted several more experiments for different business tasks:
reactivation of "dormant" clients of a clothing brand
mailing with discounts to boost sales of home textiles
In each case, we tried different approaches to formulating the request, everything was described in a detailed guide. We explained what messengers are and how to write for them, indicated the target audience, the context of the mailing, the style, the company's USP, the purpose of the messages. And here is the conclusion we came to:
ChatGPT doesn't know how to write for messengers, doesn't understand what a WhatsApp mailing is, and uses emoji correctly every other time.
The neural network cannot find out and analyze information about the brand on its own. And if you give it this information, it will not be able to pack it into a ready-made message for WhatsApp.
Automate communication with chatbots
Visual editor for creating a chatbot without programming skills
Find out more
textBack
ChatGPT for Online Education Mailings
We decided to test whether ChatGPT could invite an audience to an online event based on the information on the link . We took as a basis the mailing list from Educated Online, a platform for teaching children in schools in the USA, England and Canada. The goal is to bring people to a free conference on education in Dubai.
The neural network failed to embed the information from the link into its message and overdid it with the CTA. We conducted another experiment to create a chain of messages for an online course and things went better here. Here's what we found out in the end:
ChatGPT can create a chain of messages with buttons, but you need to specify the conditions in as much detail as possible. The word "chain" is not clea
The neural network text is long and looks more like an email newsletter. And texts for WhatsApp should be short: according to our data, messages up to 100 characters have the highest Response Rate — 83% .
In WhatsApp mailings, it is advisable to refer to the experience of past interactions with the user so that the person understands who is writing to him, remembers the company and does not block the mailing.
The main information in a messenger newsletter should be revealed in the first two sentences, and not spread across a sheet of text.
The CTA should be shorter and clearer. For example: “Order products on sale right now” + a button that you can click to go to the Yves Rocher online store.
We tried different types of requests: we told on page seo service the neural network about the context and goals of the newsletter, about the Yves Rocher company, asked to write more specifically and briefly, but still got nonsense. Finally, we threw out all the other tips and asked to add emoji. And suddenly everything became much better:
We conducted several more experiments for different business tasks:
reactivation of "dormant" clients of a clothing brand
mailing with discounts to boost sales of home textiles
In each case, we tried different approaches to formulating the request, everything was described in a detailed guide. We explained what messengers are and how to write for them, indicated the target audience, the context of the mailing, the style, the company's USP, the purpose of the messages. And here is the conclusion we came to:
ChatGPT doesn't know how to write for messengers, doesn't understand what a WhatsApp mailing is, and uses emoji correctly every other time.
The neural network cannot find out and analyze information about the brand on its own. And if you give it this information, it will not be able to pack it into a ready-made message for WhatsApp.
Automate communication with chatbots
Visual editor for creating a chatbot without programming skills
Find out more
textBack
ChatGPT for Online Education Mailings
We decided to test whether ChatGPT could invite an audience to an online event based on the information on the link . We took as a basis the mailing list from Educated Online, a platform for teaching children in schools in the USA, England and Canada. The goal is to bring people to a free conference on education in Dubai.
The neural network failed to embed the information from the link into its message and overdid it with the CTA. We conducted another experiment to create a chain of messages for an online course and things went better here. Here's what we found out in the end:
ChatGPT can create a chain of messages with buttons, but you need to specify the conditions in as much detail as possible. The word "chain" is not clea
