MailChimp Adds Generative AI Features for Email Marketing
Who will pay for all of the AI model inference costs?
Intuit MailChimp this week announced new generative AI features to help users quickly create more effective marketing emails. The company says the new Email Content Generator is one of 20 MailChimp tools that leverage AI or data science.
The solution is rolling out first in the U.S. to “select customers” starting this week. Email Content Generator leverages one of OpenAI’s GPT models, but MailChimp did not disclose which one. The company also said it had developed AI models in-house for some of its AI services.
How it Works
The new Email Content Generator will create three email copy examples based on a short description. Users can then select their preferred version and add it to their email template and edit it as necessary. The flow is similar to the Wix website copy generator tool introduced in February. MailChimp’s announcement said:
When creating an email campaign in Mailchimp’s new email editor tool, our customers can see three options of copy that match their industry, marketing intent, and tone of voice. Mailchimp customers can give a natural language prompt like, “write an email about our new product launch and offer 15% off orders today only,” and Email Content Generator will create three options that are even more targeted to the goal of their email campaign. Users can select an option generated for them that works best to drag and drop into their campaign, customize, and send to help them reach their audience and sell more.
MailChimp has presumably fine-tuned its GPT AI model implementation to follow email marketing best practices. It is unclear how the model accommodates a user’s brand and tone of voice, but it was mentioned several times.
Some marketers have expressed concern about generative AI solutions because the copy could all start sounding the same, making it harder to differentiate from competitors. If MailChimp applies a specific tone of voice customized to each brand, that may address the differentiation problem.
However, this is more likely referring to generic categories or free-form inputs such as “professional,” “fun,” “sassy,” and so on. You should expect to add your own differentiation and brand language alignment through editing.
Who Will Pay for Generative AI Everywhere?
The clear trend is that any application that requires writing or images will add generative features. While this might have been a competitive advantage a few months ago, it merely gets you competitive parity today. It is amazing how quickly this expectation has changed.
One typically ignored topic is how the companies intend to cover their new AI inference costs. As far as I can recall, with only one exception, the rollouts have uniformly positioned the new features as included in existing fees.
But, with companies such as Picsart generating millions of images and MailChimp soon to be generating millions of paragraphs of email content, the cost will become a drag on earnings. In addition, you will have some users that will “overuse” the service because it is “free to them” (i.e., included in their existing cost).
Phase two of the generative AI “feature explosion” will be higher prices or generative AI fees on top of the base software cost (i.e., higher prices 😀). There may also be a certain amount of use included for free but an upcharge for more extensive use.
The AI writing software solutions typically have a base price that includes a limit on the number of words generated. That is sure to come for the companies that see generative AI as a value-added feature. There is no free lunch.