I signed up for a newsletter last month because the signup page looked clean. The welcome email sounded like it was written for every company on earth — probably because one was.
That is when I started caring about AI in email marketing. Not because AI is useless. Because most people turn it on before they know what the email is supposed to do.
Quick Answer
Use AI to draft subject lines, shorten text you already wrote, or suggest send times on a list you understand. Do not use it to invent your offer, your tone, or a whole sequence from a blank prompt.
The mistake is turning on AI before you have one real email
I have seen small shops plug Mailchimp or Klaviyo into a list of 400 people and ask the tool to personalize everything. The merge tags work. The names show up. The email still reads like it belongs to someone else's business.
Before I touch AI in a campaign, I write one email by hand for one customer type. One problem. One reason to open it. If I cannot do that in ten minutes, AI will only speed up the confusion.
What I would let AI handle
Subject line variants for an email I already approved — maybe three options, not thirty.
A shorter version of a long update I wrote at midnight and now hate.
Segment suggestions when the platform already has purchase or click data I trust.
That is about it for week one.
What I would keep away from AI
Your whole brand voice copied from a competitor.
A five-email launch sequence when you have never sent email one.
Any line about GDPR, consent, or unsubscribe rules. Read ICO guidance on email marketing yourself or pay someone who does compliance for a living. Tools have checkboxes. Checkboxes do not save you if the list was bought or scraped.
Picking between Mailchimp, HubSpot, and the rest
Mailchimp is fine when you want simple sends and you are not running a commerce machine. HubSpot makes more sense when email sits next to CRM notes and you already live in that dashboard. ActiveCampaign shows up when people need branching automations they will actually maintain.
None of that matters if the list is cold and the offer is muddy.
Match the tool to the job you repeat every week — not the feature list on the pricing page.
The number I actually watch
Open rate tells me if the subject line lied. Click rate tells me if the body delivered. Unsubscribes tell me if I should stop sending that angle.
AI dashboards love showing lift percentages. I still pull one campaign, read the email out loud, and ask whether a tired customer would feel tricked. That test beats any score the platform prints.
