The email subject line is the single variable that separates the 100% who open and the 70% who don’t. With your qualified contact base and relevant content, if the subject line doesn’t attract, everything else is wasted — no one gets to see it.
AI has completely changed how subject lines are created and tested. What was once the copywriter’s intuitive work is now a systematic process, driven by data.
The generation process with AI
For each email you need to send, instead of writing a single subject line, you ask AI to generate twenty variations with different angles: direct benefit, curiosity, urgency, question, number, personalisation, contrast, social proof, exclusivity, and fear of missing out.
Twenty variants are created in two minutes. Writing twenty good subject lines by hand takes more than an hour.
Automatic A/B testing in three variants
From the twenty variants created, the team selects the three most promising from different angles. The send is carried out in two phases:
Phase 1: the email is sent to 30% of the list with three variants, divided into equal parts. Each third receives a different subject line.
Phase 2: four hours later, the winning subject line is identified (highest open rate) and sent to the remaining 70% of the list.
This model increases the overall send open rate by between 15% and 35%, compared to sending with a single untested subject line. Most modern email marketing platforms (ActiveCampaign, Klaviyo, Brevo, Mailchimp Pro) have this functionality built in.
Subject line personalisation at scale
The next level is subject line personalisation per contact. With generative AI integrated into the workflow, the subject line can be unique for each recipient, generated at the moment of sending, based on their history.
For a contact who has shown interest in a specific product, the subject line can mention that product. For someone at a particular stage of the funnel, the subject line can reflect that stage. For someone from a specific sector, the subject line can be tailored to that sector.
Dynamic personalisation of subject lines increases the open rate by between 20% and 40%, compared to generic subject lines personalised with just a name.
What doesn’t work
There are templates that seem like good ideas but generate fewer opens or, worse, deliverability issues:
- All-caps subject lines: spam filters penalise them
- Too many emojis: Gmail filters penalise them
- Repeatedly used false urgency: the list gets used to it and ignores it
- Misleading subject lines: they boost open rates once, but destroy trust and increase unsubscribe rates
These limits are respected in AI-generated content if the prompt restrictions are properly configured. For serious implementations, we have the full BAI Marketing system in our Email Predictivo product.