How does AI personalization in email marketing actually work?

Short answer

Most platforms use AI to help write one message faster. AeroCampaign's recommendation engine instead reads each contact's signal history — open and click behaviour, recency, list origin, CRM stage, firmographics — clusters contacts into behavioural cohorts, grounds the message in what is moving in that contact's industry, and drafts a distinct subject line and body per cohort, down to the individual where the signal justifies the cost.

Last reviewed 2026-08-24 · AeroCampaign
Key facts
  • Cohorts form from behaviour rather than from manually maintained field values.
  • Cohorts are recomputed on every send, so contacts move as their behaviour moves.
  • Generation happens at three resolutions: cohort, micro-cohort, and individual.
  • Every generated variant records the signals that produced it.
  • Nothing sends without human approval; variants land as editable drafts.
  • The engine writes against the same suppression set the sender enforces.
In detail

Generating a unique email for all 200,000 contacts on a list is technically possible and almost always wasteful. The useful question is which segments have enough signal to justify their own message. The engine chooses its resolution per segment on that basis.

Attribution matters as much as generation. Because each variant carries the signals that produced it, a lift or a flop is traceable to a reason rather than to a hunch about tone.

Get air cover for your next campaign.

Connect a provider, import a list, and send in under three minutes. Free to start, and your list size never changes the bill.

No credit card. Bring your own SendGrid or Amazon SES account.