Platform · Recursion

A system that gets better with every send.

Recursion watches your campaigns in real time, finds what's working, and tells you exactly what to change, or changes it for you. The longer it runs, the sharper it gets.

Recursion · Learning EngineLive
High impactMessaging

Reframe the subject line around the outcome, not the feature

Subject lines referencing cost savings are converting at 3.2× vs. product-feature lines across this ICP. Try leading with the outcome instead.

+18% open rate projected

Why Recursion flagged this

Analysed 1,240 sends across 6 active sequences
Compared performance by subject-line pattern
Detected 3.2× lift on outcome-framed subjects
Confidence threshold met (>85% across cohort)
How it works

Four steps. One closed loop.

Recursion runs continuously in the background. Every campaign becomes training data for the next one.

01
Ingest

Recursion reads every signal your campaigns produce: reply rates, open rates, click-through, bounce events, positive and negative sentiment in replies, and time-to-reply.

02
Analyse

It compares performance across sequences, subject lines, audiences, and send times. Patterns that humans would take weeks to spot surface in hours.

03
Suggest

Specific, reasoned recommendations land in your dashboard: a new angle for a subject line, a tighter ICP filter, a different send window, a rewrite of the hook.

04
Apply

You approve the change, or let Recursion apply it automatically. Either way, the updated logic feeds straight back into the next send cycle.

What it looks for

Every lever that moves the number.

Messaging

  • Subject line patterns
  • Hook structure
  • CTA phrasing
  • Tone and length

Audience

  • ICP filters
  • Title and seniority
  • Industry segments
  • Company size bands

Sequence

  • Touch count
  • Step spacing
  • Proof-point placement
  • Channel mix

Timing

  • Day of week
  • Time of day (local)
  • Follow-up cadence
  • Pause windows
14 days
Average time to first Recursion-driven improvement
3.2×
Lift in reply rate after 60 days of learning
100%
Of suggestions backed by campaign data, no guessing