The Outcome Ledger: How SI Learns
The memory system that makes SI smarter with every interaction
What It Is
The Outcome Ledger is SI's memory. Every time SI takes an action — sends an email, fires an SMS, triggers a call, sends a micro-conversation, escalates to a human, or even decides to skip — the action is recorded in an append-only log with complete details.
Each entry captures: the channel used, the SI flag that triggered the action (scheduled cadence, signal response, re-engagement, proactive outreach), the assets shared (content articles, videos, testimonials), the RK items referenced (talking points, conversation starters, objections), a 1-2 sentence summary of the approach, the contact's engagement score at the time of the decision, and — critically — the outcome.
Outcomes are assessed automatically after the action: did the contact open the email? Click a link? Reply? Ignore it? Did they answer the phone call? Was the handoff acted upon by the human team member?
This ledger feeds directly back into SI's next decision cycle. When SI assembles context for a contact, the outcome history is front and center — showing what has been tried, what worked, and what failed.
Why It Matters
The Outcome Ledger is what separates SI from every other automation system. Most tools send messages and forget. SI sends messages and *remembers*.
When SI sees that the last three emails to a contact were ignored but the one SMS got a reply, it knows to shift channels. When it sees that content article #42 was already sent in February and got no engagement, it never shares it again. When it sees that a micro-conversation referencing a specific Talking Point got a click-through, it knows that angle resonates with this contact.
Over time, SI builds a complete picture of what works for each individual contact. The more interactions it records, the better its decisions become. A contact with 20 entries in the outcome ledger gets more precisely targeted communication than one with 2 entries — because SI has more data about what resonates.
How It Works
What Gets Recorded
Every outcome ledger entry includes:
- Channel — email, SMS, call, micro-conversation, or human handoff
- SI Flag — what triggered the action (e.g., nurture_silent, proactive_engage, re_engage, signal_response, form_action)
- Assets Used — which content, videos, testimonials, or graphics were shared
- RK Items Used — which talking points, conversation starters, rapport questions, or objection responses were referenced
- Approach Summary — a 1-2 sentence description of the strategy (e.g., "Sent a value-driven email referencing the ROI calculator, paired with a soft CTA to book a demo")
- Engagement Score at Decision Time — a snapshot of the contact's score when SI made the decision
- Outcome — the result
Outcome Assessment
Outcomes are assessed automatically at different intervals depending on the channel:
- Email: Replied (immediate), Clicked (immediate), Opened (immediate), Bounced (72h), Ignored (72h with no opens)
- SMS: Replied (immediate), Bounced (immediate), Ignored (48h with no reply)
- Call: Answered (immediate based on call duration), Missed (48h with no record or zero duration)
- Human Handoff: Actioned (recommendation changed by a team member), Ignored (7 days with no change)
How History Feeds Forward
When SI assembles context for a contact, it includes two critical arrays from the outcome ledger:
- Assets previously used — a deduplicated list of every content, testimonial, and video ever shared with this contact
- RK items previously used — a deduplicated list of every knowledge base item ever referenced
SI is instructed to never share an asset from the used list. It will not reuse an RK item within 90 days. And if the same general approach was tried in the last 30 days, it chooses a different strategy.
This is the anti-repetition system that keeps communication fresh across months of engagement.
