Case Study 06 / ABM Programs
A GTM intelligence engine that scores accounts on evidence, and never sends without a human
An internal GTM intelligence platform takes account signals from the outside world, scores accounts with deterministic, explainable rules, uses AI for research and first drafts, and requires a person to read and approve every message before it counts as sent. Every fact in the system traces back to a source, and every outcome is recorded against the activation that produced it.
Founder track record: this work was led by Clarafold's founder before founding Clarafold, while working as marketing manager at a maritime compliance software company. The client's name is withheld until it approves publication; every figure comes from its own systems. Clarafold client engagements are added as they complete.
- of outreach human-approved
- 100%
- autonomous sends
- 0
- build phases, source to attribution
- 6
- stages in the evidence chain
- 7
01 · Client
Maritime compliance software company
Maritime compliance software and services company selling into ship owners, operators and managers. Name withheld until the client approves publication.
The challenge
In a market of a few thousand relevant accounts, knowing which ship managers to approach this quarter, and why, matters more than the volume of outreach. Account research lived in people's heads and browser tabs, scoring was a matter of opinion, and AI tools that generate outreach at scale were the wrong answer for buyers who notice a generic message instantly.
02 · System
What existed before.
- Account lists in spreadsheets with no record of why an account was prioritised
- Research repeated from scratch for every meeting
- No link between what was sent and what happened next
03 · Intervention
What was built and run.
Designed the evidence chain first
Source → fact → signal → score → intelligence → activation → outcome. Nothing enters the score without a source, and source quality is classified.
Kept scoring deterministic
Account scores come from explicit, inspectable rules rather than a model's judgment, so anyone can see why an account ranks where it does.
Used AI where it is strong
Account research, summaries and first drafts of LinkedIn messages, emails, sales briefs and content recommendations, with caching, freshness checks and a hard research budget.
Made human approval structural
There is no sending integration. A draft becomes 'sent' only when a person reads it, edits it if needed and approves it.
Recorded outcomes honestly
Outcomes are logged manually against a frozen snapshot of the context each message was approved in. Analytics stay descriptive, never auto-ranked.
04 · Outcome
What changed.
A working platform covering account and contact management, CSV import, deterministic scoring, AI research with source-quality classification, human-approved activation and outcome attribution.
Every account score can be explained in plain language, and every claim in a sales brief links back to where it came from.
The design principle Clarafold now applies to every engagement: AI for execution, people for judgment.
05 · Lessons
What we would tell you.
- Explainable beats clever in account scoring. A sales team will act on a score it can understand and argue with.
- Build the attribution model before you need it. Recording outcomes against the exact context of each message is what makes learning possible later.
- Speed is not the constraint in a small market. Relevance is.
Put It To Work