The only independent scoring platform

The Scoring Platform That
Shows Its Work.

GTM Clarity is the only independent scoring platform on the market with full transparency into how every score is calculated. No black boxes. No vendor lock-in. You see every weight, every signal, every decision — and you can guarantee the outcomes.

100%
Transparent scoring
— every weight visible
5
Data connectors
out of the box
Dual
Model architecture
— yours vs. the data
Self-Training
ML that improves
with every conversion

Transparent Scoring. Guaranteed Outcomes.

Six integrated modules — each fully transparent in how scores are calculated. No black boxes. Every weight, signal, and threshold is visible, configurable, and auditable.

Fit & Engagement Scoring

Score people, accounts, and opportunities on ICP fit (firmographic match, custom attributes) and engagement (behavioral signals with time decay). Visualize everything in a 2×2 matrix.

Buying Group Intelligence

Detect buying group roles automatically from CRM data and email patterns. See completeness scores, identify missing roles, and discover candidates from intent data and email threads.

Self-Training Propensity

XGBoost-powered ML model that predicts conversion at every funnel stage. Trains on your historical data, updates online as conversions happen, and detects drift to trigger retraining.

Dual Model Architecture

The platform maintains a data-driven model while you adjust your own. Compare both side by side — see where your intuition diverges from what the data says.

Activation & Writeback

Push scores back to Salesforce fields, trigger Customer.io segments, and send Slack alerts. Prescriptive recommendations tell your team exactly what to do next.

Contact Discovery

Surface contacts from O365 email threads and Delivr.ai intent data that aren't in your CRM. Auto-create them in Salesforce with buying group associations.

High Fit / Low Engage
34
Nurture & activate
High Fit / High Engage
12
Prioritize now
Low Fit / Low Engage
156
Deprioritize
Low Fit / High Engage
28
Investigate fit
← Engagement → ↑ Fit

See Your Entire Pipeline at a Glance

The fit vs. engagement matrix gives every team a shared language for prioritization. High fit + high engagement accounts get immediate attention. Everything else gets a clear next action.

  • Score people, accounts, and opportunities in one view
  • Configurable weights and thresholds per tenant
  • Engagement decay — recency matters more than volume
  • Tier assignment (A/B/C/D) with custom breakpoints
  • Rollup scoring from person to account to opportunity

From Raw Data to Revenue Action

Five stages transform fragmented CRM, email, and behavioral data into prioritized scoring with prescriptive next steps.

1

Ingest

  • Salesforce (CRM)
  • Customer.io (MAP)
  • Delivr.ai (Intent)
  • O365 (Email/Cal)
  • Fathom (Meetings)
2

Normalize

  • Canonical schema
  • Tenant isolation
  • Unified data layer
  • PostgreSQL + Drizzle
  • OTel tracing
3

Score

  • Fit (ICP match)
  • Engagement (decay)
  • Buying group roles
  • Propensity (ML)
  • Dual model compare
4

Analyze

  • 2×2 matrix viz
  • Buying group map
  • Completeness gaps
  • Model divergence
  • SHAP explanations
5

Activate

  • Salesforce writeback
  • Customer.io segments
  • Slack alerts
  • Gap alerts + discovery
  • Prescriptive actions

Five Connectors. One Unified Score.

Each connector ingests data through a canonical schema with OpenTelemetry tracing, tenant isolation, and encrypted OAuth tokens at rest.

Salesforce

Contacts, accounts, opportunities, contact roles, stages. OAuth 2.0 with JSForce. Full writeback for scores.

Customer.io

Email engagement, campaign data, behavioral events. Segment-based activation for scored audiences.

Delivr.ai

Off-domain behavioral identification, buying group member discovery, per-member engagement depth, and topic monitoring.

Office 365

Meeting attendance, email thread analysis, calendar cadence patterns. Discover contacts not yet in your CRM.

Fathom.video

Meeting transcripts and sentiment analysis. Understand conversation quality and buying signals from calls.

See the Full Buying Committee

GTM Clarity automatically detects buying group roles from Salesforce opportunity contact roles and email patterns. It identifies champions, economic buyers, technical evaluators, and more — then flags missing roles and suggests candidates from Delivr.ai intent data and O365 email threads.

  • Automatic role detection from CRM and email patterns
  • Completeness scoring as a factor in account score and propensity
  • Gap alerts: "Account X is missing [role]. Found 2 candidates."
  • One-click actions: add to CRM, add to sequence, assign to rep
  • Heatmap dashboard showing engagement by role and account
Account: Acme Corp — Buying Group
Champion
Sarah K. — Active
Economic Buyer
James T. — Active
Technical Eval
Missing
End User
Maria L. — Engaged
Influencer
Discovered via Delivr.ai
Blocker
Not identified
Filled Missing Discovered
Dual Model Comparison — Acme Corp
Platform Model
87%
Propensity to close
Your Model
72%
Propensity to close
Divergence: 15 points
Platform model
Your model
0% Top divergence factor: Buying group completeness weight 100%

Two Models. One Truth.

The platform maintains a data-driven propensity model trained on your historical conversion data. You maintain your own adjustable model with custom weights and rules. Run them side by side and see where your intuition diverges from what the data says.

  • XGBoost with SHAP explainability for every prediction
  • Online learning updates as conversions happen in real-time
  • Automatic drift detection triggers full retraining
  • Trains at every funnel stage: MQL, SQL, Opp, Closed-Won
  • Off-domain intent signals weighted as high-value features

Scores Are Useless Without Action

GTM Clarity pushes insights back into the tools your team already uses — with specific recommendations for what to do next.

Salesforce Writeback

Push fit, engagement, and propensity scores directly to Salesforce person, account, and opportunity fields.

  • Custom field mapping per tenant
  • Score updates on every sync cycle
  • Opportunity stage-aware scoring

Customer.io Segments

Automatically segment scored audiences in Customer.io for targeted nurture flows and activation campaigns.

  • Score-based segment creation
  • Tier-driven email sequences
  • Engagement-triggered campaigns

Slack Alerts

Real-time alerts when accounts cross scoring thresholds, buying groups become complete, or propensity spikes.

  • Configurable alert rules
  • Gap alerts with discovery suggestions
  • Prescriptive next-step actions

Deep Dives

Explore the architecture, scoring methodology, and technical details behind GTM Clarity.

Platform Architecture

GTM Clarity is built as a multi-tenant Next.js application with a Python ML sidecar, PostgreSQL data layer, and Temporal workflow orchestration.

Data Connectors

Salesforce (JSForce, OAuth 2.0), Customer.io, Delivr.ai, Office 365, Fathom.video

OAuth 2.0 AES-256-GCM encryption Canonical schema OTel spans per connector

Unified Data Layer

PostgreSQL (Neon) with Drizzle ORM. 18 tables across 17 schema files. Every query filtered by tenantId. Audit trail on all mutations.

PostgreSQL / Neon Drizzle ORM 0.45 Tenant isolation Audit logging

Scoring Engine

Pure-function scoring with zero side effects. Fit dimensions, engagement signals with decay, account/opportunity rollup, and configurable tier assignment.

Pure functions Fit + Engagement Time decay Rollup aggregation

ML Propensity Sidecar

Python FastAPI service running XGBoost for propensity predictions, River for online learning, SHAP for explainability, and MLflow for model tracking.

Python FastAPI XGBoost River (online learning) SHAP MLflow

API & Frontend

tRPC 11 for type-safe APIs with correlation IDs and OpenTelemetry tracing. Next.js 16 dashboard with Clerk multi-tenant auth (RBAC).

Next.js 16 React 19 tRPC 11 Clerk Auth Tailwind 4

Observability

Pino structured JSON logging with automatic secret redaction. OpenTelemetry spans on all tRPC procedures and external services. Health and readiness endpoints.

Pino 9 OpenTelemetry Correlation IDs OTLP export

Scoring Engine

The scoring engine is built as a set of pure functions with zero side effects — no database calls, no API calls, no randomness. This makes scores deterministic, testable, and fast.

Fit Scoring

Measures how well a person or account matches your ideal customer profile. Evaluates firmographic data, custom attributes, and ICP criteria with configurable weights per dimension.

  • Industry, company size, revenue range matching
  • Job title and seniority scoring
  • Technology stack alignment
  • Geographic fit scoring
  • Custom attribute dimensions per tenant

Engagement Scoring

Tracks behavioral signals across multiple channels with time-based decay — recent engagement carries more weight than historical activity.

  • Email opens, clicks, and replies
  • Website visits and content consumption
  • Meeting attendance and calendar patterns
  • Off-domain intent signals from Delivr.ai
  • Configurable decay curves per signal type

Rollup Aggregation

Person-level scores roll up to account and opportunity scores using configurable aggregation rules — max, average, weighted sum, or custom formulas.

  • Person → Account rollup with role weighting
  • Person → Opportunity rollup via contact roles
  • Buying group completeness as a scoring factor
  • Configurable aggregation methods per tenant

Tier Assignment

Scores map to tiers (A/B/C/D) with configurable breakpoints. The matrix visualization plots fit against engagement to create four actionable quadrants.

  • Custom tier thresholds per tenant
  • Separate tiers for people, accounts, opportunities
  • Matrix quadrant assignment with action labels
  • Historical tier tracking for trend analysis

Buying Group Intelligence

B2B deals are won by committees, not individuals. GTM Clarity maps the entire buying group, identifies gaps, and discovers candidates to fill missing roles.

Role Detection

Automatically identifies buying group roles from Salesforce opportunity contact roles, email thread analysis, and engagement pattern clustering.

  • Champion, Economic Buyer, Technical Evaluator
  • End User, Influencer, Blocker, Coach
  • Pattern matching from historical won deals
  • Email-based role inference from O365

Completeness Scoring

Every buying group gets a completeness score that factors into account scoring and propensity modeling. Incomplete groups flag for action.

  • Required roles defined per deal type
  • Completeness % visible on account cards
  • Direct input to propensity model weighting
  • Historical completeness vs. win-rate analysis

Contact Discovery

Surfaces people from Delivr.ai intent data and O365 email threads who aren't in your CRM but match missing buying group roles.

  • Delivr.ai identifies people engaging with your topics
  • O365 surfaces contacts from email threads and meetings
  • Matching algorithm pairs discovered contacts to missing roles
  • One-click add to Salesforce with group association

Gap Alerts

Proactive notifications when a high-value account is missing key buying group roles, with recommended candidates to fill them.

  • "Account X is missing [role]. 2 candidates found."
  • Pushed to Slack and Salesforce
  • Per-member engagement depth from Delivr.ai
  • One-click: add to CRM, add to sequence, assign to rep

ML Propensity Pipeline

A self-training machine learning pipeline that predicts conversion likelihood at every funnel stage, with a dual model architecture that gives you both data-driven truth and adjustable control.

Training Pipeline

XGBoost models trained on your historical conversion data. Hybrid approach: online learning for real-time updates plus periodic full retraining for drift correction.

  • Train at every conversion point: MQL, SQL, Opp, Closed-Won
  • Configurable funnel stages mapped from Salesforce
  • Cold-start strategy for new tenants
  • MLflow for model versioning and tracking

Online Learning

River-based online learning updates the model incrementally as new conversions arrive — keeping predictions fresh without waiting for a full retrain.

  • Real-time model updates on conversion events
  • Feedback loop: scored data becomes training signal
  • Drift detection triggers automatic retraining
  • Continuous accuracy monitoring

Dual Model Architecture

Two models run in parallel: the platform's data-driven model and your adjustable version. Compare them to understand where intuition and data diverge.

  • Platform model trained across all tenant data
  • Customer model with adjustable weights, thresholds, rules
  • Side-by-side comparison view
  • Divergence analysis surfaces key differences

Explainability

Every propensity prediction comes with SHAP explanations showing which features drove the score — making the model transparent and actionable.

  • SHAP values for every prediction
  • Top contributing features ranked
  • Feature importance across the model
  • Delivr.ai intent signals as high-weight features

Security & Compliance

Built with enterprise security from day one — multi-tenant data isolation, encrypted credentials, audit logging, and comprehensive observability.

Multi-Tenant Isolation

Every database query is filtered by tenantId. There is no way to access another tenant's data, models, or configurations.

  • Row-level tenant isolation on all 18 tables
  • Tenant-scoped ML model training
  • Separate encryption keys per tenant
  • RBAC via Clerk with role-based permissions

Encryption & Secrets

All OAuth tokens and API keys are encrypted at rest using AES-256-GCM. Pino logging automatically redacts password, token, secret, and authorization fields.

  • AES-256-GCM encryption for stored credentials
  • Automatic secret redaction in all logs
  • No secrets in code — env-based configuration
  • Input sanitization on all API boundaries

Audit & Observability

Every mutation logs to an audit trail via tRPC middleware. Full OpenTelemetry tracing with correlation IDs that flow from browser to ML sidecar.

  • Audit log table captures all data mutations
  • Correlation IDs on every request (UUID)
  • OpenTelemetry spans on all external services
  • Health and readiness endpoints with degraded mode

Testing & CI

759 automated tests across 63 test suites. CI pipeline with type checking, linting, vulnerability scanning, and security audits on every commit.

  • Vitest unit tests + Playwright E2E tests
  • TypeScript strict mode — no implicit any
  • CI vulnerability scanning on dependencies
  • API rate limiting and input sanitization

Stop Guessing. Start Scoring.

GTM Clarity gives your revenue team a shared, data-driven framework for prioritizing every account in your pipeline.

Get Started →