Skip to main content

Scoring Configuration

GTM Clarity scores every entity on two axes -- Fit (how well a prospect matches your ICP) and Engagement (how actively they interact with your brand). These combine into a unified score that places each entity into a quadrant tier.

Scoring configuration is stored per-tenant in the scoring_configs table and is fully customizable.

Scoring Pipeline Overview​

Fit Configuration​

Fit scoring measures how closely an account matches your Ideal Customer Profile. The configuration is defined by the FitConfig type:

ICP Profile​

ParameterTypeDefaultDescription
industriesstring[][]Target industries (exact match)
minEmployeesnumber50Minimum employee count
maxEmployeesnumber1000Maximum employee count
minRevenuenumber1,000,000Minimum annual revenue (USD)
maxRevenuenumber100,000,000Maximum annual revenue (USD)
customAttributesarray[]Additional scored fields

Custom Attributes​

Custom attributes allow scoring on any field in the account data:

{
  "customAttributes": [
    {
      "field": "technology_stack",
      "values": ["Salesforce", "HubSpot", "Marketo"],
      "weight": 15
    },
    {
      "field": "region",
      "values": ["North America", "EMEA"],
      "weight": 10
    }
  ]
}

Fit Dimension Weights​

Weights control how much each dimension contributes to the overall fit score. They are expressed as percentages and should sum to 100:

DimensionDefault WeightDescription
industry30Industry match
companySize25Employee count within range
revenue25Annual revenue within range
customAttributes20Custom field matches

Engagement Configuration​

Engagement scoring measures behavioral signals with time-based decay. The configuration is defined by the EngagementConfig type:

Core Parameters​

ParameterTypeDefaultDescription
decayHalfLifeDaysnumber30Days until signal value halves
signalWeightsRecord<string, number>See belowWeight per activity type
channelCapsRecord<string, number>See belowMax score per channel
qualityHierarchyRecord<string, number>See belowSignal quality ranking (0-1)

Signal Weights​

Each activity type has a configurable weight. Default weights are all 1.0:

SignalDefault WeightDescription
demo_request1.0Demo or trial request
form_fill1.0Form submission
email_click1.0Email link click
email_open1.0Email open
page_view1.0Website page view
ad_click1.0Advertisement click
event_attend1.0Event attendance
content_download1.0Content download
meeting_attend1.0Meeting attendance

Channel Caps​

Channel caps prevent any single channel from dominating the engagement score:

ChannelDefault CapDescription
email30Email interactions
web25Website activity
ads15Advertising engagement
events20Event participation
content20Content consumption
meetings25Meeting engagement
other15Uncategorized

Quality Hierarchy​

The quality hierarchy assigns a multiplier (0-1) reflecting signal intent strength:

SignalQuality ScoreIntent Level
demo_request1.0Highest
event_attend0.9Very High
meeting_attend0.9Very High
form_fill0.8High
content_download0.7Medium-High
email_click0.6Medium
ad_click0.5Medium
email_open0.4Low-Medium
page_view0.2Low

Decay Formula​

Engagement scores decay exponentially over time:

decayed_score = raw_score * e^(-lambda * days_since_activity)

Where lambda = ln(2) / decayHalfLifeDays. With the default 30-day half-life, a signal loses 50% of its value after 30 days and approximately 75% after 60 days.

tip

Set a shorter half-life (e.g., 14 days) for fast-moving sales cycles, or a longer one (e.g., 90 days) for enterprise deals with extended timelines.

Combined Weights​

The combined score blends fit and engagement:

ParameterDefaultDescription
fitWeight1Relative weight for fit
engagementWeight1Relative weight for engagement

The formula: combined = (fitScore * fitWeight + engagementScore * engagementWeight) / (fitWeight + engagementWeight)

Quadrant Thresholds and Tier Assignment​

Entities are assigned to quadrants based on their fit and engagement scores:

QuadrantFitEngagementTier Label
Fast-Track>= threshold>= thresholdhot
Investigate< threshold>= thresholdwarm
Nurture>= threshold< thresholdcool
Deprioritize< threshold< thresholdcold

Default thresholds:

ParameterDefaultDescription
fitThreshold50Fit score boundary
engagementThreshold50Engagement score boundary
hysteresisBand10Points beyond threshold needed to change tier

Hysteresis​

Hysteresis prevents entities from oscillating between tiers when scores hover near thresholds. An entity must move beyond the threshold by the full hysteresis band (default: 10 points) before changing tier. For example, a hot entity must drop below 40 (not just 49) to become cold.

info

Hysteresis is applied automatically by the TierAssigner class and is tracked in the score breakdown as hysteresisApplied: true/false.

Rollup Strategies​

Account and opportunity scores are aggregated from their associated people scores:

Account Rollup​

StrategyDescriptionBest For
championHighest individual score wins (default, weight: 0.6)Small buying groups
weighted-averageAll contacts contribute proportionallyLarge accounts
max-breadthRewards engagement breadth across contacts (cap: 1.5x)Enterprise deals

Opportunity Rollup​

ParameterDefaultDescription
fitSourceinherit-accountInherit fit from parent account
stageVelocityWeight0.1Weight for deal stage velocity
contactEngagementStrategyweighted-averageHow contact scores aggregate

Configuration Storage​

Scoring configs are stored in the scoring_configs table:

ColumnTypeDescription
idtext (UUID)Primary key
tenant_idtextTenant scope
typetextfit, engagement, or combined
configjsonbFull configuration object
versionintegerConfiguration version (default: 1)
is_activebooleanWhether this config is active
warning

Only one active configuration per type per tenant is allowed (enforced by a unique index on tenant_id + type). Changing the config creates a new version -- the old config is deactivated but retained for audit purposes.