Engagement Scoring

Engagement scoring measures how actively a person interacts with your brand across multiple channels. It applies exponential decay for recency, a quality hierarchy to weight high-value actions above passive ones, and per-channel caps to prevent single-source score domination.

Core Concepts

The engagement engine processes a person's activity history through three layers:

1. Signal quality --- Each activity type has a quality multiplier reflecting its intent signal strength
2. Exponential decay --- Recent activities contribute more than older ones
3. Channel caps --- No single channel can contribute more than its cap to the total score

Signal Quality Hierarchy

Every activity type has a quality multiplier between 0.0 and 1.0. Higher multipliers indicate stronger buying intent:

Activity Type	Quality Multiplier	Channel	Intent Signal
demo_request	1.0	web	Highest
event_attend	0.9	events	Very high
meeting_attend	0.9	meetings	Very high
form_fill	0.8	web	High
content_download	0.7	content	Medium-high
intent_signal	0.7	intent	Medium-high
email_click	0.6	email	Medium
ad_click	0.5	ads	Medium
email_thread	0.5	email	Medium
email_open	0.4	email	Low-medium
page_view	0.2	web	Low
email_sent	0.1	email	Minimal

Excluded Signal: meeting_recorded

The meeting_recorded activity type is intentionally excluded from scoring. Recording a meeting is a system action, not a buying signal. This avoids inflating engagement for accounts using Fathom.video.

Unknown activity types default to a quality multiplier of 0.2 and are assigned to the "other" channel.

Exponential Decay

Activity scores decrease over time following an exponential decay curve. The formula is:

decayMultiplier = 2^(-daysSince / halfLifeDays)

Where:

• daysSince is the number of days between the activity and the reference date
• halfLifeDays is the configurable half-life (default: 30 days)

Decay Curve Characteristics

Days Since Activity	Decay Multiplier (30-day half-life)
0 (today)	1.000
15	0.707
30	0.500
60	0.250
90	0.125
120	0.063

Choosing Your Half-Life

• Short sales cycles (< 30 days): Use a half-life of 14--21 days to emphasize very recent activity
• Medium sales cycles (30--90 days): The default of 30 days works well
• Long enterprise cycles (90+ days): Consider 45--60 days to keep older signals relevant

Edge Cases

• Future activities (negative daysSince): Treated as day 0, no decay applied
• Half-life of 0: Returns a decay multiplier of 0 (all signals discarded)
• The multiplier is always clamped to [0, 1]

Per-Channel Caps

Each scoring channel has a maximum score contribution to prevent a single data source from dominating:

Channel	Default Cap	Mapped Activity Types
email	30	email_open, email_click, email_thread, email_sent
web	25	page_view, form_fill, demo_request
meetings	25	meeting_attend
intent	25	intent_signal
events	20	event_attend
content	20	content_download
ads	15	ad_click
other	15	Any unmapped activity type

The theoretical maximum engagement score is the sum of all channel caps: 175, but since the final score is clamped to 100, multi-channel engagement is what drives a person toward the top of the scale.

Scoring Formula

For each activity, the per-signal score is:

signalScore = signalWeight * qualityMultiplier * decayMultiplier

Signals are grouped by channel, summed within each channel, and capped:

channelScore = min(sum(signalScores), channelCap)

The final engagement score is:

engagementScore = min(100, sum(channelScores))

Worked Example

A person has three activities:

Activity	Days Ago	Quality	Decay (30d)	Signal Score
demo_request	5	1.0	0.891	0.891
email_click	10	0.6	0.794	0.476
page_view	45	0.2	0.354	0.071

Channel aggregation:

• web: demo_request (0.891) + page_view (0.071) = 0.962 (cap 25: capped at 0.962)
• email: email_click (0.476) (cap 30: capped at 0.476)

Total engagement score: 0.962 + 0.476 = 1.4 (low because only 3 activities)

Real-World Scores

In practice, active prospects have dozens or hundreds of activities. A person with 20+ activities across multiple channels over the past 30 days will typically score 40--70.

Configuration Reference

interface EngagementConfig {
  decayHalfLifeDays: number;           // default: 30
  signalWeights: Record<string, number>;  // default: all 1.0
  channelCaps: Record<string, number>;    // see table above
  qualityHierarchy: Record<string, number>; // see table above
}

Signal Weights

Signal weights (signalWeights) are multipliers applied before quality and decay. They default to 1.0 for all activity types. You can increase weights for high-value signals or decrease them to dampen noisy sources:

{
  "signalWeights": {
    "demo_request": 2.0,
    "page_view": 0.5
  }
}

This would double the contribution of demo requests and halve page views.

Related Pages

• Scoring Overview --- Where engagement fits in the pipeline
• Fit Scoring --- The other scoring axis
• Tier Assignment --- How engagement + fit determine tiers
• Account Scoring --- How person engagement rolls up