Relationship Graph Interpretation Guide
You have computed relationship data including clusters, bridges, and contact tiers. Your job is to interpret this data meaningfully—find the story in the numbers.
Understanding Contact Scoring
Longevity-First Philosophy
The scoring algorithm intentionally values consistency and depth over recency. This means:
- A contact you've emailed monthly for 2 years but not in 3 months = HIGH score
- A contact you just discovered and exchanged 5 emails with this week = LOWER score
Score Components
- Consistency (0-30): Emails spread across many months beats burst patterns
- Historical Peak (0-20): Past engagement matters even if currently dormant
- Relationship Age (0-15): Multi-year relationships score higher
- Recency (0-15): Bonus for active, but NO penalty for dormant
- Depth (0-20): Back-and-forth conversations, thread depth
Contact Statuses
- Active: Recent communication (last 180 days)
- Dormant: No recent communication, lower overall score
- Dormant-Important: No recent communication, BUT high score—these are lapsed important relationships
Interpreting Clusters
Domain-Based Clusters
Contacts grouped by email domain often represent:
- Work clusters: People at the same company
- Organization clusters: School, church, club membership
- Family clusters: Shared family domain (rare but exists)
What to look for:
- Cluster size: Large clusters suggest significant organizational involvement
- Multiple work clusters: Career changes, consulting, multiple affiliations
- Small focused clusters: Tight-knit groups
Interpreting Cluster Names
The cluster name comes from the domain. Translate for the user:
acmecluster → "Your Acme colleagues"stanfordcluster → "Your Stanford connections"familynamecluster → Possibly family members
Bridge Contacts
Bridge contacts connect multiple clusters—they know people in different parts of the user's life.
What bridges indicate:
- Social connectors: People who introduce others
- Multi-role relationships: Colleague who became friend
- Transitional contacts: Connected during career/life change
How to present:
"[Name] bridges your [Cluster A] and [Cluster B] networks—they might be how these worlds connect."
Contact Tiers
Inner Circle (Top 5 by Score)
These are the user's most significant email relationships based on consistency, depth, and history.
Key insight: These might not be the most RECENT contacts—they're the most SUSTAINED.
How to present:
"Your strongest email relationships are with [names]. These aren't just frequent contacts—they're consistent over time."
Dormant-Important
High score but inactive for 6+ months. These are lapsed important relationships.
What they represent:
- Old close colleagues who changed jobs
- Friends who moved away
- Professional relationships that went quiet
- Collaborators from past projects
Opportunity: These are people the user SHOULD reconnect with. The relationship existed; it just went quiet.
How to present:
"You have [N] dormant-important contacts—people you had strong relationships with who've gone quiet. [Name] hasn't been in touch since [date] but you exchanged [N] emails over [N] months. Worth reconnecting?"
Regular Contacts
Active mid-tier relationships. These are the "working network"—people in regular rotation.
Occasional Contacts
Lower frequency but still meaningful. Don't dismiss these—sometimes the best relationships don't require constant contact.
Patterns to Identify
Relationship Balance
- One-way relationships: User receives but doesn't reply, or sends without response
- Mutual relationships: Bidirectional flag in data
What imbalance might mean:
- Newsletters/notifications (expected one-way)
- Unanswered outreach (possible concern)
- Ignored requests (relationship health)
Conversation Depth
- Thread depth: How many back-and-forths per conversation
- Long threads with specific people: Deep collaboration or personal connection
- Short threads: Transactional relationships
Temporal Patterns
- Burst patterns: Suddenly active then quiet (project-based relationship)
- Steady patterns: Regular contact over time (ongoing relationship)
- Seasonal patterns: Holiday cards, annual check-ins
Presentation Approach
Lead with Insights, Not Data
Don't just list clusters and scores. Tell the user something they might not know:
Bad: "You have 3 clusters: acme (12 contacts), bigcorp (8 contacts), school (5 contacts)"
Good: "Your email life has three main worlds: your current team at Acme (12 people), connections from your BigCorp days (8 people still in touch), and a small but consistent group from school (5 people you've stayed connected with for years)."
Highlight Surprises
- Inner circle members the user might not consciously recognize as "important"
- Dormant-important contacts they may have forgotten
- Bridges that explain how different parts of their life connect
Be Tactful About Gaps
- Don't point out that someone "should" be in their inner circle
- Don't judge relationship frequency
- Present data neutrally; let user interpret meaning
Writing the Analysis
Structure your relationship analysis as:
## Your Email Network
### Inner Circle
[Describe top relationships with context about why they score high]
### Your Worlds
[Describe clusters as "worlds" the user operates in]
### Bridges
[If meaningful bridges exist, explain how they connect worlds]
### Worth Reconnecting?
[List 2-3 dormant-important contacts with brief context]When Data is Limited
If the user has:
- Few contacts: Note that analysis is limited; they might not use email for primary communication
- No clusters: Could mean freelancer, remote worker, or non-organizational email use
- All dormant: Either email isn't their channel, or there's a communication gap to address