Mixpanel Project Topologies (NDE Analytics)

Mixpanel Project Topologies (NDE Analytics)

See Best Practices for Decision Pages and Tags for groups
Legend: not started IN PROGRESS STALLED decided

Status

IN PROGRESS

Description

As part of the migration to Primo NDE, SILS will need to choose one of two options for managing Primo analytics: “Independent institutions” or “Centralized”.

Decision summary

 

Owning group

Discovery

Approver

 

Consulted

 

Informed

 

Decision-making process

Discussion and consensus

Priority

Medium

Target decision date

september 2026

Date decided

[type // to add Date]

Recommendation

Impact

Stakeholder group

Impact

Stakeholder group

Impact

Discovery

 

SILS OT

data retention; user privacy

SILS OC

potential administration

Reasoning

 

Background

Mixpanel is a product analytics tool used to track how users interact with a system like NDE. It captures specific actions (events) such as searches, record views, and link clicks, along with contextual details (properties) like institution, resource type, or search scope. This allows libraries to analyze user behavior patterns rather than just aggregate counts, providing insight into how discovery is actually being used.

In an independent model, each institution manages its own Mixpanel setup, defining its own events and data structure, which allows flexibility but results in data that cannot be easily compared across sites. In a centralized model, institutions agree on shared event definitions and tracking standards, enabling cross-institution analysis and reporting, but requiring coordination and limiting local variation.

Independent vs Centralized Mixpanel

This is a data governance and implementation model decision, not just technical setup.


Independent Mixpanel

Each institution runs its own analytics setup.

Structure:

  • separate Mixpanel projects per IZ

  • each institution defines its own:

    • events

    • properties

    • dashboards

Implications:

  • full local control

  • no required coordination

  • data is not inherently comparable across institutions

Example:

  • Berkeley tracks search_performed

  • Davis tracks searchExecuted
    → same concept, different data → cannot aggregate cleanly


Centralized Mixpanel

All institutions contribute to a shared analytics model.

Structure:

  • shared project or coordinated projects

  • common:

    • event names

    • property definitions

    • tracking logic

Implications:

  • enables cross-institution comparison

  • requires agreement and coordination

  • less local flexibility

Example:

  • all institutions use search_performed

  • all include institution, scope, resource_type
    → data can be aggregated and compared


Core Difference

Aspect

Independent

Centralized

Aspect

Independent

Centralized

Control

Local

Shared

Coordination

None

Required

Data comparability

Low

High

Flexibility

High

Lower

Setup complexity

Lower

Higher

Options Considered

Aspect

Independent Mixpanel

Centralized Mixpanel

Aspect

Independent Mixpanel

Centralized Mixpanel

Setup effort

Lower; each IZ can implement quickly without coordination

Higher; requires upfront agreement on events, properties, and structure

Ongoing maintenance

Decentralized; each IZ maintains its own setup

Shared responsibility; changes require coordination

Flexibility

High; institutions can track whatever they want

Lower; must adhere to shared definitions

Data consistency

Low; events and properties may differ across IZs

High; standardized schema across all institutions

Cross-institution comparison

Difficult or unreliable

Straightforward and meaningful

Governance overhead

Minimal

Significant; requires agreement and change control

Speed of changes

Fast; local decisions only

Slower; changes may need consortium approval

Risk of drift over time

High; schemas diverge further

Lower; shared standards reduce drift

Quality of analytics

Variable; depends on each IZ

More consistent and reliable overall

Troubleshooting

Localized; easier to isolate issues

More complex; issues may affect multiple IZs

Long-term value

Limited for consortium-wide insights

Strong for system-wide analysis and decision-making

 

Dependencies

 

Questions to consider

Ex Libris Documentation

Action Log

Action/Point Person

Expected Completion Date

Notes

Status

Action/Point Person

Expected Completion Date

Notes

Status

 

 

 

 

 

 

 

 

The SILS mission is to transform library services and operations through innovation and collaboration. The future is shared!
Question? Contact AskSILS-L@ucop.edu