Tribar StudioCareers
Back to all tribes

Pulse

Everything measurable. Analytics. Experiments. A/B testing. ML. Predictions.

Mission

Pulse answers "Did it actually work?" instead of "I think it worked."

Pulse is the empirical conscience of Tribar. They design experiments, measure outcomes, build models, and ensure every decision is grounded in data. They turn "I think" into "we know." Pulse measures across three layers: intrapersonal (individual feelings, habits, margin), interpersonal (team dynamics, collaboration quality), and collective (system-wide health, market position).

Did it actually work?

Pulse's founding question — replacing opinion with evidence.

What we measure

Intrapersonal

Individual habits and margin — is the product becoming part of the user's routine, or is it a chore?

How: DAU/WAU/MAU ratios, session frequency distribution, power user curve, "time-to-ritual" (days until daily active)Where: Product analytics dashboards, cohort retention reportsWhen: Daily for core metrics; weekly for cohort analysis; monthly for power user analysisSignal: Power user ratio (top 10% by session count) declines for 2 consecutive weeks, or "time-to-ritual" exceeds 14 daysInstrument: Amplitude/Mixpanel, cohort retention model, power user segmentation, habit loop tracking

Emotional trajectory — how does the user's sentiment change over their lifecycle?

How: Longitudinal sentiment tracking (per-user NPS over time), emotion tagging on support interactions, churn prediction scoresWhere: User-level sentiment timeline, support ticket sentiment pipeline, churn risk modelWhen: Per interaction; weekly cohort sentiment summary; monthly lifecycle analysisSignal: Sentiment decline of >0.5 on a 10-point scale over 30 days for a cohortInstrument: CSAT surveys, NLP sentiment model on tickets, churn prediction model (ML)

Cognitive load — how much mental effort does the product demand?

How: Task completion rate, time-on-task, error rate, undo/redo frequency, help article views per taskWhere: Per-task analytics, session replay tags, help center analyticsWhen: Per task type; reviewed weekly for new flows; monthly for existing flowsSignal: Any critical task falls below 70% completion rate or average time exceeds 2x historical baselineInstrument: RUM data, custom analytics events, help center click tracking, session replay AI tags
Interpersonal

Collaboration quality — how well do teams using Tribar products work together?

How: Shared resource creation rate, comment-to-resolution time, collaboration graph density, cross-team feature adoptionWhere: Collaboration-specific analytics, shared workspace metrics, network analysis of user interactionsWhen: Weekly for active features; monthly for network analysisSignal: Collaboration graph density decreases, or comment-to-resolution time increases >50%Instrument: Collaboration event pipeline, network graph analysis, response-time tracking

Feedback loops — are users responding to each other's contributions?

How: Reply rate, reaction rate, thread depth, co-editing frequency, @mention densityWhere: Real-time collaboration analyticsWhen: WeeklySignal: Reply rate drops below 30%, or average thread depth < 2Instrument: Social graph analysis, interaction density metrics, real-time events pipeline
Collective

System-level health — is the overall ecosystem growing in the right direction?

How: Overall retention curves (Day 1/7/30/90), North Star metric progression, segment growth rates, ecosystem network effectsWhere: Executive dashboard, investor metrics deck, quarterly business reviewWhen: Daily for North Star; weekly for retention curves; monthly for segment growth; quarterly for ecosystem analysisSignal: North Star metric plateaus for >30 days, or any segment shows negative growth for 2 consecutive monthsInstrument: North Star metric dashboard, retention curve model, segment growth tracker, network effect coefficient model

Experimental velocity — how many decisions are empirically grounded?

How: Number of experiments run per sprint, percentage of decisions informed by experiment, experiment-to-change ratioWhere: Pulse experiment registry, decision log, product management dashboardWhen: Per sprint; reviewed quarterlySignal: <50% of significant product decisions have experimental backingInstrument: Experiment registry, decision impact tracker, experimental coverage dashboard

Model drift and prediction accuracy — are our ML models still reliable?

How: Prediction accuracy over time, feature importance stability, data distribution shifts (PSI), feedback loop latencyWhere: ML model monitoring dashboard, data quality metricsWhen: Continuous for production models; weekly review; monthly deep diveSignal: Model accuracy drops >5% or PSI exceeds 0.2 — retraining triggered automaticallyInstrument: ML monitoring platform, data quality pipeline, automated retraining triggers

Rhythms

Metric review

Weekly

60 minutes. Pulse walks through every North Star metric and sub-metric. Is anything moving unexpectedly? Delta and Atlas attend.

Catches problems early. No metric goes unexamined for more than a week.

Experiment council

Weekly

30 minutes. Every proposed experiment is reviewed by Pulse for statistical validity, sample size, and ethical considerations.

Prevents bad experiments from wasting time or misleading the company.

Causal review

Monthly

Pulse reviews every "significant" metric movement from the past month and determines: is this causal or correlational?

Prevents spurious correlations from driving strategy.

Model audit

Monthly

Every production ML model is audited for accuracy, bias, drift, and fairness.

Models degrade. Pulse catches it before users feel it.

Data literacy workshop

Quarterly

Pulse trains a different tribe on experimental design, statistical literacy, and metrics interpretation.

Data literacy is not optional at Tribar.

Tools & artifacts

Metric tree

framework

The complete hierarchy of every metric Tribar tracks — from North Star through input metrics to diagnostic counters. Every metric has a definition, owner, and review cadence.

PulseAtlasEcho

Experiment platform

platform

Self-serve A/B testing platform with statistical engine, sample size calculator, and automated result reporting.

PulseDelta

ML model registry

platform

Every production model registered with metadata: version, training data range, accuracy, drift threshold, owner.

PulseDelta

Dashboards

platform

Role-specific dashboards for every tribe. Executives see North Star + key results. Teams see input metrics. Engineers see system health.

PulseAll tribes

Decision log

document

Every significant decision recorded with: what was decided, what evidence supported it, what experiment validated it, what we expected to happen, what actually happened.

PulseAtlasThread

Responsibilities

  • Analytics infrastructure and data pipelines
  • A/B testing and experimentation platform
  • Machine learning and predictive models
  • Metrics definition and tracking across all tribes
  • Dashboards, reporting, and alerting
  • Statistical rigor and methodological review
  • Behavioral modeling and user segmentation analytics
  • Causal inference and attribution analysis

Deliverables

Experiment designs and results
Dashboards and monitoring systems
Predictive models and ML services
Causal analysis reports
Metrics definitions and SLAs
Alerting rules and on-call runbooks
Success metric

Decision quality — how many product decisions are backed by experimental evidence?

Relationships

How Pulse connects to every other tribe — what flows, when, and through what ritual.

Feeds
EchoPer experiment; weekly summary

Empirical validation of Echo's hypotheses, behavioral data for user models, anomaly detection on user segments

Ritual:Metrics review
If this breaks: Echo's insights stay untested — signal remains unvalidated
Shadow risk: Pulse may use data as a verdict and make partners feel observed, ranked, or reduced to performance.
Repair: Name what the metric cannot see, let the affected tribe interpret first, and agree what evidence could change Pulse's view.
AtlasWeekly metrics review; quarterly deep dive

Experiment results, market metric trends, competitive benchmarks, causal analysis of strategic moves

Ritual:Metrics review
If this breaks: Atlas makes strategy calls based on gut, not data
Shadow risk: Pulse may use data as a verdict and make partners feel observed, ranked, or reduced to performance.
Repair: Name what the metric cannot see, let the affected tribe interpret first, and agree what evidence could change Pulse's view.
DeltaPer experiment completion

Experiment results, performance regression alerts, feature adoption metrics, rollout recommendations

Ritual:Experiment readout
If this breaks: Delta doesn't know if what they shipped actually worked
Shadow risk: Pulse may use data as a verdict and make partners feel observed, ranked, or reduced to performance.
Repair: Name what the metric cannot see, let the affected tribe interpret first, and agree what evidence could change Pulse's view.
OrbitWeekly

Customer health scores, usage anomaly alerts, adoption trend reports, churn prediction signals

Ritual:Customer health review
If this breaks: Orbit doesn't see customer problems until the customer calls
Shadow risk: Pulse may use data as a verdict and make partners feel observed, ranked, or reduced to performance.
Repair: Name what the metric cannot see, let the affected tribe interpret first, and agree what evidence could change Pulse's view.
Receives
The tribe under strain

To be unassailable because the evidence is on their side.

This is a protective pattern, not a verdict. It describes what can happen when the tribe's gift is driven by fear, anger, grief, scarcity, or status.

What activates it

A decision relies on intuition, emotion, incomplete data, or a metric Pulse does not trust.

How it can appear

Pulse can detach from human impact, keep score, delay action for proof, or use measurement to expose rather than learn.

What it costs

What cannot be measured loses legitimacy; partners manage the metric instead of sharing uncertainty.

The integrated gift

Evidence with humility: making uncertainty visible while keeping lived experience inside the model.

Core question

Did it improve anything?

Information pipeline
1
EchoUnderstand humans
2
AtlasUnderstand markets
3
DeltaBuild solutions
4
ForgeCreate leverage
5
PulseMeasure outcomes
6
OrbitEnsure adoption
7
ThreadPreserve knowledge
8
HorizonExplore the unknown
9
VaultRun the business
Operating modes
Explorer

Discovers new signals — what should we be measuring that we aren't?

Builder

Builds data pipelines, ML models, analytics infrastructure

Optimizer

Refines experiments for statistical power, reduces noise, improves model accuracy

Guardian

Ensures data integrity — prevents p-hacking, survivor bias, misleading metrics

Catalyst

Makes data accessible — dashboards, alerts, self-serve analytics for every tribe

Visionary

Sees where ML and data science can create step-function improvements, not incremental gains