Design an experimentation dashboard for Sales Navigator

Problem Statement Description

Product context: LinkedIn is Microsoft's professional network; its products include profiles, feed, jobs, recruiting, LinkedIn Learning, sales tools, messaging, and ads.

LinkedIn’s Sales Navigator team runs experiments across B2B sales workflows such as onboarding, account and lead discovery, search, recommendations, alerts, outreach, CRM sync, and product education for users who are still learning how to get value from the tool. Internal PMs, data scientists, engineers, and go-to-market stakeholders need a dashboard that helps them understand whether an experiment is healthy, trustworthy, and decision-ready.

Today, experiment interpretation can be difficult because Sales Navigator impact is not captured by a single click or session metric. The product serves individual sellers, sales managers, and enterprise customers, so the dashboard must connect short-term product behavior with longer-term value, while accounting for cohorts, seat types, account context, network effects, and professional trust.

This is a metrics design exercise. Focus on what the experimentation dashboard should measure, how metrics should be defined and instrumented, what comparisons should be possible, and how the dashboard should help teams make launch, iteration, or rollback decisions without over-optimizing for misleading signals.

The dashboard design should consider:

- Primary success metrics for Sales Navigator experiments and how their denominators should be defined.

- Activation, adoption, engagement, retention, and monetization signals across individual users, teams, and enterprise accounts.

- Cohorts such as new versus experienced users, role type, subscription tier, company size, region, CRM-connected users, and learning/onboarding stage.

- Instrumentation requirements for exposure, eligibility, assignment, user actions, account-level outcomes, and delayed conversion events.

- Experiment health checks, including sample ratio mismatch, traffic allocation, event logging quality, latency, and statistical confidence.

- Guardrail metrics related to member trust, spam-like behavior, InMail quality, unsubscribe or complaint signals, degraded search relevance, and customer support issues.

- Decision usefulness for PMs and executives, including how the dashboard should support ship, iterate, hold, or stop decisions.

- Trade-offs between fast experimentation readouts and longer-cycle B2B sales outcomes that may take weeks or months to observe.

The goal is to frame a dashboard that enables LinkedIn teams to evaluate Sales Navigator experiments rigorously, compare impact across relevant user and account segments, and make confident product decisions that improve seller productivity while protecting customer trust and long-term business value.