Lead Product Designer · 2016 — 2018
Salesforce Einstein Analytics
Einstein Analytics is Salesforce's analytics cloud, organized in three layers: prebuilt Analytics Apps to consume, App Studio and Data Explorer to produce, and a Data Manager to enable. As lead product designer I owned end-to-end design across App Studio — where teams build, style, and share dashboards and apps — and Data Explorer, the self-service workspace for visual and predictive analysis.
The work below follows the analyst's journey: consume prebuilt apps, build and customize dashboards in App Studio, then explore raw data, model it, and predict outcomes in Data Explorer.
Analytics Apps
Prebuilt, customizable dashboards for sales and service teams — analytics-ready on day one, from a sales home and rep leaderboard to a service agent performance view.
Understanding the analyst
Multiple design sprints with PMs, engineers, and customers mapped the analyst's journey — and an analytics methodology map (predictive vs. non-predictive, data-rich vs. data-poor) that became the backbone of the information architecture.
App Studio — App Home
The home for browsing, managing, and creating apps: details, usage analytics, and an activity timeline at a glance, plus a guided flow to spin up a new app from a template.
App Studio — Styling & conditional formatting
Edit and style any chart on a dashboard — change chart type and data fields, pick a color scheme, and apply conditional formatting rules that respond to the data with custom colors.
App Studio — Sharing & version history
Collaborate on an app: share with fine-grained access, browse a full version timeline, and compare or restore any previous version side by side.
Data Explorer — Connect & prepare
Bring data in from dozens of sources, preview and profile it, then prepare fields — converting measures to dimensions, binning, and renaming — before exploring.
Data Explorer — Explore & visualize
Pick a chart type or a predictive model, then explore visually — Einstein surfaces trend lines and statistical insights inline, and trellising breaks the data apart by dimension.
Data Explorer — Model & predict
Build a regression model in a few clicks, read a plain-language report on its quality and significance, then predict outcomes interactively by changing the inputs.