Statsig
SaaS unicorn for enterprises
Statsig is an experimentation and product analytics platform that empowers leading companies like OpenAI. It enables teams to run experiment at scale, turning raw data into actionable insights.
Timeline
4 month
May - Aug 2025
Team
2 Engineers
1 Product Manager
2 Data Scientists
Focus
SaaS
Data Visualization
Role
Product Design
Design System
Launched
8.5.2025
I designed conversion driver 0 -> 1; An analytic tool that pinpoints why users convert or drop off, by automatically surfacing the key factors behind those outcomes
Outcome 1
Proactive insights surfaced in Conversion Driver
Expandable cards let users quickly scan key metrics and dive deeper into detailed analyses when needed.
Outcome 2
Streamlined access to deep-dive data
Filtering by analysis groups and one-click shortcuts make it easier to surface targeted insights and dive into granular data with less friction.
SOLUTION IMPACT
30% faster
in decision making, proactive insights enabled teams to spot opportunities and risks more quickly.
22.3% increase
in engagement with analytics tools, turning a previously underutilized feature into a core part of daily workflows
8 weeks
from concept to launch, doubling the usual speed-to-market and enabling teams to act on insights months earlier
HOUSTONโฆWE HAVE A PROBLEMโฆ
Fragmented data, slow decisions
Enterprise User, June 6
We can see people dropping off before finishing sign-ups or purchasesโฆ but we wanna know why
Letโs add Conversion Driver โ maybe build it like our competitors so we can close the gap ASAP?
Product Manager, 12:39PM
Wait.. but what do users actually want to get out of this โ is it more numbers and data, or clear next steps?
Elaine's mind, Product Designer, 2:00 PM
LOOK INTO COMPETITORS
The true value is turning data into actionable insights and making exploration faster and easier.
DESIGN PROCESS - HOW TO SURFACE DATA
IDEA 1
Single Table With Filtering
low dev-cost
Using an existing table component save cost and time on dev end
space-saving
A compact tabular layout kept multiple variables visible in one place
insight gap
hard to quickly identify the most meaningful drivers
IDEA 2
Two Table View + Summary
reduced cognitive loads
surfacing summaries instead of raw data, help lowering mental effort
clearer decision making
Splitting insights into two categories (โConvertedโ vs. โDrop-offโ) made it easier to see patterns at a glance
transparency gap
lack of break-down or deep-dive options make users question where summaries come from
IDEA 3
Insight Ladder
reduced cognitive loads
Strikes a middle ground between high-level insights and the option to deep-dive
space efficiency
Maintains a compact layout while still layering in more depth when needed
more dev demand
A new, unbuilt design pattern that needs upfront investment โ but can scale across other analytics surfaces long term
Dropdown - V1
"how to save up more vertical space"
Restructured the confusion matrix to save vertical space, but learned data scientists preferred the conventional layout for easier interpretation
Dropdown - V2
"how can we help user read and spot important metric"
Used color coding and added graphs to help non-technical users quickly grasp key information and ease interpretation
Dropdown - V3
dynamic drop-downs
actionable summaries
correlation indicators
balanced layout
Put analytical tool on same surface to
reduce context-switching
AH-OH PLOT TWIST!
๐ตโ๐ซ
๐ค
Dev Concern + Edge Case Scenario
Raj
Data Scientist
11:28AM
If the funnel grows longer, the chart drops out of view โ then Iโm still stuck scrolling up and down to piece things together.
Pierre
Eng Lead
11:20AM
Not a fan โ putting the chart on the same surface would complicate things on the engineering side, since it would overlap with other datasets and be difficult to pull off
Control tabs
Pierre
Eng Lead
3:20PM
I prefer this design, good discoverability + easy to implement
But it clutters the top rail and blurs the distinction between configuration tabs and analysis features, would confuse our users
๐๐
The expectation has already been set with the interactions we introduced earlier
Plus, testing showed users already have the instinct to click on the chart to explore, so discoverability shouldnโt be a problem
Pierre
Eng Lead
4:50PM
๐๐ผ๐
Contextual menu + Modal
DESIGN FOR THE FUTURE
Weโre not just solving for today โ the design needs to scale as we add more functionality down the road!
Product Manager, 10:39AM
Makes sense โ I will see where this pattern could be integrated to make it even more scalable and consistent across the product
Elaine, Product Designer, 12:00 PM
Create components
Design guidelines
Scale the design across data-analytics surfaces for a cohesive experience
Brief look back
My Key Takeaways
Thriving in ambiguity: Learned to break down vague, high-level asks into actionable steps, using quick alignment loops with PMs and engineers to clarify priorities
Designing at speed: Built and iterated fast, balancing scrappy solutions with thoughtful design to ship in 8 weeks from concept to launch without compromising usability
Impact focused thinking: Learned to prioritize user outcomes and business impact over design perfection, focusing on what truly moves the needle
