CASE STUDIES · LIVE TOOLS · SHARED EXPLANATORY APPROACH

Case studies that show the
approach in action.

A concept test and NBA shot logs — built with the same shared architecture across commercial research and operational data. Two domains, one workflow — proof that the pattern travels.

Browse the Case Studies
2

Live Working Examples

2

Domains Already Demonstrated

1

Workflow That Travels

Browse the live studies

Start with the domain that interests you most, or compare the two to see how the same underlying approach travels.

Commercial Research

Concept Test

In a 2025 craft-beer concept test, 54.7% of beer-buying consumers said they would purchase the brand. The case study asks what package perceptions drove that intent and how it shifts across income, region, and brand awareness.

Dataset

Beer concept-test consumer survey

Main outcome

Top-2-box purchase intent

Sample size

803 respondents

What you'll see

The full Mirror workflow on a commercial concept test, brand-team-ready

  • Eight outcome models on the concept test (purchase intent at top-2-box, top-1, and ordered cuts; package appeal; brand fit; high-quality; premium; and most-purchased brand)
  • Built in partnership with emi Research Solutions on anonymized survey data
  • Brand-team-ready: results in points and percentages, not log-odds
View Concept Test
Or jump straight to a tool:
Survey Explorer Model Builder Inside the Models All-or-Nothing Simulator Model Checks Fine-Tuning Simulator Non-Response Test
Sports Analytics

NBA 3-Point Shots

In the 2014–15 NBA season, players made about 35% of their three-point attempts. The case study asks what shot conditions drove that rate and what might have changed it.

Dataset

SportVU shot logs

Main outcome

Three-point make rate

Sample size

~33,000 attempts

What you'll see

The same workflow on a different data shape — event-level, no sampling

  • One outcome model (three-point make rate)
  • Season and data-provenance framing for the SportVU sample
  • Multiple ways to simulate shot-condition changes
  • Direct comparison of exploration, tuning, and model-building paths
View NBA Case Study
Or jump straight to a tool:
Shot-Log Explorer Model Builder Inside the Models All-or-Nothing Simulator Model Checks Fine-Tuning Simulator

What the studies have in common

Each case study follows the same broad sequence from outcome to explanation to scenario testing.

1

Start with the outcome

A visible topline anchors the study and gives users a concrete entry point.

2

Inspect the source data

Users can move from descriptive distributions to more focused analytical questions.

3

Model the drivers

The explanatory layer makes it easier to see what measured factors matter under stated rules.

4

Test scenarios

Interactive tools let users see how outcomes change when conditions are altered.

Working with Electric Insights

Want a case study like these for your question?

The two examples are built on public data. The Services page describes the three ways the same approach can be applied to your data, your question, and your decision.

Discuss Your Project

Start with the case that interests you most

Both studies are live. Browse the background first, or jump straight into a tool.

Explore concept test Explore NBA