Building a $30M
Experimentation Program
Enterprise experimentation from scratch across 5 regulated energy retail brands — Reliant Energy, Direct Energy, Green Mountain Energy, Cirro Energy, and Discount Power — serving 7 million customers in 24 states.
Five Brands. Zero
Shared Methodology.
NRG Energy operates 5 national energy retail brands across 24 states, each with different customer bases, regulatory environments, and acquisition funnels. When I joined, experimentation was fragmented — under 20 tests per year with no connection to revenue outcomes.
The C-suite couldn't answer a basic question: what is the ROI of our experimentation program? Without that answer, the program would never get the resources or executive sponsorship it needed to scale.
From Foundation
to Compound Growth
Building the Program from Zero
Inherited fragmented testing across 5 brands with no shared methodology. Audited every existing test, identified the measurement gaps, and designed the hypothesis engine that would connect behavioral insights to revenue forecasts. Established statistical standards: minimum detectable effect, power calculations, and sequential testing protocols.
From 20 to 100+ Tests Per Year
Rolled out the PRISM method across all 5 regulated energy brands — Reliant Energy, Direct Energy, Green Mountain Energy, Cirro Energy, and Discount Power. Built the revenue-per-experiment metric that made the program legible to the C-suite. Trained cross-functional teams on behavioral science–driven hypothesis writing.
AI-Augmented Experimentation
Introduced AI-assisted workflows — automated hypothesis generation from behavioral heuristics, analysis acceleration, and CDP-driven personalization. Cut analysis time by 40% while maintaining statistical rigor. Established AI governance frameworks for experimentation at enterprise scale.
What Hundreds of Experiments
Actually Tested
Test-level details stay internal — but the themes show how the program thinks. Every hypothesis names a behavioral mechanism before it runs, and every readout holds to the same statistical standard: Bayesian and sequential analysis, pre-set minimum detectable effects and power requirements, guardrail metrics, and false-discovery-rate control.
Enrollment-Flow Simplification
Removed unnecessary steps, form fields, and distractions from online enrollment. Double-digit lift in enrollment confirms, validated at >99% Bayesian probability.
Choice Architecture on Plan-Comparison Charts
Clearer visual hierarchy, plan labels, and benefit callouts on product charts. Single-digit lifts in enrollment starts with ~35% improvement in chart-to-enrollment progression.
Call-Channel Capture & Mobile Navigation
Surfaced a prominent acquisition phone number and mobile sticky navigation for high-intent visitors. Roughly 3× lift in incremental call sales — a digital test that recovered an offline channel.
Friction & Trust in Credit and Deposit Steps
Reframed identity-verification copy around approval benefits. Mid-single-digit lift in the lower-friction path, reducing deposit friction across the funnel.
De-Emphasizing Low-Commitment Options
Demoted a low-commitment payment option from a prominent button to a text link. Measurable reduction in deposit holds — a cost-savings win, not a conversion win.
Personalization via CDP Segmentation
AI-assisted dynamic content and real-time behavioral segmentation via Tealium CDP and Optimizely. Double-digit uplift in personalized lead acquisition.
Every test moves through the same 12-step operating pipeline — from idea intake to executive readout and handoff. See the pipeline →
The Principles
Behind the Numbers
Every experiment at NRG starts with a named behavioral mechanism. Here are four that drove the largest revenue impact.
Loss Aversion
Reframed rate plan choices from potential gains to potential losses. 'You're currently overpaying $X/mo' outperformed 'Save $X/mo' across all 5 brands.
Choice Architecture
Reduced plan options from 8+ to 3 recommended tiers with smart defaults. Conversion rate increased by double digits by eliminating decision paralysis.
Social Proof
Added local social-proof cues ('most customers in your area chose this plan') to rate comparison pages. Trust signals reduced time-to-decision by 22%.
Anchoring
Positioned the highest-value plan first in the comparison flow. Average revenue per customer increased without changing the underlying rate structures.
Built on the
PRISM Method
The NRG experimentation program runs on PRISM — a behavioral science–driven process that predicts revenue before a test runs and verifies it after. Every experiment follows the same five-step cycle.
See the PRISM Method →See all case studies
Explore the geo-experimentation work at Silicon Valley Bank, see individual test-level evidence in the Experiments hub, or read weekly breakdowns on Lean Experiments.
Revenue Frameworks
for Growth Leaders
Every week: one experiment, one framework, one insight to make your marketing more evidence-based and your revenue more predictable.
Free · No spam · Unsubscribe anytime
Read the archive
200+ issues of experiments, frameworks, and field reports from inside a Fortune 150 growth team.
Open Substack (opens in new tab)