A/B Testing
Compares treatment and control groups to estimate whether an intervention produced a measurable effect on a predefined metric.
- Used for
- Testing interventions, comparing variants, estimating effect size and supporting rollout decisions.
TECHNICAL INDEX
Methods
A practical index of statistical, quantitative and AI methods used across the portfolio. Each method links to case studies where it is applied.
This is not a theoretical glossary. Methods are described in the context of measurable questions, assumptions, uncertainty and decision criteria.
Experimentation
Methods for planning, running and interpreting controlled tests with explicit metrics and decision criteria.
Multivariate Testing
Evaluates multiple factors or variants simultaneously to understand individual and combined effects.
- Used for
- Testing combinations of changes, identifying interaction effects and avoiding isolated one-variable conclusions.
Related case studies
Tags
Experimental Design
Defines hypotheses, metrics, treatment/control logic, randomization, sample size and decision criteria before analysis begins.
- Used for
- Reducing bias, preventing ambiguous conclusions and making experiments auditable.
Related case studies
Tags
Power analysis
Estimates the sample size required to detect a minimum meaningful effect with a predefined level of statistical power.
- Used for
- Planning experiments before launch and avoiding underpowered tests.
Related case studies
Tags
Sample Size Calculation
Determines how many observations are needed to estimate effects with acceptable precision and reliability.
- Used for
- Experiment planning, survey planning and statistical validity checks.
Related case studies
Tags
Sequential Testing
Allows experiment monitoring over time while controlling the risk of false positives caused by repeated looks at the data.
- Used for
- Experiment monitoring, early stopping and controlled decision-making during live tests.
Related case studies
Tags
Bayesian Testing
Estimates uncertainty using probability distributions and updates beliefs as evidence accumulates.
- Used for
- Probabilistic experiment interpretation and decision-making under uncertainty.
Related case studies
Tags
Multiple Testing Correction
Adjusts statistical decision criteria when multiple hypotheses are tested at the same time.
- Used for
- Reducing false discoveries in multivariate tests and multi-metric experiments.
Related case studies
Tags
Causal Inference
Methods for estimating whether changes can be attributed to interventions rather than bias or noise.
Causal Inference
Estimates whether an observed change can be attributed to an intervention rather than bias, trend, seasonality or random variation.
- Used for
- Impact measurement, intervention evaluation and decision support when causality matters.
Uplift Analysis
Measures the incremental effect of an intervention compared with what would have happened without it.
- Used for
- Estimating incremental impact, treatment effectiveness and rollout value.
Related case studies
Tags
Counterfactual Forecasting
Estimates what would likely have happened without an intervention, creating a comparison baseline for impact measurement.
- Used for
- Evaluating interventions when randomized experiments are not available.
Related case studies
Tags
Statistical Modeling
Methods for temporal, risk and uncertainty analysis in structured quantitative problems.
Time Series Impact Analysis
Analyzes temporal data to separate intervention effects from trends, seasonality and noise.
- Used for
- Measuring impact over time and evaluating changes in business or operational metrics.
Related case studies
Tags
Forecasting
Estimates future values of a metric using historical patterns, uncertainty and model assumptions.
- Used for
- Planning, scenario evaluation and performance monitoring.
Related case studies
Tags
Survival Analysis
Models time-to-event outcomes such as time until churn, failure, default, renewal or process abandonment.
- Used for
- Estimating risk over time and comparing survival patterns across groups.
Related case studies
Tags
Kaplan-Meier
Estimates survival probabilities over time without assuming a specific parametric distribution.
- Used for
- Visualizing time-to-event behavior and comparing groups.
Related case studies
Tags
Cox Proportional Hazards
Models the relationship between covariates and event risk over time.
- Used for
- Estimating hazard ratios and identifying factors associated with higher or lower risk.
Related case studies
Tags
Optimization
Methods for comparing scenarios, preferences, prices and portfolio decisions before action.
Price Elasticity
Estimates how demand changes in response to price variation.
- Used for
- Pricing decisions, revenue analysis and scenario simulation.
Related case studies
Tags
MaxDiff
Measures relative preference or importance by asking respondents to choose the most and least important options.
- Used for
- Prioritizing attributes, features, offers or portfolio elements.
Related case studies
Tags
Conjoint Analysis
Estimates how different attributes and levels contribute to preference or choice.
- Used for
- Portfolio design, pricing scenarios and offer optimization.
Related case studies
Tags
TURF Analysis
Identifies the combination of options that maximizes total reach across a target population.
- Used for
- Portfolio optimization and offer mix selection.
Related case studies
Tags
Market Share Simulation
Simulates expected share under different offer, price or attribute scenarios.
- Used for
- Comparing scenarios before making commercial or portfolio decisions.
Related case studies
Tags
Quantitative Intelligence
Methods for segmentation, prioritization, value modeling and measurable action planning.
Clustering
Groups observations based on similarity across quantitative variables.
- Used for
- Finding statistically meaningful segments and patterns in structured data.
Related case studies
Tags
RFM Segmentation
Segments entities based on recency, frequency and monetary value.
- Used for
- Prioritization, value analysis and retention strategy.
Related case studies
Tags
Churn Prediction
Estimates the probability that an entity will stop using, buying, renewing or remaining active.
- Used for
- Risk prioritization, retention planning and early-warning systems.
Related case studies
Tags
LTV Analysis
Estimates the expected value generated over the lifetime of a relationship.
- Used for
- Resource allocation, ROI analysis and long-term value prioritization.
Related case studies
Tags
ROI Optimization
Compares expected return against cost to prioritize actions with the highest measurable value.
- Used for
- Budget allocation, scenario comparison and performance optimization.
Related case studies
Tags
NLP & AI Automation
Methods for turning unstructured text and knowledge bases into traceable analytical outputs.
Sentiment Analysis
Classifies text according to sentiment polarity or intensity.
- Used for
- Turning large volumes of text into quantitative indicators.
Related case studies
Tags
Topic Modeling
Identifies recurring themes or latent topics in collections of text.
- Used for
- Summarizing unstructured data and tracking patterns across documents.
Related case studies
Tags
Entity Extraction
Extracts named entities such as organizations, locations, products, dates or technical terms from text.
- Used for
- Structuring unstructured documents and supporting search, classification or reporting.
Related case studies
Tags
Summarization
Condenses long text into shorter summaries while preserving the main information.
- Used for
- Reducing review time and creating executive summaries from large text collections.
Related case studies
Tags
RAG
Combines retrieval from a knowledge base with language model generation to answer questions using source-grounded context.
- Used for
- Manual search, internal knowledge assistants and documentation-based chatbots.
Related case studies
Tags
Vector Search
Finds semantically similar documents or passages using embeddings.
- Used for
- Knowledge retrieval, document search and RAG systems.