EXPERIMENTATION
Test Your Best Ideas in Production
Empower anyone to gather data on how a new feature is performing. Test your hypotheses on real users, measure the impact and gain confidence you're making the right changes.
Why Experimentation?
Deliver new features with confidence by rapidly learning from your users and systems before a full rollout. Vet new functionality by testing it against a baseline option and measure its effects on system performance and cost, in addition to your business metrics. Experimentation will give your team the confidence and data to launch the right features no matter if it’s a simple front-end user experience upgrade or something more complex like an enhanced back-end algorithm or microservice.
Fail Fast, Iterate Rapidly
Test new ideas quickly, eliminate the bad ones and deliver value to your users faster.
Predict Performance Impacts
Gauge how new features will affect your system’s reliability once they are fully rolled out in production, without guessing.
Empower Any Team to Experiment
With feature flags at the core of every new change, anyone can set up, monitor and visualize experiments without additional developer overhead.
Declare a Winner & Take Action
Start serving the winning variation immediately by simply updating the flag targeting rule to 100%. No need to remove the experiment or redeploy in order to take action.
Multivariate Flags and Experiments
Test and evaluate multiple variants in each experiment.
Multivariate Flags
Custom Metrics
Targeting Rules
Experiment Lifecycle
Experiment Results
Real-Time Updates
Multivariate Flags and Experiments
Test and evaluate multiple variants in each experiment.
Custom Metrics
Define your own measurements and set baselines for comparison.
Powerful Targeting Rules
Define specific users or groups to include in an experiment.
Control the Experiment Lifecycle
Pause, resume, and reset experiments as desired.
Experiment Results
Visualizations to help you evaluate trends across different variations.
Real-Time Updates
Take action immediately after declaring a winner.
The First Experimentation Tool for the Entire Team
Not every new feature requires a user experiment. But when you need to learn from your users, do it rapidly with Experimentation from LaunchDarkly. Finally, a platform that makes running experiments a frictionless experience for every team, from Development and Operations teams to Product & Marketing teams.
Development
Run an experiment to measure application response time and see if it increases with a new feature.
Operations
Run an experiment to track an increase in infrastructure costs after releasing new functionality.
Product & Marketing
Run an experiment to test different feature variations or copy changes, to see if users engage more and convert.
Experimenting is Easy with Feature Flags at the Core
1. Create a Flag
Every new feature should already be wrapped in flag. Looks like you’re ready to go!
2. Assign Metrics
Define the data you want to collect: revenue impacts, page load time, server cost, etc.
3. Start Recording
Fine tune your experiment audience and start recording data for each assigned metric.
Robust Experiments Powered by Feature Flags
Flag + Metrics = Experiment
Every new feature benefits from a flag by giving you total control. By associating metrics with a feature flag you’re able to measure incredibly robust things like revenue impacts, page load time and server cost.
Truly Full-Stack Capabilities
Run experiments on anything in your stack from a simple front-end change to more in-depth back-end features like how an infrastructure change can impact page load time.
Seamless Development Workflows
Easily incorporate testing and experimentation capabilities into your existing software delivery workflows creating a frictionless experience for development teams that want to gather more data.
Conversion & Numeric Baselines
Set a baseline or ‘control’ variation and measure how far your new feature deviates from a baseline value for things like server cost and revenue.
Control the Entire Lifecycle
Granular User Targeting - Use powerful segmentation rules to control exactly which groups of users you want to include or exclude from an experiment.
Start, Stop & Reset - Take full control over your experiments by being able to pause, resume, reset or change the baseline for any metric associated with a flag at any time during any experiment. Your team will have granular control over exactly what data is being recorded and for how long.
Visualize & Take Action - View how your experiments are trending in real-time to easily declare a winning variation and take action in 200ms by simply updating the flag targeting rules.
