Corrily Experimentation Platform
Why use Corrily for your pricing, packaging and discount experiments?
Corrily provides an end to end experimentation platform for all things pricing, packaging, discounts and promotions. Corrily’s experimentation platform abstracts the complexity involved in running experiments in-house, and manages everything for you. Moreover, our machine learning models are pre-trained, and continuously improving over time based on your business-specific data and anonymous data across our entire customer base. This, in combination with our data-efficient sampling algorithms, results in 30-40% improvement in experiment convergence over traditional AB/n experiments.
What can you manage using Corrily’s experimentation platform?
Experiments in Corrily can be set-up and managed using our dashboard or the API:
- Defining the experiment objective
- Managing control and experiment treatments
- Orchestrating experiments i.e. traffic allocation, adding and disabling particular experiment treatments
- Measuring experiment results across most common metrics such as ARPU, LTV, Conversion, Retention, MRR, ARR etc.
- Calculating statistical significance and Confidence Intervals
- Post-experiment clean-up
- Rolling out winning experiment treatments
Different types of Experiments in Corrily
Corrily’s Experimentation platform allows you to run various kinds of experiments.
- Pricing Experiments: to find the best price for a set of users. Pricing experiments can be used to test different price points, different strikethrough prices, as well as price formatting i.e. price rounding strategies and currencies shown
- Packaging Experiment: to find the best combination of products and plans for a set of users. Packaging experiments can also be used to test things such as copy variations for a product, different trial length etc.
- Discount Experiments: to find the optimal discount for a set of users. Discount experiments can be used to test optimal discount percentages or amounts or coupons
Things to know before setting up an experiment
A few things to note while running an experiment with Corrily:
- Experiment Objective: Corrily supports various experiment objectives that you can choose from, such as Optimizing ARPU, LTV, Conversion or Retention. Make sure you are selecting right experiment objective
- Experiment Targeting: Experiments can be run on a subset of products, countries and audiences. We recommend spending some time thinking through the experiment architecture, especially if you plan on running multiple experiments. At Corrily, our Pricing and Data Science experts are more than happy to consult.
- Experiment States: An experiment can have multiple states:
Completed. You can set-up an experiment in the
Draftstate, submit it for
Reviewbefore going live. Once an experiment is live, you can decide to
Pausean experiment (say during critical business periods), and
Resumeit later. Experiments can move to a
Completedstate when they reach desired statistical significance, or can be forced to end subjectively
- Traffic Allocation: Experiment traffic allocation can be manual or dynamic i.e. optimized automatically by Corrily’s machine learning models
- An experiment in Corrily does not necessarily need to have a control group