Multi-Armed Bandit

The term "Multi-Armed Bandit" comes from the field of probability theory and statistics, and it's often used in the context of reinforcement learning and A/B testing.

The name is derived from a hypothetical scenario involving a gambler at a row of slot machines (also known as "one-armed bandits"), who must decide which machines to play, how many times to play each machine and in which order to play them, to maximize his total reward.

In the context of online experiments, a multi-armed bandit algorithm dynamically adjusts the traffic allocation towards different variations based on their performance. This is different from traditional A/B testing where traffic is split evenly and the allocation does not change during the test.

How it Works

A multi-armed bandit algorithm works by continuously adjusting the traffic towards the best-performing variations until it can confidently pick the best variation. The winning variation will then receive 100% of the traffic.

For example, if a given variant has a 60% probability of being the best, a multi-armed bandit algorithm like Autotune will provide it 60% of the traffic. At a high level, the multi-armed bandit algorithm works by adding more users to a treatment as soon as it recognizes that it is clearly better in maximizing the reward (the target metric).

Throughout the process, higher-performing treatments are allocated more traffic whereas underperforming treatments are allocated less traffic. When the winning treatment beats the second-best treatment by enough margin, the process terminates.

Example

Let's say you're running an online store and you want to test three different designs for your checkout page to see which one leads to the highest conversion rate.

Instead of splitting your traffic evenly between the three designs (as you would in a traditional A/B test), you could use a multi-armed bandit algorithm to dynamically adjust the traffic allocation based on the performance of each design.

If design A is performing well and has a 70% chance of being the best, the algorithm will allocate 70% of your traffic to design A. If design B has a 20% chance of being the best, it will receive 20% of the traffic, and design C, with a 10% chance of being the best, will receive the remaining 10% of the traffic.

This way, you're maximizing your conversions while the test is still running, instead of waiting until the end of the test to make changes based on the results.

When to Use Multi-Armed Bandit

Multi-armed bandit algorithms are particularly useful when you want to minimize the opportunity cost of showing sub-optimal variations to your users.

They're also useful when you want to run continuous experiments without a fixed end date, or when the performance of variations changes over time. However, they're not as useful when you want to measure the impact of a variation on multiple metrics, as they can only optimize for a single metric.

Loved by customers at every stage of growth

See what our users have to say about building with Statsig
OpenAI
"At OpenAI, we want to iterate as fast as possible. Statsig enables us to grow, scale, and learn efficiently. Integrating experimentation with product analytics and feature flagging has been crucial for quickly understanding and addressing our users' top priorities."
Dave Cummings
Engineering Manager, ChatGPT
SoundCloud
"We evaluated Optimizely, LaunchDarkly, Split, and Eppo, but ultimately selected Statsig due to its comprehensive end-to-end integration. We wanted a complete solution rather than a partial one, including everything from the stats engine to data ingestion."
Don Browning
SVP, Data & Platform Engineering
Recroom
"Statsig has been a game changer for how we combine product development and A/B testing. It's made it a breeze to implement experiments with complex targeting logic and feel confident that we're getting back trusted results. It's the first commercially available A/B testing tool that feels like it was built by people who really get product experimentation."
Joel Witten
Head of Data
"We knew upon seeing Statsig's user interface that it was something a lot of teams could use."
Laura Spencer
Chief of Staff
"The beauty is that Statsig allows us to both run experiments, but also track the impact of feature releases."
Evelina Achilli
Product Growth Manager
"Statsig is my most recommended product for PMs."
Erez Naveh
VP of Product
"Statsig helps us identify where we can have the most impact and quickly iterate on those areas."
John Lahr
Growth Product Manager
"The ability to easily slice test results by different dimensions has enabled Product Managers to self-serve and uncover valuable insights."
Preethi Ramani
Chief Product Officer
"We decreased our average time to decision made for A/B tests by 7 days compared to our in-house platform."
Berengere Pohr
Team Lead - Experimentation
"Statsig is a powerful tool for experimentation that helped us go from 0 to 1."
Brooks Taylor
Data Science Lead
"We've processed over a billion events in the past year and gained amazing insights about our users using Statsig's analytics."
Ahmed Muneeb
Co-founder & CTO
SoundCloud
"Leveraging experimentation with Statsig helped us reach profitability for the first time in our 16-year history."
Zachary Zaranka
Director of Product
"Statsig enabled us to test our ideas rather than rely on guesswork. This unlocked new learnings and wins for the team."
David Sepulveda
Head of Data
Brex
"Brex's mission is to help businesses move fast. Statsig is now helping our engineers move fast. It has been a game changer to automate the manual lift typical to running experiments and has helped product teams ship the right features to their users quickly."
Karandeep Anand
President
Ancestry
"We only had so many analysts. Statsig provided the necessary tools to remove the bottleneck. I know that we are able to impact our key business metrics in a positive way with Statsig. We are definitely heading in the right direction with Statsig."
Partha Sarathi
Director of Engineering
"Statsig has enabled us to quickly understand the impact of the features we ship."
Shannon Priem
Lead PM
Ancestry
"I know that we are able to impact our key business metrics in a positive way with Statsig. We are definitely heading in the right direction with Statsig."
Partha Sarathi
Director of Engineering
"Working with the Statsig team feels like we're working with a team within our own company."
Jeff To
Engineering Manager
"[Statsig] enables shipping software 10x faster, each feature can be in production from day 0 and no big bang releases are needed."
Matteo Hertel
Founder
"We use Statsig's analytics to bring rigor to the decision-making process across every team at Wizehire."
Nick Carneiro
CTO
Notion
"We've successfully launched over 600 features behind Statsig feature flags, enabling us to ship at an impressive pace with confidence."
Wendy Jiao
Staff Software Engineer
"We chose Statsig because it offers a complete solution, from basic gradual rollouts to advanced experimentation techniques."
Carlos Augusto Zorrilla
Product Analytics Lead
"We have around 25 dashboards that have been built in Statsig, with about a third being built by non-technical stakeholders."
Alessio Maffeis
Engineering Manager
"Statsig beats any other tool in the market. Experimentation serves as the gateway to gaining a deeper understanding of our customers."
Toney Wen
Co-founder & CTO
"We finally had a tool we could rely on, and which enabled us to gather data intelligently."
Michael Koch
Engineering Manager
Notion
"At Notion, we're continuously learning what our users value and want every team to run experiments to learn more. It's also critical to maintain speed as a habit. Statsig's experimentation platform enables both this speed and learning for us."
Mengying Li
Data Science Manager
Whatnot
"Excited to bring Statsig to Whatnot! We finally found a product that moves just as fast as we do and have been super impressed with how closely our teams collaborate."
Rami Khalaf
Product Engineering Manager
"We realized that Statsig was investing in the right areas that will benefit us in the long-term."
Omar Guenena
Engineering Manager
"Having a dedicated Slack channel and support was really helpful for ramping up quickly."
Michael Sheldon
Head of Data
"Statsig takes away all the pre-work of doing experiments. It's really easy to setup, also it does all the analysis."
Elaine Tiburske
Data Scientist
"We thought we didn't have the resources for an A/B testing framework, but Statsig made it achievable for a small team."
Paul Frazee
CTO
Whatnot
"With Warehouse Native, we add things on the fly, so if you mess up something during set up, there aren't any consequences."
Jared Bauman
Engineering Manager - Core ML
"In my decades of experience working with vendors, Statsig is one of the best."
Laura Spencer
Technical Program Manager
"Statsig is a one-stop shop for product, engineering, and data teams to come together."
Duncan Wang
Manager - Data Analytics & Experimentation
Whatnot
"Engineers started to realize: I can measure the magnitude of change in user behavior that happened because of something I did!"
Todd Rudak
Director, Data Science & Product Analytics
"For every feature we launch, Statsig saves us about 3-5 days of extra work."
Rafael Blay
Data Scientist
"I appreciate how easy it is to set up experiments and have all our business metrics in one place."
Paulo Mann
Senior Product Manager
We use cookies to ensure you get the best experience on our website.
Privacy Policy