More Powerful Experiments and Personalization at Scale with Amplitude and Split

The very best products are able to serve the optimal experience for each individual user at the right time through deep customer insights and precise, adaptable targeting. 

Split has partnered with Amplitude to let you do just that, by utilizing Amplitude’s behavioral cohorts and predictive cohorts  directly within Split.

Save Time and Avoid Errors

With this new integration, there is no need to manually create and manage target lists in several places, saving you time and removing the chance of errors. Once integrated, your Amplitude cohorts can be brought into Split as a segment via a one-time sync or a scheduled sync that updates at a specific time interval (hourly or daily). For example, you can test out a new support feature on users who recently used support (based on a cohort synced hourly) or frequently used support during a specific time range in the past (based on a cohort synced just once).

Personalize Customer Experiences

By sending Amplitude cohorts to Split, you can target customers with relevant flag treatments based on their past or predicted behavior. With behavioral cohorts, you can fine-tune your target audiences based on user attributes (such as age and location), users events (such as cart abandonment and past purchases), and timeframe. If you’re set on a specific goal (such as conversions, upgrades, or dollar spend), you can also choose to create a predictive cohort based on your desired outcome and Amplitude will automatically build a machine learning model to predict the users most likely to achieve that outcome.

When used as targeting rules for features flags in Split, these cohorts allow you to identify sample populations for beta testing and personalize customer experiences at scale. For example, if targeting with a predictive cohort most likely to upgrade their plan, you can drive these customers towards upgrading using feature flags to target them free trials and other in-app offers. You can use a behavioral cohort to identify power users for a specific feature and offer them beta access to experiment with a new version of that feature. The possibilities are endless!  

Boost Your Experimentation Program

Alongside the vast benefits to your customers and your business, integrating cohorts into your processes can also improve your experimentation program. With Amplitude’s behavioral analysis and predictive insights, you can gain new ideas to drive future experiments. By understanding the relation between user actions and your desired outcomes, you can generate new hypotheses for improving your customer journeys, and test them out by experimenting within Split.

Targeting specific audiences using cohorts can also speed up your testing velocity. You can increase the power of your experiments with feature flags and cohorts by targeting the customers most likely to interact with, and be impacted by, the new feature being tested. This enables you to get statistically significant results faster without needing to increase the number of users exposed, keeping costs down (in the case of free trials and offers, for example) and without accepting additional risk by exposing all of your users to your new features.

To start sending Amplitude cohorts to Split today, check out our documentation.

Learn More About Delivering Impactful Features with Experimentation

Join us at Split’s upcoming virtual user conference, Flagship 2021 from March 16-17. Can’t wait? Neither can we! In the meantime, we have plenty of related resources to keep you busy below:

And as always, we’d love to have you follow along as we share new stories. Check us out on YouTube, Twitter, and LinkedIn!