Personalization in the style of Netflix and Amazon—optimizing the digital experience for the right user with the right message at the right time— is the dream of every marketer.
But the effectiveness of personalization hinges on the data you have access to. By themselves, demographic or behavioral or algorithmic data aren’t enough. To achieve true 1:1 personalization, you need all three: only then can you deliver to your users tailored experiences that maximize conversion rates.
Amplitude Recommend is the first self-serve personalization platform to work for every job, stage, and stakeholder of the personalization workflow—all without needing a data scientist or engineer:
- Segmentation (right user): Marketing can identify the best segments for behavioral personalization via cohorts and computations
- Personalization (right message): Product can determine the next best action for 1:1 personalization via predictions and recommendations
- Delivery (right time): Engineers can export data into their digital channels for real-time personalization via APIs and syncs
NOTE: Recommend is an evolution and expansion of Amplitude's Engage add-on. For that reason, Engage is no longer available.
Also, Recommend is not currently available to customers based in the European Union.
Many companies approach personalization by simply showing content items in random arrays, or at best, they sort by frequency. This is not true personalization: these recommendations are not personalized in any meaningful way, and these companies are losing out on incremental revenue gains a more systematic approach would deliver.
By contrast, Amplitude Recommend transforms a product from static to dynamic experiences, driving a 15% to 30% lift in conversions in the process.
How Recommend delivers personalized experiences to your users
While Amplitude is built around user-specific data, Recommend is rooted instead in the concept of cohorts, groups of users who have something in common. When you apply predictions and recommendations to a cohort, you can give that cohort the deeply personalized experience that maximizes lift.
Recommend lets you segment your users via cohorts or computations.
Amplitude supports two types of cohorts: behavioral and predictive.
Behavioral cohorts are clusters of users grouped together based on their past actions. Create them by segmenting users by events they fired or did not fire, whether they were active or inactive, and the user properties they have.
Behavioral cohorts are perfect for engagement-based targeting when you want to reach users at specific stages of your user lifecycle: perhaps they added something to their cart but did not purchase in the last 24 hours, for example, or they’ve recently activated a subscription.
Predictive cohorts are clusters of users grouped together based on the actions Amplitude anticipates they will take sometime in the future. Create a predictive cohort by segmenting users by their percentile likelihood: for example, whether they are in the top 10% likelihood to activate, or bottom 25% likelihood to churn.
Use these cohorts for activation, retention, or engagement based marketing when you want to drive users to the next stage of a user lifecycle: this would include cohorts of users likely to subscribe next week, or users likely to watch a second show next week.
NOTE: Cohorts created in Amplitude can be sent to all the integrations listed here.
All customers on an Amplitude Growth or Enterprise plan have access to unlimited cohorts.
Computations work by transforming an event or event property into a computed user property. You can use the computed property as a configurable filter in any Amplitude chart for analysis, or as a personalization tool by syncing it to an external destination.
For example, segmenting your users by the number of purchases they’ve made is feasible with cohorts. However, you’d have to create a different cohort to correspond to each possible number of purchases. To streamline the process, you could use computations to create a new, computed property that stores this information as a single integer, for each user. Then segment your users based on their values for this property, taking one step to do what would have previously taken several.
All customers on an Amplitude Growth or Enterprise plan have access to 10 computed properties. You can upgrade your plan to include unlimited computed properties. Contact your Amplitude representative for more details.
In Recommend, personalization is accomplished through the use of predictions and recommendations.
When you specify a predictive goal, like a user’s likelihood to subscribe or watch a second show, Recommend leverages all your historical behavioral data to predict future performance based on past performance. Every user is assigned an individual probability to reach your specified goal in the next seven days, recalculated every hour.
Accuracy metrics and predicted-vs-actuals are always front-and-center. You can explore and build cohorts by percentile of user probability. And you can analyze which behaviors and user properties were most important to the prediction for distinguishing high vs low probability users.
You can create unlimited predictions, but only 30 can be active—in other words, refreshing hourly—at any given time. Use predictions to identify the users most likely to take an action, and trigger a personalized communication to them right before they do.
Predictions are only available to Recommend customers.
Once you’ve identified a predictive goal for your users, the next step is making the recommendations that are most likely to drive users to reach it. Once you’ve specified which event / event property combinations you’re interested in, Amplitude’s AutoML system will determine which items are most likely to maximize each user’s predictive goal, and then get those items in front of the user. The entire process takes minutes instead of weeks, with minimal to no code.
Recommendations are only available to Recommend customers.
Amplitude Recommend delivers data via manual or automated syncs, and via two APIs.
When you sync your cohorts to a destination, like Facebook or Braze, all the userIDs/emails/mobileIDs in your Amplitude cohort will be exported to that destination.
Amplitude supports two types of syncs: on-demand, and automated. On-demand syncs are ad-hoc, one-time syncs, useful for audience testing and one-off campaigns. Automated syncs are scheduled on a daily or hourly cadence—so as your cohort audience membership changes, or the underlying predicted probabilities of the user change, Amplitude will automatically adjust their cohort membership in connected destinations as well. No more CSV downloads or manual syncs required—whenever your users take an action in your app, they’ll be automatically synced to your respective ad, email, or testing platform.
All customers on an Amplitude Growth or Enterprise plan will have access to unlimited one-time syncs and five automated syncs. You can upgrade your plan to include unlimited automated syncs. Contact your Amplitude representative for more details.
In addition to directly exporting data via syncs to your marketing destinations, you can integrate Amplitude data directly into your internal systems or app via two APIs:
- Use the Cohorts API to list all your cohorts in Amplitude, export a cohort, or upload a cohort. Simply query for a specific cohort, and all the users currently in it will be returned.
- Use the Profile API to list data by user. Simply query by userID or deviceID, and you can return their user properties, cohorts, computed properties, predictions, or recommendation values in real time.
All customers on an Amplitude Growth or Enterprise plan have access to 500 Cohort API calls. You can upgrade your plan to include unlimited Cohort API calls and Profile API calls. Contact your Amplitude representative for more details.
Check out our entire suite of documentation for Amplitude Recommend:
- Computations: Create new user properties on the fly
- Predictions: Use Amplitude's AI to help maximize lift
- Build a prediction
- Use prediction-based cohorts in your campaigns
- Recommendations: Help users reach the goals you've set for them
- Build a recommendation
- Use recommendations in personalization campaigns
- Sync your recommendations and cohorts to third-party destinations
- The Profile API