We're Inventing Technology that Solves Chronic Customer Data Challenges

Generic Amperity Image of 3 connected data points on a grey background


Patents Stable Id

Stable ID

Patent No. 10,503,696 Maintaining cluster stability over time, so Amperity can reliably serve as your consistent customer ID.

Read the paper
Patents Generating Training Data

Generating Training Data

Patent No. 10,509,809 Optimized method for sampling and labeling records, with a variety of positive effects.

Read the paper
Patents Multiple Databases

Multiple Databases

Patent No. 10,599,395 Parameterizing probabilistic clustering to allow for multiple databases.

Read the paper
Patent Transitive Stitching

Transitive Stitching & Fusion

Patent No. 10,853,033 Two methods for combining and merging customer data sources, where n>2.

Patent Clustering

Clustering of Data Records with Hierarchical Cluster IDs

Patent No. 10,922,337 Clusterings at multiple confidence levels to enable downstream applications with varying precision/recall trade-off needs.

Patent Conflict Resolution

Cluster Conflict Resolution

U.S. Patent No.: 11,003,643 Our identity matching system is not just aiming to detect pairs of records associated with the same identity, but rather to produce coherent identity clusters.

Patent 11080043

Version Control of Configuration Tree

Patent No. 11,080,043 Enables Amperity to track, analyze and, if necessary, revert changes in data with a single button click.

Amperity Patent Badge for "Validating and Merging Configuration Tables"

Validating and Merging Configuration Tables

Patent No. 11,294,666 Allows for rapid cloning and experimentation without impacting the production tenant.

Pending Patents

Patent Pending

Proper Handling of Supersized Clusters

A graph theory-based algorithm partitions large clusters, enabling high quality identity resolution results even in the presence of low quality data.

Patent Pending

Predict Customer Lifetime Value with Unified Customer Data

A special modeling approach that leverages this richer source of customer data to improve predictive accuracy.

Patent Pending

Generative-Discriminative Ensemble Method for Predicting Lifetime Value

We use an ensemble model with both generative and discriminative models to produce robust customer lifetime value predictions for different types of data.

Patent Pending

Generating Affinity Groups with Multinomial Classification and Bayesian Ranking

Product affinity groups are constructed with the ranking of the customer’s likelihood of buying a given product. This ranking is derived from two other models using a Bayesian approach.

Patent Pending

Recommended Audience Size

We recommend the optimal campaign audience size by learning the right trade-offs between the marketing cost and revenue.

See how we put these patents to use

Learn about the platform

Web Pingpong Stitch

Recommended Content