Outlier Detection in Data With Amazon Lookout for Metrics

Facilitate data processing and build highly accurate machine learning models to detect outliers and anomalies in data faster, more efficiently, and at scale

Identify and resolve issues
affecting your business
with AI & AWS

Organizations in data-rich, highly dynamic industries such as e-commerce, gaming, adtech, and telco are moving away from manual tracking and rules-based algorithms for monitoring revenue, customer engagement and operational metrics. They seek to empower their analytics & anomaly detection initiatives, and avoid service interruptions caused by inefficient data processing and alert fatigue. With Amazon Lookout for Metrics, organizations can better assure the stability of applications and networks, and ensure a seamless customer experience by detecting and diagnosing anomalies (i.e. outliers from the norm) in business and operational time series data.

What Is Amazon Lookout for Metrics?

Amazon Lookout for Metrics is a fully managed machine learning service that makes it easy for businesses to more efficiently detect outliers in their data making sure that all important metrics are monitored in near real-time and under control. With the service, every developer can build highly accurate models for anomaly detection, without any machine learning expertise, to improve accuracy, accelerate time-to-detection, and deliver actionable insights. This allows businesses to react in a rapidly changing environment ensuring that all services, including customer-facing applications, work properly.

Amazon Lookout for Metrics Is Used In

& Finance


Finding outliers and anomalies,
and comparing such metrics as
revenue, cost, profit, and pricing

IT & Tech


Finding anomalies and
analyzing CPU performance,
latency, memory consumption,
and availability

Marketing, Sales,
and Advertising


Finding outliers and anomalies
impacting product usage,
web page views, churn rate,
and CTR

Supply Chain Ops
& Logistics


Ensuring operations reliability
by detecting anomalies in
inventory levels, returns,
delivery time, and more

Key Features of
Amazon Lookout for Metrics

Machine learning-powered, automated detection of outliers and anomalies in real-time data
Ability to use historical data to train and tune models for real-time anomaly detection
Simplified on-the-go ML model training when customers provide little to no data
Outlier detection in time-series data to determine their root causes across multiple metrics and domains
Grouping and ranking of results, to monitor chosen metrics and to streamline reporting of outliers
Automated scaling to meet customer needs with simple API, with in-console testing and deployment options
Ability to continuously improve accuracy via a closed feedback loop

How It Works

With real-time data ingested to Amazon Lookout for Metrics, the service automatically selects the ML algorithm, called a detector, that best fits your data and use case, and starts detecting outliers based on chosen KPIs. If an outlier is detected, it is ranked by severity, and notification is sent. You can provide feedback on outliers to keep improving the model.


Provectus — An Amazon Lookout for Metrics Launch Partner

Provectus is proud to be an Amazon Lookout for Metrics launch partner. We are committed to helping our customers leverage Amazon Lookout for Metrics to make the most of their live and real-time data, to improve customer experience, sales analytics, and marketing intelligence.

Machine Learning Competency Partner

Provectus is an AWS Premier Consulting Partner with a strong track record of data analytics and anomaly detection solutions. AWS Data & Analytics, DevOps, and Machine Learning Competency Partner.