Services

At Portendo, we design and implement large scale & high performance machine learning platforms, helping our clients building their own capabilities to run and improve them. Our at-risk pricing scheme makes our value proposition unique.

What is Machine Learning

Machine learning learns from experience – in form of data - to detect and exploit behavior patterns automatically

Input
(in form of data)
Inference
(algorithms)
Action plan
(from predictions)

Which customers should I cold-call to cross-sell products?

Which price should I charge for every product in my catalog?

Which credit applications are frauds?

Machine learning is most effective when:

  • Human expertise does not exist  >> optimizing a complex insurance website
  • Humans are unable to explain their expertise  >> speech recognition
  • Data volume, complexity and speed increase   >> trading, Black Friday catalog pricing
  • There are business benefits to personalizing the solution  >> stratified medicine

It is very different and complementary to:

  • Regular data analytics which tends to focus mainly on human driven analysis to support management decisions
  • Transactional and expert systems (e.g., payroll, pricing rule engines) which focus on following pre-defined procedures
  • Forecasting which is mostly concerned with large trends and not with personalized, fine-grained, behaviors
  • Traditional statistics which are mostly focused on answering precise questions with small data sets and well-specified models (e.g., does this drug performs well or not)

Banking

Increase profits by reducing your overhead expenses, avoiding fraudulent applications, anticipating defaults and optimizing your marketing efforts.

Uses of Machine Learning algorithms for Financial Institutions

What's Predicted

Direct Action

Which loan applicants will default on their loans?

Reject the loan application

Which transactions or applications for credit are fraudulent?

Reject the transactions/applications or screen the transactions/applications with human auditors

Which homeowners will refinance their mortgages with a competing bank?

Perform the most cost effective action plan

Which late-paying customers will better respond to collection efforts, especial collections deals or will not respond at all?

Perform the most cost effective action plan

Which customers will close their bank accounts? Which ones can be persuaded to stay? How?

Retention efforts targeted to at-risk customers

Detect errors & novelties in trade tickets for a trading middle-office before processing occurs

Zoom on unusual trades and resolve errors immediately before processing the ticket

Insurance Companies

Increase profits by reducing your loss ratio, avoiding fraudulent applications and optimizing your pricing scheme.

Uses of Machine Learning algorithms for Insurance Companies

What's Driven

Use

Which claims are fraudulent? (auto, accident, bodily injury, robbery, fire, etc.)

Reject the claims or screen them with human auditors

Which insurance applications are fraudulent?

Reject fraudulent applications

What is the estimated loss ratio of a specific customer?

Price the insurance product accordingly

Detect patterns in claim processing, audits and insurance applications

Automate applications and claims handling procedures, save time and reduce overhead

E-Commerce

Increase revenues by maximizing cross selling, email campaigns performance, traffic allocation and by optimizing marketing strategies.

Uses of Machine Learning Algorithms for Retail E-commerce

What's Driven

Use

What products will a customer most likely buy? What about during or after the purchase?

Deploy a real-time recommendation algorithm to maximize cross-selling revenue

What pricing method is the most effective to maximize profits?

Develop a dynamic pricing algorithm that adjusts product pricing in real time

Identify the different customer segments and sub-segments for a recent product

Refine product & marketing to better reach them

Detect patterns in claim processing, audits and insurance applications

Increase website sales faster

Target email recipients, timing and formatting for marketing email campaigns

Maximize performance of email campaigns

Target phone prospects for a product selling call desk

Maximize performance of call desk

Telecom

Increase profits by reducing churn, maximizing cross selling, optimizing email campaigns performance.

Uses of Machine Learning Algorithms for Telecom

What's Driven

Use

How to maximize marketing effectiveness for each type of consumer?

Deploy real time marketing initiative from consumer location, travel patterns, shopping behavior, socioeconomic and financial info, etc.

Which consumers will switch telecom provider during the next 90 days?

Deploy different marketing initiatives to retain the consumer and reduce churn (discounts, better deal, etc.)

Which customers are involved in communications fraud?

Intervene to stop fraud and stop delivering the service

What pricing method is the most effective to maximize profits?

Develop a dynamic pricing algorithm that adjusts product pricing in real time for each type of costumer

Why Us?

  • Our team combines more than 25 years of experience in machine learning, business strategy and technology & operations.
  • We go at risk: our pricing scheme is composed by a small fixed fee and a variable success fee. You only pay for the value you receive.
  • We help companies building their own capabilities to run and improve their machine learning algorithms.