Analysis module icon

Personalized newspaper experience for users

Programatically matching users with content, based on analysis of their behavior, driving subscriptions and revenue
Based on a persistent user identity that survives cache clearance and cookie deletion, the system matches users with content by analysing their behavior and predicting their response, suggesting content with a high percentage likelihood to encourage subscription or upgrade.

The Client

We created the system for a marketing company that caters to some of the biggest names in American publishing. Thanks to our reputation, we were approached to build a system that addresses some of the key challenges faced by publishers as they digitize their operations.

Business challenge

As publishers digitize they must find new ways of reaching out to readers, and focus on building a revenue model based on subscriptions in a media marketplace that has never been more crowded and competitive.
1
Revenue model
Publishers are shifting back to a model powered by their readers, rejecting the advertising-based model of the early digital years. This requires a new approach to readers and to acquiring and keeping subscriptions.
2
Converting subscribers
Converting casual readers to subscribers in the digital world is both more necessary and more difficult than signing up regular newspaper readers. Publishers must source the tech and skills to convert and retain a new generation of consumers.
3
Consumer expectations
Customer expectations have been conditioned by other digital media platforms such as Netflix. Readers now expect personalization as standard, and are no longer satisfied with a one-size-fits-all approach.
4
Content responsiveness
Publishers need to know which content is successful with which readers, in order to plan future content creation and distribution strategies. Without attribution and analysis, they cannot build the publication their most profitable readers want.

When they attempt to address these concerns, publishers typically encounter technical issues including:

Persistent user identification icon
Persistent user identification
Identifying anonymous, non-logged in users on the site across sessions. These are conversion targets, but while they remain anonymous there can be no data-driven conversion rate optimization (CRO) aimed at them.
Support for different clients and structures icon
Support for different clients and structures
When building tools for publishers, agencies encounter difficulties in creating a tool that supports multiple website structures and client backend logic flows.
Data acquisition granularity and volume  icon
Data acquisition granularity and volume
To be effective, the system must collect and analyse massive tranches of granular pageview data for large numbers of users, including anonymous users and subscribers.
Inbound sales icon
Inbound sales
Integration across the stack is a vital component of integration across departments; in particular, inbound sales needs a responsive, real-time window into the insights generated by the system.
The goal was to create a system that addressed these concerns, equipping publishers for digital success.

In particular, it was necessary to deliver:

  • Increased sales efficiency
  • Transparency of data and decision-making facilitation based on analysis
  • Automated suggestions to readers and campaign running
  • Calculation of likelihood to subscribe
  • Personalized content consumption patterns

Personalizing publishing CRO at scale with machine learning

Lightpoint created a ML-based tool that matches users with content based on analysis of their previous behavior, and creates proposals designed to appeal to individual users, programmatically and at scale.
CRO illustration
1. A reader visits the publisher’s site.
2. The backend matches that reader with a persistent identity inside the tool and an analysis of their previous onsite behavior.
3. Based on this, the system offers that reader content with a higher probability to encourage conversion.
4. The system also generates tailored calls to action and personalized proposals emphasizing the content most likely to generate conversions.
5. Post-subscription, the process continues, with relevant offers displayed to segmented subscribers to encourage retention and upgrade, and sales involved in upselling and cross-selling where appropriate.

Key features of the system

Single user identity icon
Single user identity
Even anonymous users have a single, persistent identity across sessions, which survives cache clearance and enables long-term data collection.
Rapid sales response icon
Rapid sales response
The system is built to help increase the speed of sales reps to warm customers, as well as putting customer information at their disposal to improve closure rate.
Marketing campaign automation icon
Marketing campaign automation
Offers and analysis are built into marketing campaign design, allowing automated launch and monitoring.
Report generation and dashboarding icon
Report generation and dashboarding
The system converts analysis into reports delivered via a dashboard, providing the insights necessary for data-driven strategy.

Lightpoint built the system with:

2 Front -end Developers, 1 Back-end Golang Developers, 1 Data Scientist / Python developer, 1 Manual QA

Tech stack:

Database:

Google bigquery, Mongo, Postgres

Back-end:

Go (REST WebAPI), Python, NodeJs

UI:

TypeScript

Other:

GCP, GA, Google Data studio

Results

The product has become one of the most popular in the US media market, and continues to develop and expand its functionality. Our customers became the largest players in the American media market with a combined market share of around 60%. Customers of all sizes, not just enterprise clients, saw sharp increases in revenue and subscriber numbers.

Ready to talk about a similar solution for your business?

Get in touch with us at:
Full name
Email
Phone
Description