Hyperlocal Engagement with Our Audience Service

Introduction

In the ever-evolving landscape of digital marketing, the pursuit of pinpoint accuracy in connecting with the ideal audience has sparked a major change: “hyperlocal targeting”. This innovative strategy is revolutionising advertising by enabling advertisers to focus their efforts on specific geographic areas, and more critically, on finely segmented user cohorts within those regions.

Benefits of hyperlocal targeting

• Maximising ROI : Rather than dispersing resources across broad demographics, this strategy focuses resources precisely where they’re most impactful.
• Empowering CPaaS Systems :these systems leverage the granular data obtained through hyperlocal targeting to offer advertisers the means to communicate more efficiently and effectively with their intended audience

What Are User Cohorts

User cohorts refer to segmented groups of individuals who exhibit similar characteristics, behaviours, or attributes. These similarities might revolve around demographics, interests, purchasing behaviors, or even geographic location. When harnessed effectively, these cohorts serve as the foundation for targeted marketing endeavours.

• Demographic Cohorts: These groups are categorised based on age, gender etc or other demographic markers.
• Psychographic Cohorts: These cohorts are characterised by similar lifestyle choices, values, attitudes, and interests. For instance, foodies, fitness enthusiasts etc
• Geographic Cohorts: Segments based on geographic location, enabling hyperlocal targeting.

Problem Statement

Facilitating Hyperlocal Targeting and Real-time Audience Customisation

Airtel’s forward-thinking vision encompasses a drive toward hyperlocal targeting and real-time campaign personalization, aligning seamlessly with the core objectives of our project. At its core, our project aims to revolutionise the advertising landscape by building an audience service platform that caters to these specific objectives.

Project Requirements and Objectives

• Precision and Efficiency: The platform’s core objective is to create an audience service that offers precision of targeting at a hyperlocal level.
• Real-time Insights: Providing advertisers with real-time cardinality ensures they have immediate access to audience insights, enabling quick and informed decisions in campaign creation.
• Seamless Integration: Easy Integration with CPaaS systems ensures a seamless execution of campaigns, facilitating direct and personalised communication to targeted audiences.

Challenges in Handling High Volume Data and Network-Based Location Mapping

• Data Processing Bottlenecks:

Dealing with vast amounts of user data poses challenges in processing and analysis specially when required in real time.Scaling data processing pipelines to handle this volume efficiently without compromising speed or accuracy is a primary concern.

• Periodic Data Refresh Cycles:

Periodic refreshment of such huge data without introducing any downtime to the application was a big challenge.Implementing pipelines to automate the periodic refreshment of data ensures that cohorts stay up-to-date with the latest user behaviours and trends, maintaining the relevancy and accuracy of targeting strategies.

• Algorithmic Efficiency:

Implementing data segmentation algorithms that effectively categorize users into precise cohorts demands sophisticated methodologies. Within this process, robust data governance practices will be imperative to ensure the reliability, quality, and ethical use of data. Equally important is the need to ensure that these algorithms remain efficient and scalable, especially when handling large datasets

• Precise User Location Mapping via Network Towers

Network tower data might face signal interferences or inaccuracies, leading to erroneous location mapping. Filtering out noise and ensuring data accuracy is crucial for reliable location-based targeting.

• Security & Privacy

One of our paramount principles was data anonymisation. Keeping privacy in mind no data is circulated outside our ecosystem. User cohorts, profile all are not shared with any other application.

Solution Implementation

• Utilising Data Science Models and Analytics Techniques

Leveraging diverse data science models and analytics techniques to categorize users into cohorts based on behavioural patterns, demographics, and location preferences ensures a granular segmentation process.

• Hive as a Large Data Store

Storing these segmented cohorts in Hive provides a scalable and structured data repository, enabling efficient data management and accessibility for future analytics and campaign targeting.

• Automated Data Refreshment:

To tackle this challenge, we devised an efficient approach involving table updates and data partitioning in Hive. Initially, we created Parquet files that contained the refreshed data. Using data partitioning and sharding techniques, we seamlessly swapped these updated partitions within our data structure. This method ensured a smooth and timely update process while maintaining the integrity and consistency of our data

Parquet is a columnar storage format that’s designed to be highly efficient for analytics workloads. While Parquet files themselves are immutable

Real-time Cardinality with Trino: Why we used Trino

Distributed SQL Query Engine : Queries are divided into smaller tasks and executed in parallel across multiple worker nodes, enhancing overall query speed and performance.

Architecture: Trino follows a distributed architecture, employing a coordinator-worker model. The coordinator manages queries, while multiple worker nodes execute query fragments in parallel across the data sources.

Enhanced Location Mapping: Collaborating with networks to obtain precise user location details enriches the cohort data, enabling more accurate hyperlocal targeting and personalised campaigns.

Architecture

Conclusion

Redefining Advertising Precision with Hyperlocal Targeting

The implications of our approach are profound, with advertisers witnessing heightened ROI, more engaging campaigns, and the delivery of contextually relevant ads. As we continue to refine algorithms, enhance location mapping, and leverage cutting-edge technologies, the future promises even more targeted, personalised, and effective advertising strategies.