Turning messy data into a single customer view

Three blue digital panels with light streaks and binary code on a white background
Blue “Onward” wordmark on a white background

About this project:

Analyse and merge messy, duplicated data into a single source of truth for people and addresses.

"We all know that clean data and strong governance is crucial to automation and scaling AI - this project has helped Pinboard Consulting achieve just that!"


Marie May

Head of Nurture


  • Reach out to Marie to find out how Nurture can help your business to achieve goals
Smiling woman in a black sleeveless top, standing against a light wall with floral line art.

Executive Summary

To Achieve a Single Customer View

Objective:

The challenge of this project was to use, test and validate a deterministic, rules based, data matching engine to determine and measure its efficacy in matching bad, partial, or otherwise compromised data.


Solution:
Our Nurture student analysed the data then used matching rules to find the best way to resolve the unresolved entities. She then used a test harness to validate the output before using AWS Athena to perform detailed matching validation.


Results:

  1. Far better understanding of the difficulties of implementing matching software
  2. A good set of analysis outputs on the data under test
  3. A clear set of matching rules giving good results across the dataset




Forming a Single View of Customer is a challenge for companies across many industries. Most companies have multiple data sources across multiple divisions or lines of business, they also have multiple external sources of data.

 

Each of those sources can (and often do) provide a different view of the end customer – whether that’s an individual or a company. This could include different spellings, different levels of data completeness, different levels of data quality, different country / location sources, and so on.

 

The challenge of this project was to use, test and validate a deterministic, rules based, data matching engine to determine and measure its efficacy in matching bad, partial, or otherwise compromised data.

Forming a single view of our customers