Unilever - Developing a Scalable Digital Voice of the Consumer Capability
Name: Anand Patel (Digital Voice of the Consumer team)
Title: Analytics Manager
Country: United Kingdom
Description: Everyday 2.5 billion people use a Unilever product to look good, feel good or get more out of life. Our purpose is to make sustainable living commonplace. We are home to some of the UK’s best- known brands like Persil, Dove and Marmite, plus some others that are on their way to becoming household favourites like Seventh Generation and Grom. We have always been at the front of media revolutions whether that be the 1st print advertisements in the 1890s or in 1955 when we became the 1st company to advertise on British TV screens. Experimentation and bravery drive us and have helped us become one of the UK’s most successful consumer goods companies.
AI Democratization & Inclusivity
As a consumer obsessed business, Unilever had an ambition to ensure all business decisions are centered around our consumers. In order to do this, we must listen to, understand, and adapt to the changing needs and wants of our consumers.
Our challenge was to develop a scalable digital voice of the consumer capability, that would enable us to interpret large, complex and unstructured consumer feedback datasets through leading data science techniques and serve relevant consumer insights to business decision makers for actioning.
The solution would be developed in partnership with Unilever’s Quality function, to enable them to identify product issue and opportunities areas to improve and innovate based on consumer comments. The comments would be sourced from social media, product reviews, and Unilever’s customer engagement centers.
We therefore required a platform that would enable us to build an industrialized AI solution end-to-end, including data merging, cleansing, manipulation, modeling, and output. The platform should enable the processing of large volumes of datasets in near real-time and enable the development and deployment of sophisticated AI-driven natural language processing models.
The AI models will be key to automation and deriving prescriptive insights from the large unstructured datasets. Models should also be made available to be used by the rest of the business for other products or analysis if needed. The developed flow show be repeatable and performed on a cloud set up, to benefit from distributed data storage and processing.
An industrialized solution called Digital Voice Of the Consumer (DVOC) was developed to fulfill the ambition. With Dataiku as the underlying platform, we were able to develop an automated and scaled solution that is updated daily and democratizes consumer insights to Unilever’s entire organisation.
Through developing multiple flows in Dataiku, we were able to bring together an array of internal and external consumer feedback datasets and enrich with additional product data. The datasets were then cleansed and structured using a combination of Dataiku’s pre-built nodes and Python scripts.
Machine learning-based natural language models were developed using leading-edge methods and enabled unsupervised learning to identify the key topics and themes consumers were talking about. Deep learning algorithms were also developed for sentiment classification.
Through the versatility and intuitive nature of Dataiku, the flow was developed for data scientists and analysts with all levels of experiences and provided a great way for more junior team members to upskill themselves. Developed machine learning models have been deployed and also made available to the rest of the Unilever community through the plugins feature on Dataiku.
The overall DVOC flows are complex, however the Dataiku platform enables us to visually display the flow and create groupings and scenarios to cluster related notes together – making it easier to make developments, test and diagnose bugs and errors.
Our Digital Voice Of the Consumer solution is now live in over 60 markets, 50 factories across the globe and available in more than 20 languages with an active user base of over 2,000 people.
The tool is embedded into the day-to-day operations of the Quality team, and is also widely used across other business functions including marketing, research & development, and customer development teams.
There have been over 750 business insights logged from the capability to date with over 500 of these actioned and closed. DVOC insights have been actioned across Unilever’s business to make a real impact on consumers. Examples include:
Helping Unilever develop and launch new products in line with what consumers really want.
Optimise existing products so that are more suited for eCom by improving product packaging to reduce leakage and breakages.
Redesign of products and packaging to make them more sustainable.
Helping fight counterfeits.
Tracing back issues to specific Unilever factories and enabling Unilever’s business to be agile during the Covid outbreak by reacting to changes in consumer behaviors.
This has led to tangible business results to deliver costs savings, increased profits and ultimately improved product quality and better consumer experiences.Individual cases have unlocked costs savings of up to €350k each.