Linda Hoeberigs, Sr Manager, Data Science & AI, CMI PDC
Ishit Ghandi, CMI PDC Hive Data Scientist
Bhavya Kantiwal, CMI PDC Hive Data Scientist
Amit Shirkane, CMI PDC Hive Data Science Manager
Anwita-Ajit Bhure, CMI PDC Hive Data Scientist
Anda Ziemele, CMI PDC Data Scientist
Chandan Agarwal, CMI PC Director, PC and Advanced Analytics
Chhavi Agarwal, CMI PDC Manager India
Adrian Oruclar, CMI PDC Labs Director
Roja Kompalli, UniOps Sr. Solutions Architect
Unilever is one of the world’s leading suppliers of Beauty & Personal Care, Home Care, and Foods & Refreshment products, with sales in over 190 countries and products used by 3.4 billion people every day. We have 148,000 employees and generated sales of €52.4 billion in 2021. Over half of our footprint is in developing and emerging markets. We have around 400 brands found in homes all over the world – including iconic global brands like Dove, Lifebuoy, Knorr, Magnum, OMO, and Surf; and other brands such as Love Beauty & Planet, Hourglass, Seventh Generation, and The Vegetarian Butcher.
Unilever's Marketing and R&D teams needed to figure out what trends exist with consumers in the market that could justify the launch of new products. To do this, it was critical to identify where product trends first pop up geographically, to which countries they are transmitted, and where they tend to end up as other markets follow. Once this knowledge was acquired, the business could then track the originator countries for search and product launch patterns that could inspire Unilever's product development practices and ensure their extension across markets in a timely manner.
The main challenge was to ensure that this market scan could be conducted across 50+ countries, where different languages are spoken, and categories in Unilever's portfolio, both on demand and in an efficient time period, without sacrificing a detailed data- and statistically-driven approach.
The objective was to build an automatic app that would output what countries are the respective originators, transmitters, and followers of trends relating to a given sub-category (e.g., tea, face masks).
Based on a user’s inputted seed keywords and key concepts describing the category that has relevant interest and growth across the last four years — the app will automatically picture a time series graph that matches the provided terms. It describes how trends travel around the world based on consumer searches and product launches, which are based on various statistical methods.
Initially, we tried to obtain this via a flow and plugin. However, the output was not visually appealing enough for Unilever’s marketing teams. As a result, the same solution was rebuilt in a webapp, which was a new feature at the time.
The vision for the app was to offer users the opportunity to conduct this analysis at any time and across any conceptual space of Unilever’s business for any relevant markets, all in a matter of hours. The analysis is based on validated statistical methods run on top of search and product launches time series data.
Once it’s clear how the trends are traveling, the business can then understand what trends are popping up in originator markets and ensure those are materialized into Unilever products (if they’re not already) and transmitted to typical follower markets of the category under analysis.
Business Area: Marketing/Sales/Customer Relationship Management/Product & Service Development
Use Case Stage: Built & Functional
This solution is unique as it is, to our knowledge, the first capability of its kind on the market. It is a realistic solution in the sense that it uses not only search data, but also product launch data, providing a robust vision of consumer intent. What’s more, it is a statistically validated solution with state-of-the-art methods that analyze causality across time-series data in its foundation.
Our solution is also scalable, as it’s built as a webapp that can run in a matter of hours for any categories and markets required. Anyone can use the app front-end to generate the required results, with or without a programming background.
Its outputs are equally friendly to non-technical users. They are provided as Excel or image files which can be easily integrated into any typical presentation format, such as PowerPoint, thereby reducing the gap between the app’s users and senior business stakeholders who require these insights to make business decisions.
This app has reduced the lead time for insight into how trends travel by three months. This has a direct impact on the cycle-time of new product developments and market launches. As an illustrative example of insights provided, thanks to the Trends Geo-Tracking app, the business is now aware that shampoo trends come mainly from the US and Indonesia markets, with countries such as Canada, Mexico, Brazil, Thailand, and Australia being the transmitters, and France and Japan being the followers. Several other studies of this nature have been produced equally Additionally, there are relevant insights that help the Marketing and R&D teams at Unilever to pick the right markets to work with depending on the maturity stage of any given trend or product.
Furthermore, our India CMI PDC team was able to use Trends GeoTracking to understand color cosmetics which helped them both to understand the color palettes, as well as the style of applying, and how those would spread from country to country, and state to state. This gave our Indian business an 18-month lead time to get products into the market at the right moment. Out of the 10 predicted trends and styles, eight arrived within a month of the predicted time frame, putting Unilever’s in-market success rate ahead of the competition.
All of this now takes a matter of hours, versus the 10 days, it would have taken to run this analysis before we turned it into a repeatable process with Dataiku.
Value Brought by Dataiku:
Dataiku allowed us to create this end-to-end pipeline within a single tool, which would have otherwise been very difficult, as it brings together visual input, data gathering, data wrangling, statistics, and interactive visual output.
Dataiku helped us to fully automate the GeoTracking process, which can be run in a couple of hours, which would have otherwise taken 10 days using our offshore engine.
- Save time
- Trends — scale across multiple markets