Pratima Aiyagari

09th May 2022

YZR makes you wiser! Or, why we invested in YZR

Modern companies have long been generating large amounts of data. But before truly analyzing data, it first needs to be cleansed to generate truly actionable insights. Along with Orange Ventures, we were delighted to co-lead the $12m Series A investment round in YZR, which helps normalize data sets. We welcome Sébastien, Jean-Philippe, Yosr, Arjun and the entire YZR team to the Nauta Family!

Meet the leading team behind YZR – Sébastien Garcin, Jean-Philippe Poisson, Yosr Mhiri and Arjun Chatterjee.

Modern Data Stack

Within any organisation, various business processes, sensors, and other devices generate data. In the case of sensor or real-time, the data needs to be safely transported before it is prepped and stored. In other cases, it needs to be discovered and ingested into appropriate data stores. With the growing urgency to move applications to the cloud, enterprises have a hybrid legacy and cloud application mix, producing ever more data.

Add to this mix the complex and data-heavy machine learning applications and related operations and you get an evolving data environment. A now common part of the enterprise backdrop as they evolve and adapt to the modern data stack.

Quantity vs quality

Within the data value chain – generating data, transporting it, storing, transforming and analysing for insights – the insights are truly valuable only if the data is clean. Business intelligence and analytics that run on messy data are entirely useless. There is a computer science phrase for this: “Garbage in, garbage out.” The cleansing of the data that is created within an organization is the crucial first step, which will result in valuable and actionable business insights.


Where does YZR fit in

YZR was founded by Sébastien Garcin and Jean-Philippe Poisson from a common pain point they were facing when working on various digitisation projects across L’Oréal’s 60+ sub-brands. They kept facing messy heterogeneous data sets which required data normalisation.

Data normalisation is the process of prepping any kind of (business) data to be ready, accurate, and well-organised for reports, machine learning, analysis, predictions, and business intelligence functions.

They quickly realised that without the right normalisation and preparation of the 1000s of SKUs across various L’Oréal brands, the actual business use cases of the digitisation projects and the resulting business intelligence, insights, and optimisations proposed were just a faraway dream. They scoured the market for a suitable solution but only encountered Master Data Management Solutions, which only solve part of the problem.

Hence, in 2019 they founded YZR (pronounced ‘wiser’) to solve this issue of semantic data normalization.

YZR ingests raw and messy data sets either directly from the various data sources or from the customer’s ETL/data lakes via its API connection or batch processes. Then, the API uses AI to do the following:

  1. Automatic Standardisation: replace the same attribute that might have been written in different formats in different data sets with one global type
    • yoghurt, yogrt, yogurt, yogurtz, yogut, yog = yoghurt
  1. Smart Labelling: creating smart labels and grouping attributes to the appropriate abstraction layer requested by the business while taking context into account
    • strawberry/apple/mint –> flavour (if the segment seems to relate to food)
    • rose/mint/apple –> colour (when related to clothing or non-edible items)
  1. Fuzzy Matching: which is a de-duplication of similar items with the same underlying meaning based on language analysis
    • waterproof = wtprf = waterpf

Upon normalizing the data sets, it then returns this cleaned up and normalised data set to your data lake where it can flow as usual into other systems that rely upon this data. Precious for any business which uses a large amount of semantic data, YZR’s first clients were retail and e-commerce leaders such as Monoprix or La Redoute. YZR accompanied them in automatically standardizing heterogeneous product data from their suppliers. This more qualitative and granular data is then used to create next-generation pricing or customer loyalty policies.

What we loved about this investment opportunity

The explosion of data from all processes within the enterprise is evident. Evolving machine learning use cases and their adoption in enterprises are picking up pace. Among other reasons, the stellar team at YZR and an attractive market were some of the reasons we decided to invest in YZR.

Other reasons include:

  • Continued data explosion and the garbage in, garbage out problem – The ongoing rise and creation of ever more datasets is compounding the age-old problem of garbage-in, garbage-out that affects all software systems and is especially detrimental to AI and ML models that can be completely ruined by learning from or running on the unclean training sets. YZR and its data normalisation solution are tackling this problem head-on across all verticals.
  • Strong market pull – 100% of Chief Data Officers across all industries know this is a large issue: data preparation and normalisation are currently their key priorities. Without having access to tools that help them resolve their data and business needs, data teams have been forced to create internal tooling and Excel workarounds just to keep up with the demands of the business. Due to this, most data teams are in a state of continuous firefighting rather than pro-active data harmonisation.
  • Extremely sticky product – By adding value at the data level, YZR is positioning itself as a foundational layer on top on which other systems and business processes rely on. As such, it is very easy to land and expand with more data use cases but also embeds YZR deep in a company’s data infrastructure stack.
  • An ambitious and world-class team – The team consists of world-class founders and talent, who bring a breadth of experience in scaling large organisations, digital transformations at enterprise level, as well as technical know-how to the table. They have banded together on this ambitious problem and are building out the missing step that was preventing the value of data to be unlocked.

We are very happy to welcome the YZR team to the Nauta Family! And we are delighted that we were invited to join YZR on this journey. We are extremely bullish on the data infrastructure and tooling market.

More about YZR on their website here:

See a demo here.