TotalEnergies

Energy · International

Data access: days to minutes

TotalEnergies builds internal data marketplace, cutting data access from days to minutes

500+ users, 500+ connected assets within months of launch. Internal eSpark platform on AWS.

Friction

Data was fragmented across country and business units. Access to critical data took days and required extensive coordination. AI and analytics could not scale without a shared data foundation.

Breakthrough

An internal data marketplace (eSpark) where data owners publish data as products and consumers subscribe, like an internal app store for data. Technology and culture change combined.

Impact

Data access reduced from days to minutes. Approximately 200 users and 500 connected assets in first 4 months. 500+ users shortly after. Duplicate data infrastructure costs significantly reduced.

Unlock the full analysis with breakthrough, impact, what made it smart and its technical approach below!

Name

Organisation name

Problem

TotalEnergies had data fragmented across country and business units. Access to critical data took days and required extensive coordination. Without data ownership and governance, AI could not be applied at scale.

What made it smart

The marketplace model: treating data as a published product rather than a shared resource. This creates accountability through data owners, discoverability, and reuse. New roles and governance changed the culture around data as well as the technology.

Technical approach

eSpark: cloud-based data marketplace on AWS. Amazon S3 for storage, publish/subscribe mechanism for data products, centralised governance with IAM and auditing. Data owners publish sets as data products; consumers search and subscribe via a portal. AWS Kinesis for real-time data streams.

Strategic lesson

Data platforms fail when they are infrastructure projects. They succeed when they are product and culture projects, with clear ownership and consumption incentives.

Reflection question

How much time does your organisation spend finding and preparing data, before anyone can actually use it to make a decision?