Stephen Elson

I led a large, multi-year strategic initiative to rebuild the Financial Times' content publishing platform, greatly enhancing the speed and reliability of publishing content to users, and augmenting it with extensive metadata to enhance discoverability.

The new platform was delivered incrementally while running the old system in parallel and involved migrating hundreds of API customers over the course of several years as the capabilities of the platform grew.

The primary customers were other Financial Times' teams, who used the new APIs to build customer-facing products, and B2B customers wishing to syndicate the Financial Times' content or leverage the metadata for trading purposes.

The platform employed a Cloud-Native technical stack, utilizing AWS, microservices, Golang, Kubernetes, Docker, Neo4j, and Elasticsearch.

In 2009 the BBC was using dozens of different content management systems to drive various parts of bbc.co.uk. I saw this was hugely inefficient and expensive, and proposed a solution to the BBC Future Media board. Previous attempts to solve the problem had been abandoned due to the new CMS imposing a single content model and becoming a bottleneck for delivery.

My proposal would deliver a more flexible SaaS tool, which freed development teams to define their own content models and receive content via a REST API. The CMS did not provide any HTML rendering or other page building, leaving products free to implement this however they preferred.

Serious emphasis was given to providing editorial users with a simple, elegant, and usable interface. Significant effort went into research, testing, and ongoing engagement with users. This combination of stripped back technical features and editorial usability proved to be very successful. The system is still in place today and manages all of bbc.co.uk.

In 2017 I secured £800,000 funding from the Google Digital News Initiative to increase engagement with Financial Times content by creating intelligence about the quality of metadata tagging and making it available to the newsroom.

The Financial Times had a new web site that used metadata as one of the primary mechanisms for joining users with content. However, editorial tools for applying metadata were old and needed improvement. I worked with data scientists to generate models about how metadata quality was likely to affect user engagement and generate suggestions for specific actions that could be taken to improve the metadata.

The team then created dashboards and real-time notifications for journalists, highlighting the quality of the metadata tags they were applying and providing contextual suggestions about improvements they could make. The trial ran for nine months, and the system was trialed with one of the main editorial desks responsible for metadata, after which a significant improvement in KPIs was recorded. Findings from the trial were fed into a new internally funded project to create a new generation of metadata tagging tools.