How the ONS is Making UK Government Decision-Making More Data-Driven

Tomas Sanchez Lopez PhD, Chief Data Architect at the Office for National Statistics, outlines the role data design processes and domain-oriented data platforms play in the ONS’ data strategy 

From measuring COVID-19 infection rates to delivering the national census, to tracking the impact of inflation on the cost of living – the Office for National Statistics (ONS) provides the authoritative source of critical data guiding government policymaking. 

As the ONS’ Chief Data Architect, Tomas Sanchez PhD played a pivotal role in designing the systems and processes necessary to track these data accurately on a national scale.  

Ahead of Corinium’s 2022 CDAO UK events, he shares how he’s approached key data management challenges since joining the organization in 2017, as well as what data leaders in private corporations should take from his experiences.  

“We will be focusing on also making data available, with good practices in terms of data linking [and] with good practices in terms of metadata” 

Tomas Sanchez PhD, Chief Data Architect, ONS 

“There is no ‘one size fits all’ approach to formulating a data strategy and ensuring data quality,” says Dr Sanchez. “Data management and data architecture are a long game. 

“You have to invest a substantial amount of time understanding the organization’s data needs so that you propose ideas that make sense. And it takes even more time to make sure that the data management practices are actually being followed.” 

Three Domain-Oriented Data Platforms 

When he joined the ONS, Dr Sanchez spent his first two and a half years setting up its data architecture function and formulating its data strategy, which the organization published in 2019.  

One focus for Dr Sanchez during this time was the question of how to meet the needs of the different stakeholder groups who use ONS data. These include ONS analysts, independent researchers and government policymakers. 

To provide the right levels of access and data to each of these groups, the ONS is developing three domain-oriented data platforms: 

• The Secure Research Service gives accredited researchers outside of the ONS access to de-identified and unpublished datasets to drive research for the public good
• The Data Access Platform is an internal ONS system for analyzing data from the ONS and its partners. One recent project was running Census 2021, the first digital-by-default census across England and Wales, which saw 97% of households responding
• The Integrated Data Service is a cloud-based platform that’s being developed to foster data sharing across government departments and give government analysts access to government data to inform policymaking

“The [Integrated Data Service] is a proposal the ONS has put to the treasury for creating a platform for government for sharing data,” Dr Sanchez explains. “We are the delivery partner for that, and all the principles of our data strategy are being applied to that proposal.” 

This approach mirrors those of ‘data mesh’ advocates. They argue that data leaders in corporate settings will also benefit from designing domain-oriented data ecosystems that provide data and analytics capabilities to different stakeholder groups based on their needs. 

Putting Data Design First 

In addition to providing colleagues with the right access to the right data, Dr Sanchez and his team are responsible for ensuring those datasets are high quality and provide the right context for decision-making. 

To achieve this, every project at the ONS starts with a period of ‘data design’. This covers the design of data ingestion processes, the translation of data into the right formats for the ONS’ internal data platform and the automation of data management and quality assurance steps in the data pipeline. 

"We get either a specification of the datasets or samples of the datasets (or both) and create the pipeline and automate the pipeline as much as possible,” Dr Sanchez explains. “So that, hopefully, when we get the full datasets of millions or billions of rows, most of the processing can be done automatically.” 

For complex projects such as censuses, getting this process right is essential for ensuring the resulting data can be analyzed to uncover the insights the ONS wants to uncover.  

 “A lot of times with data quality, people just think about collecting data and analyzing it. They forget about all the preparation that goes in the middle” 

Tomas Sanchez PhD, Chief Data Architect, ONS 

However, at the height of COVID-19, the ONS fast-tracked the data design on some projects, so they could provide officials with information more quickly to inform the pandemic response. As a result, the team then had to then refine how they were collecting the data as they went. 

“By definition, if you want to do data strategically, you have to do a data design exercise,” Dr Sanchez recalls. “That takes time, and time is something that we did not have.” 

Moving forward, Dr Sanchez hopes to retain the greater cross-departmental collaboration and agility his team achieved during this period, while still being strategic about data design. 

He recommends that companies that wish to use data to transform their businesses also focus on getting these foundations right before rushing to analyze their data for useful insights. 

He concludes: “Of course, what you do with your algorithms is important. But if you just have very bad quality data, no matter what you do with the algorithm, the output is going to be bad.”

ONS Chief Data Architect Tomas Sanchez PhD is speaking at Corinium's 2022 CDAO UK conference. For more information and to reserve your place, click here now.