Need for speed: using real-time data to improve government delivery  

By on 24/01/2024 | Updated on 24/01/2024
New York City. Photo by Pierre Blaché via Pexels

As governments work to meet citizens’ ever higher expectations of public services, they increasingly need good data fast. At a Mastercard webinar, experts discussed the growing appetite for real-time and near real-time data, how to reap the benefits – and avoid the pitfalls

As governments around the world battled to contain the spread of COVID-19, keep populations safe and safeguard economies, it became imperative to collect data that showed the picture ‘as it is now’ rather than ‘as it was then’.

And, in this fast-changing and uncertain world, the appetite for real-time and near real-time insights has only grown. During a Mastercard webinar, public and private sector experts from the US, UK and Canada explored how data could be collected from a range of sources to improve decision-making and the delivery of public services, and touched on AI, training, and the importance of people.

Neil McIvor, director, chief data officer and chief statistician at the UK Department for Education, began by highlighting that now more than ever, governments need to be able to access timely statistics, so that they know, for example, what child poverty looked like last month, not 18 months ago.

In order to facilitate this, he said governments and departments need to think about end-to-end data pipelines – getting the data, putting it somewhere, and then using it – and how these overlap with business processes.

There was also a need to brush up on basic principles, he said: those working for the UK government talk “an awful lot” about how to use data but “often forget” about the basics.

“We put operationally critical data into loads of ungoverned spreadsheets that drive error and need cottage industries of people to pull it together, quality assure it and publish it,” he said.

Neil McIvor

Focusing on data gathering in particular, McIvor said if needed data doesn’t exist, questions should be asked about who collects it and what the incentives are to do so.

“I really want schools to be learning establishments, I don’t want them to be data factories, so I really think about the burdens and the impact,” he said.

McIvor ended his opening comments with reference to a daily school attendance data project he had devised with his team. In six months, they stood up a system to collect daily pupil attendance in near real time, which automatically pulls attendance registers from schools every night, does the “data munching”, and provides bespoke dashboards to help schools and the relevant authorities decide how to tackle absenteeism. 

Of the schools in England, 85% opted in to the project. This translates to a flow of 14,000 bits of data into the Department for Education every day, the collection of seven billion records in the department’s data ‘lake house’ (which had to be reengineered for the project), and the creation of 19,000 fully automated bespoke dashboards.

And during teacher strikes, McIvor explained, the team was able to scrape the data every hour, pulling in 80m records on affected days and publishing data on the impact of the strikes at 4pm each afternoon.

New York City insights

The next panellist, Stephen Keefe, vice president, business development, North America, at Mastercard – the knowledge partner for the webinar – spoke about the company’s work with city authorities.  

He explained that in the US alone, the company’s payments network captures around 40 billion transactions annually, and that it works with cities around the world to help them use near real-time data to improve delivery.

Authorities’ need for more near real-time data rose during the COVID pandemic and has also been driven by the increase in natural disasters, Keefe said. By layering over information about events, they can get near real-time indicators of how a situation is progressing and what the danger level is, and over the longer term, how to get on the path to recovery and “what levers a city can pull in order to enact change”.

Stephen Keefe

City authorities can also take economic spending data and combine it with other types of near real-time data broken down by geographic quadrants – whether it be a neighbourhood or a particular street – to measure the impact of particular interventions, such as whether there is an uplift in spending or in traffic if a bike or bus lane is introduced.

“We’re really getting more advanced in understanding those things – policymakers now have the insights to go to their citizens and show them what’s working and what’s not,” Keefe said.

He gave the example of a project Mastercard has been working on with the City of New York. Authorities had pedestrianised some commercial corridors during the Christmas holidays and, while those that were running the programme thought it was driving benefits, they didn’t have the data to prove it. Gathering and analysing the relevant data revealed a US$3m uplift in spending and allowed them to monitor what would happen if they partially closed a street to vehicles, or if it was pedestrianised for only certain hours of the day.

By leaning on data, the city now has insights to better tailor and promote Christmas and other marketing campaigns.

COVID-19 the catalyst 

The webinar audience heard more about the use of data in the Big Apple from Martha Norrick, chief analytics officer of the City of New York, who shared her team’s experience of the coronavirus pandemic and their work with Mastercard.

There are almost 300,000 municipal employees working for the city and Norrick’s team acts as the central coordinating body for data analytics, overseeing data collected in the city from a wide range of sources, from emergency call lines to sensors.

She agreed with Keefe that the COVID-19 pandemic had been the catalyst for the city and other public sector bodies to work more closely with the private sector on data projects. 

Prior to the pandemic, the speed of data creation and collection had been governed by business processes and use cases. When the pandemic hit, “there was this incredible hunger for better information at a faster pace,” Norrick said.

All of a sudden, her team were being asked “a million questions a day” from how to measure school pupil attendance remotely to how to ascertain whether people who would usually apply for benefits at government offices were doing so online.

With the system the way it was at the time – which relied heavily on quarterly tax receipts, for example – they realised they weren’t going to be able to answer such questions without “different and more creative uses of data”.

So, Norrick’s team set up the Recovery Data Partnership as part of which they searched for sources of data that weren’t generated via governmental processes, and which would, ideally, unlock near real-time insight, whether related to the pandemic’s impact on health, the economy or beyond.

Martha Norrick

This led the city to develop its partnership with Mastercard, initially to establish what was happening in the areas of the city that would usually be full of commuters spending money on coffee, lunch and other goods and how spending patterns had changed.

Through working with Mastercard, what they found was that while there was a huge decline in spending in the central business district, it wasn’t being lost to the city, but was instead being spent in other, residential, boroughs.

“That was a very important understanding for the city – that not all of this economic activity had disappeared, [the pandemic] just sort of shifted geographic patterns of spending in the city.”

What is important, Norrick highlighted, is not collecting data for the sake of it but being able to learn from it: “Are you able to really create meaning out of this data as it’s flowing in?”

She ended her opening comments with a related, and important, point: “One thing that I find myself doing sometimes is helping people consider whether or not real-time data is actually what they need. They think they want it but sometimes the question that they’re trying to answer is not actually answerable in real time. And if you answered in real time, you might get it wrong, because there just hasn’t been enough information to really discern a pattern in a meaningful fashion.”

Data sharing and the ‘art of the possible’

Elise Legendre, chief data officer and director general, data policy and transformation at Agriculture and Agri-Food Canada, also agreed that the pandemic had created a pull to be more creative, inventive and innovative with data.

In her department’s case, what was particularly important during that time was understanding how the agriculture industry was responding to travel restrictions that meant the temporary foreign workers it relies upon heavily couldn’t enter the country.

It also helped the organisation redefine the way it shared data and overcome the challenges of sharing data between departments and ministries.

“I would say the crisis showed you the art of the possible when you really need to get to something, and how much more collaborative we can really be when push comes to shove,” Legendre said, pointing out that data work is primarily not about technology but about people.

“Data is a team sport. And when you’re talking about policy and questions that we need to answer, I think it’s very important to focus the mind.”

Elise Legendre

As a scientific department, collecting data isn’t a problem for Agriculture and Agri-Foods Canada – the main challenge is “making sense of what we have… understanding the potential of it is important”. 

“I think that the true value and the real impact comes when we’re able to collect data once and use it for multiple purposes and being able to leverage it to help answer other questions,” she said. “It’s a really interesting mix of soft skills and people skills, but also that technical knowledge to understand what questions you’re trying to answer.”

Getting data use right in the public sector is “a critical challenge that we have to take on because it speaks about the relevance of the public service – our ability to be able to explain what we do and the impact of the investments that we make, and not just of the dollar,” she added.

People often search for the “perfect data”, she highlighted, but what’s important is to find that “sweet spot” at which you have enough “to get going, to start peeling and taking the layers off and understanding the issue”.

This way, government can “make sure that our policies are actually having the impact that we want them to have, and if not, to be able to course correct quickly enough”.

A single source of truth, avoiding ‘nice to haves’, and AI

During the Q&A portion of the webinar, in which panellists answered questions from the live audience, someone asked about creating a single source of truth, to which Norrick offered a caveat.

“The trouble comes when people think of datasets as representations of reality when actually the data generation process means that we’re seeing a little window into what’s actually happening,” she said.

This was a point Legendre agreed with. “A single source of truth will remove a lot of noise in the organisation,” she said, adding that it helps whoever the data custodian is to know what they have, where it comes from and what the limitations are. “But it’s when you start comparing with other things that you can start doing some interesting and creative analysis.”

Keefe made the point that it is possible to interpret something that is not representative of the dataset in question and that that is why, in the case of data procured from the private sector, a partnership approach is “critically important”. Only the entity that has collected the data can provide the detail on what the nuances and caveats are so it’s helpful to be able to go to them and say: “This is what we think the insights are telling us – is that true?”

The discussion moved on to how to prevent officials wasting time on collecting ‘nice to have’ data that won’t necessarily yield benefits.

Building on his opening comments, McIvor’s suggestion was to look at the cost. “What’s it going to cost us in pounds and pence or dollars and cents? What’s it going to cost us in time? What’s it going to cost us in resources to be able to do this? What’s the opportunity cost? What else could you be doing with those people and that money? Be very clear and open on the burden versus benefit dynamic.”

Most people think data is a free good – “if I wave my wand, someone over there is going to do it for me” – but this is not the case.

Next came discussion about the impact of AI. Norrick said that, for her, the most exciting application for large language models in government is their ability to make sense of and find patterns in large amounts of unstructured data, such as the free-text boxes on forms filled out by citizens.

However, she added that she would be hesitant, at present, to use AI in a way that drives operational decision-making, because of “the proclivity to hallucinate and a lot of things that we’re still learning about these models”.

Legendre said it was imperative to ensure that governments are not feeding AI models with biased data that could perpetuate systemic issues. Technology must, she said, be used “for the greater good”.

Open data and learning to ‘draw the signal from the noise’

Another question from the audience centred on open data and how to manage the sharing of data with the public.

“I think the era of just dumping all of your data on the internet with no additional context or ability to interpret it is not serving anybody,” Norrick said. As such, the City of New York is working on providing a “robust data dictionary” with every openly available dataset, as well as details of the responsible person should a member of the public have any questions.

The authority also runs the Open Data Ambassadors programme, through which volunteers – including city employees, librarians, civic technology enthusiasts and others – are trained to help people think about and use open data accurately and responsibly.

McIvor added that what is important is to understand how people absorb information and that, as such, his colleagues spend time with psychologists to discern the user persona, and to figure out how best to help people with basic usage principles and to learn to “draw the signal out from the noise”.

Panellists also touched on the need to increase the basic level of data literacy across the public sector.

McIvor noted that the UK government had launched the One Big Thing programme, through which every civil servant was asked to complete a day’s worth of data training in autumn 2023, and added that “if you’re creating silos of data experts, you’re not democratising the data”.

Legendre made the point that “a lot of people are afraid of data” and demystifying data – “getting people comfortable with data, understanding the limits, understanding that it’s a tool, it’s not the be all and end all of everything, is important”.

She added: “It’s a lot about the behaviour. Everybody is a data person in an organisation whether they know it or not. So making them embrace that, and their responsibility [for] the quality of the data that they put in the system, is important as well.”

Panellists agreed that data generation should be done in partnership with the frontline employees who interact with citizens and undertake much of government’s data collection.

It’s about demonstrating “what we’re going to do with that data and why the quality of that data is important,” Norrick said, adding: “Entering lots of data into a system can be boring and thankless. How do we connect with and demonstrate to the people who are responsible for collecting data and to the citizens that are giving up their information that we’re showing what is happening with that information and how we’re using it to improve services?”

Keefe agreed that a “transaction based data relationship is not the best way. It’s more of a continuous partnership – data engagement is the right way to approach it”.

Legendre explained why it is important to bring people together from different perspectives, so they can both define the question data can answer and find solutions. “Understanding everybody’s contribution, and how much you can lighten the load on the frontline, how much you can lighten the load on your recipient and provide that information back to them is incredible. I think that’s where you get buy-in pretty quickly.”

McIvor concluded the session with a final note about open data: “Having open data that people can trust is the foundation of a democratic process and a democratic system, because then, if you put the insight out there… people can formulate their own views or opinions based on facts and not fake facts.”

Indeed, the public sector’s own growing appetite for real-time and near real-time data feeds just as neatly into the democratic process – and making decisions that will affect service delivery based on fact, not fiction.

You can watch the full Just in time: how real-time data can improve government delivery webinar on our dedicated event page. The webinar, hosted by knowledge partner Mastercard with support from Global Government Forum, was held on 28 November 2023.

About Knowledge Partner Mastercard

Leave a Reply

Your email address will not be published. Required fields are marked *